From c4ee6d9b73300d906b8df4602157d646df2415ef Mon Sep 17 00:00:00 2001 From: Robert Barron Date: Sun, 30 Jul 2023 00:41:10 -0700 Subject: [PATCH 001/342] xyz_grid: allow varying the seed along an axis along with the axis's other changes --- scripts/xyz_grid.py | 18 ++++++++++++++++-- 1 file changed, 16 insertions(+), 2 deletions(-) diff --git a/scripts/xyz_grid.py b/scripts/xyz_grid.py index 1010845e5..4eb1b197d 100644 --- a/scripts/xyz_grid.py +++ b/scripts/xyz_grid.py @@ -416,6 +416,10 @@ class Script(scripts.Script): with gr.Column(): include_lone_images = gr.Checkbox(label='Include Sub Images', value=False, elem_id=self.elem_id("include_lone_images")) include_sub_grids = gr.Checkbox(label='Include Sub Grids', value=False, elem_id=self.elem_id("include_sub_grids")) + with gr.Column(): + vary_seeds_x = gr.Checkbox(label='Vary seed on X axis', value=False, elem_id=self.elem_id("vary_seeds_x")) + vary_seeds_y = gr.Checkbox(label='Vary seed on Y axis', value=False, elem_id=self.elem_id("vary_seeds_y")) + vary_seeds_z = gr.Checkbox(label='Vary seed on Z axis', value=False, elem_id=self.elem_id("vary_seeds_z")) with gr.Column(): margin_size = gr.Slider(label="Grid margins (px)", minimum=0, maximum=500, value=0, step=2, elem_id=self.elem_id("margin_size")) @@ -475,9 +479,9 @@ class Script(scripts.Script): (z_values_dropdown, lambda params:get_dropdown_update_from_params("Z",params)), ) - return [x_type, x_values, x_values_dropdown, y_type, y_values, y_values_dropdown, z_type, z_values, z_values_dropdown, draw_legend, include_lone_images, include_sub_grids, no_fixed_seeds, margin_size] + return [x_type, x_values, x_values_dropdown, y_type, y_values, y_values_dropdown, z_type, z_values, z_values_dropdown, draw_legend, include_lone_images, include_sub_grids, no_fixed_seeds, vary_seeds_x, vary_seeds_y, vary_seeds_z, margin_size] - def run(self, p, x_type, x_values, x_values_dropdown, y_type, y_values, y_values_dropdown, z_type, z_values, z_values_dropdown, draw_legend, include_lone_images, include_sub_grids, no_fixed_seeds, margin_size): + def run(self, p, x_type, x_values, x_values_dropdown, y_type, y_values, y_values_dropdown, z_type, z_values, z_values_dropdown, draw_legend, include_lone_images, include_sub_grids, no_fixed_seeds, vary_seeds_x, vary_seeds_y, vary_seeds_z, margin_size): if not no_fixed_seeds: modules.processing.fix_seed(p) @@ -648,6 +652,16 @@ class Script(scripts.Script): y_opt.apply(pc, y, ys) z_opt.apply(pc, z, zs) + xdim = len(xs) if vary_seeds_x else 1 + ydim = len(ys) if vary_seeds_y else 1 + + if vary_seeds_x: + pc.seed += ix + if vary_seeds_y: + pc.seed += iy * xdim + if vary_seeds_z: + pc.seed += iz * xdim * ydim + res = process_images(pc) # Sets subgrid infotexts From aab385d01b4311726127397552d791f4d71b7147 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Sun, 3 Sep 2023 11:56:02 +0900 Subject: [PATCH 002/342] thread safe extra network list_items --- extensions-builtin/Lora/ui_extra_networks_lora.py | 10 +++++----- modules/ui_extra_networks.py | 2 ++ modules/ui_extra_networks_checkpoints.py | 6 +++--- modules/ui_extra_networks_hypernets.py | 5 +++-- modules/ui_extra_networks_textual_inversion.py | 5 +++-- 5 files changed, 16 insertions(+), 12 deletions(-) diff --git a/extensions-builtin/Lora/ui_extra_networks_lora.py b/extensions-builtin/Lora/ui_extra_networks_lora.py index 55409a782..e9f300621 100644 --- a/extensions-builtin/Lora/ui_extra_networks_lora.py +++ b/extensions-builtin/Lora/ui_extra_networks_lora.py @@ -66,11 +66,11 @@ class ExtraNetworksPageLora(ui_extra_networks.ExtraNetworksPage): return item def list_items(self): - for index, name in enumerate(networks.available_networks): - item = self.create_item(name, index) - - if item is not None: - yield item + with self.thread_lock: + for index, name in enumerate(networks.available_networks): + item = self.create_item(name, index) + if item is not None: + yield item def allowed_directories_for_previews(self): return [shared.cmd_opts.lora_dir, shared.cmd_opts.lyco_dir_backcompat] diff --git a/modules/ui_extra_networks.py b/modules/ui_extra_networks.py index 063bd7b80..564bab7fe 100644 --- a/modules/ui_extra_networks.py +++ b/modules/ui_extra_networks.py @@ -1,6 +1,7 @@ import os.path import urllib.parse from pathlib import Path +from threading import Lock from modules import shared, ui_extra_networks_user_metadata, errors, extra_networks from modules.images import read_info_from_image, save_image_with_geninfo @@ -94,6 +95,7 @@ class ExtraNetworksPage: self.allow_negative_prompt = False self.metadata = {} self.items = {} + self.thread_lock = Lock() def refresh(self): pass diff --git a/modules/ui_extra_networks_checkpoints.py b/modules/ui_extra_networks_checkpoints.py index ca6c26076..2753214fa 100644 --- a/modules/ui_extra_networks_checkpoints.py +++ b/modules/ui_extra_networks_checkpoints.py @@ -30,9 +30,9 @@ class ExtraNetworksPageCheckpoints(ui_extra_networks.ExtraNetworksPage): } def list_items(self): - names = list(sd_models.checkpoints_list) - for index, name in enumerate(names): - yield self.create_item(name, index) + with self.thread_lock: + for index, name in enumerate(sd_models.checkpoints_list): + yield self.create_item(name, index) def allowed_directories_for_previews(self): return [v for v in [shared.cmd_opts.ckpt_dir, sd_models.model_path] if v is not None] diff --git a/modules/ui_extra_networks_hypernets.py b/modules/ui_extra_networks_hypernets.py index 4cedf0851..411b4f111 100644 --- a/modules/ui_extra_networks_hypernets.py +++ b/modules/ui_extra_networks_hypernets.py @@ -31,8 +31,9 @@ class ExtraNetworksPageHypernetworks(ui_extra_networks.ExtraNetworksPage): } def list_items(self): - for index, name in enumerate(shared.hypernetworks): - yield self.create_item(name, index) + with self.thread_lock: + for index, name in enumerate(shared.hypernetworks): + yield self.create_item(name, index) def allowed_directories_for_previews(self): return [shared.cmd_opts.hypernetwork_dir] diff --git a/modules/ui_extra_networks_textual_inversion.py b/modules/ui_extra_networks_textual_inversion.py index 55ef0ea7b..d25b45d61 100644 --- a/modules/ui_extra_networks_textual_inversion.py +++ b/modules/ui_extra_networks_textual_inversion.py @@ -29,8 +29,9 @@ class ExtraNetworksPageTextualInversion(ui_extra_networks.ExtraNetworksPage): } def list_items(self): - for index, name in enumerate(sd_hijack.model_hijack.embedding_db.word_embeddings): - yield self.create_item(name, index) + with self.thread_lock: + for index, name in enumerate(sd_hijack.model_hijack.embedding_db.word_embeddings): + yield self.create_item(name, index) def allowed_directories_for_previews(self): return list(sd_hijack.model_hijack.embedding_db.embedding_dirs) From 25de9a785cc9e93c16626db6ab5b16824443de53 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Sat, 9 Sep 2023 16:56:19 +0900 Subject: [PATCH 003/342] Revert "thread safe extra network list_items" This reverts commit aab385d01b4311726127397552d791f4d71b7147. --- extensions-builtin/Lora/ui_extra_networks_lora.py | 10 +++++----- modules/ui_extra_networks.py | 2 -- modules/ui_extra_networks_checkpoints.py | 6 +++--- modules/ui_extra_networks_hypernets.py | 5 ++--- modules/ui_extra_networks_textual_inversion.py | 5 ++--- 5 files changed, 12 insertions(+), 16 deletions(-) diff --git a/extensions-builtin/Lora/ui_extra_networks_lora.py b/extensions-builtin/Lora/ui_extra_networks_lora.py index e9f300621..55409a782 100644 --- a/extensions-builtin/Lora/ui_extra_networks_lora.py +++ b/extensions-builtin/Lora/ui_extra_networks_lora.py @@ -66,11 +66,11 @@ class ExtraNetworksPageLora(ui_extra_networks.ExtraNetworksPage): return item def list_items(self): - with self.thread_lock: - for index, name in enumerate(networks.available_networks): - item = self.create_item(name, index) - if item is not None: - yield item + for index, name in enumerate(networks.available_networks): + item = self.create_item(name, index) + + if item is not None: + yield item def allowed_directories_for_previews(self): return [shared.cmd_opts.lora_dir, shared.cmd_opts.lyco_dir_backcompat] diff --git a/modules/ui_extra_networks.py b/modules/ui_extra_networks.py index 564bab7fe..063bd7b80 100644 --- a/modules/ui_extra_networks.py +++ b/modules/ui_extra_networks.py @@ -1,7 +1,6 @@ import os.path import urllib.parse from pathlib import Path -from threading import Lock from modules import shared, ui_extra_networks_user_metadata, errors, extra_networks from modules.images import read_info_from_image, save_image_with_geninfo @@ -95,7 +94,6 @@ class ExtraNetworksPage: self.allow_negative_prompt = False self.metadata = {} self.items = {} - self.thread_lock = Lock() def refresh(self): pass diff --git a/modules/ui_extra_networks_checkpoints.py b/modules/ui_extra_networks_checkpoints.py index 2753214fa..ca6c26076 100644 --- a/modules/ui_extra_networks_checkpoints.py +++ b/modules/ui_extra_networks_checkpoints.py @@ -30,9 +30,9 @@ class ExtraNetworksPageCheckpoints(ui_extra_networks.ExtraNetworksPage): } def list_items(self): - with self.thread_lock: - for index, name in enumerate(sd_models.checkpoints_list): - yield self.create_item(name, index) + names = list(sd_models.checkpoints_list) + for index, name in enumerate(names): + yield self.create_item(name, index) def allowed_directories_for_previews(self): return [v for v in [shared.cmd_opts.ckpt_dir, sd_models.model_path] if v is not None] diff --git a/modules/ui_extra_networks_hypernets.py b/modules/ui_extra_networks_hypernets.py index 411b4f111..4cedf0851 100644 --- a/modules/ui_extra_networks_hypernets.py +++ b/modules/ui_extra_networks_hypernets.py @@ -31,9 +31,8 @@ class ExtraNetworksPageHypernetworks(ui_extra_networks.ExtraNetworksPage): } def list_items(self): - with self.thread_lock: - for index, name in enumerate(shared.hypernetworks): - yield self.create_item(name, index) + for index, name in enumerate(shared.hypernetworks): + yield self.create_item(name, index) def allowed_directories_for_previews(self): return [shared.cmd_opts.hypernetwork_dir] diff --git a/modules/ui_extra_networks_textual_inversion.py b/modules/ui_extra_networks_textual_inversion.py index d25b45d61..55ef0ea7b 100644 --- a/modules/ui_extra_networks_textual_inversion.py +++ b/modules/ui_extra_networks_textual_inversion.py @@ -29,9 +29,8 @@ class ExtraNetworksPageTextualInversion(ui_extra_networks.ExtraNetworksPage): } def list_items(self): - with self.thread_lock: - for index, name in enumerate(sd_hijack.model_hijack.embedding_db.word_embeddings): - yield self.create_item(name, index) + for index, name in enumerate(sd_hijack.model_hijack.embedding_db.word_embeddings): + yield self.create_item(name, index) def allowed_directories_for_previews(self): return list(sd_hijack.model_hijack.embedding_db.embedding_dirs) From f5959c1c3022c454de22fab749d0f06ab3219868 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Sat, 9 Sep 2023 17:05:50 +0900 Subject: [PATCH 004/342] thread safe extra network using list --- extensions-builtin/Lora/ui_extra_networks_lora.py | 3 ++- modules/ui_extra_networks_hypernets.py | 3 ++- modules/ui_extra_networks_textual_inversion.py | 3 ++- 3 files changed, 6 insertions(+), 3 deletions(-) diff --git a/extensions-builtin/Lora/ui_extra_networks_lora.py b/extensions-builtin/Lora/ui_extra_networks_lora.py index 55409a782..e74daa770 100644 --- a/extensions-builtin/Lora/ui_extra_networks_lora.py +++ b/extensions-builtin/Lora/ui_extra_networks_lora.py @@ -66,7 +66,8 @@ class ExtraNetworksPageLora(ui_extra_networks.ExtraNetworksPage): return item def list_items(self): - for index, name in enumerate(networks.available_networks): + names = list(networks.available_networks) + for index, name in enumerate(names): item = self.create_item(name, index) if item is not None: diff --git a/modules/ui_extra_networks_hypernets.py b/modules/ui_extra_networks_hypernets.py index 4cedf0851..5f5904915 100644 --- a/modules/ui_extra_networks_hypernets.py +++ b/modules/ui_extra_networks_hypernets.py @@ -31,7 +31,8 @@ class ExtraNetworksPageHypernetworks(ui_extra_networks.ExtraNetworksPage): } def list_items(self): - for index, name in enumerate(shared.hypernetworks): + names = list(shared.hypernetworks) + for index, name in enumerate(names): yield self.create_item(name, index) def allowed_directories_for_previews(self): diff --git a/modules/ui_extra_networks_textual_inversion.py b/modules/ui_extra_networks_textual_inversion.py index 55ef0ea7b..40ab0aca3 100644 --- a/modules/ui_extra_networks_textual_inversion.py +++ b/modules/ui_extra_networks_textual_inversion.py @@ -29,7 +29,8 @@ class ExtraNetworksPageTextualInversion(ui_extra_networks.ExtraNetworksPage): } def list_items(self): - for index, name in enumerate(sd_hijack.model_hijack.embedding_db.word_embeddings): + names = list(sd_hijack.model_hijack.embedding_db.word_embeddings) + for index, name in enumerate(names): yield self.create_item(name, index) def allowed_directories_for_previews(self): From e785402b6acca12108e15224ff80d58817ab3c27 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Sat, 9 Sep 2023 17:28:06 +0900 Subject: [PATCH 005/342] return nothing if not found --- extensions-builtin/Lora/ui_extra_networks_lora.py | 3 ++- modules/ui_extra_networks_checkpoints.py | 7 ++++++- modules/ui_extra_networks_hypernets.py | 9 +++++++-- modules/ui_extra_networks_textual_inversion.py | 6 +++++- 4 files changed, 20 insertions(+), 5 deletions(-) diff --git a/extensions-builtin/Lora/ui_extra_networks_lora.py b/extensions-builtin/Lora/ui_extra_networks_lora.py index e74daa770..dac90a86c 100644 --- a/extensions-builtin/Lora/ui_extra_networks_lora.py +++ b/extensions-builtin/Lora/ui_extra_networks_lora.py @@ -17,6 +17,8 @@ class ExtraNetworksPageLora(ui_extra_networks.ExtraNetworksPage): def create_item(self, name, index=None, enable_filter=True): lora_on_disk = networks.available_networks.get(name) + if lora_on_disk is None: + return path, ext = os.path.splitext(lora_on_disk.filename) @@ -69,7 +71,6 @@ class ExtraNetworksPageLora(ui_extra_networks.ExtraNetworksPage): names = list(networks.available_networks) for index, name in enumerate(names): item = self.create_item(name, index) - if item is not None: yield item diff --git a/modules/ui_extra_networks_checkpoints.py b/modules/ui_extra_networks_checkpoints.py index ca6c26076..35e958a00 100644 --- a/modules/ui_extra_networks_checkpoints.py +++ b/modules/ui_extra_networks_checkpoints.py @@ -15,6 +15,9 @@ class ExtraNetworksPageCheckpoints(ui_extra_networks.ExtraNetworksPage): def create_item(self, name, index=None, enable_filter=True): checkpoint: sd_models.CheckpointInfo = sd_models.checkpoint_aliases.get(name) + if checkpoint is None: + return + path, ext = os.path.splitext(checkpoint.filename) return { "name": checkpoint.name_for_extra, @@ -32,7 +35,9 @@ class ExtraNetworksPageCheckpoints(ui_extra_networks.ExtraNetworksPage): def list_items(self): names = list(sd_models.checkpoints_list) for index, name in enumerate(names): - yield self.create_item(name, index) + item = self.create_item(name, index) + if item is not None: + yield item def allowed_directories_for_previews(self): return [v for v in [shared.cmd_opts.ckpt_dir, sd_models.model_path] if v is not None] diff --git a/modules/ui_extra_networks_hypernets.py b/modules/ui_extra_networks_hypernets.py index 5f5904915..74f7d8472 100644 --- a/modules/ui_extra_networks_hypernets.py +++ b/modules/ui_extra_networks_hypernets.py @@ -13,7 +13,10 @@ class ExtraNetworksPageHypernetworks(ui_extra_networks.ExtraNetworksPage): shared.reload_hypernetworks() def create_item(self, name, index=None, enable_filter=True): - full_path = shared.hypernetworks[name] + full_path = shared.hypernetworks.get(name) + if full_path is None: + return + path, ext = os.path.splitext(full_path) sha256 = sha256_from_cache(full_path, f'hypernet/{name}') shorthash = sha256[0:10] if sha256 else None @@ -33,7 +36,9 @@ class ExtraNetworksPageHypernetworks(ui_extra_networks.ExtraNetworksPage): def list_items(self): names = list(shared.hypernetworks) for index, name in enumerate(names): - yield self.create_item(name, index) + item = self.create_item(name, index) + if item is not None: + yield item def allowed_directories_for_previews(self): return [shared.cmd_opts.hypernetwork_dir] diff --git a/modules/ui_extra_networks_textual_inversion.py b/modules/ui_extra_networks_textual_inversion.py index 40ab0aca3..71c38fabc 100644 --- a/modules/ui_extra_networks_textual_inversion.py +++ b/modules/ui_extra_networks_textual_inversion.py @@ -14,6 +14,8 @@ class ExtraNetworksPageTextualInversion(ui_extra_networks.ExtraNetworksPage): def create_item(self, name, index=None, enable_filter=True): embedding = sd_hijack.model_hijack.embedding_db.word_embeddings.get(name) + if embedding is None: + return path, ext = os.path.splitext(embedding.filename) return { @@ -31,7 +33,9 @@ class ExtraNetworksPageTextualInversion(ui_extra_networks.ExtraNetworksPage): def list_items(self): names = list(sd_hijack.model_hijack.embedding_db.word_embeddings) for index, name in enumerate(names): - yield self.create_item(name, index) + item = self.create_item(name, index) + if item is not None: + yield item def allowed_directories_for_previews(self): return list(sd_hijack.model_hijack.embedding_db.embedding_dirs) From 74b80e72115af46bf1c04167a30f9ec5025cb464 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Tue, 12 Sep 2023 09:29:07 +0900 Subject: [PATCH 006/342] add comment --- extensions-builtin/Lora/ui_extra_networks_lora.py | 1 + modules/ui_extra_networks_checkpoints.py | 1 + modules/ui_extra_networks_hypernets.py | 1 + modules/ui_extra_networks_textual_inversion.py | 1 + 4 files changed, 4 insertions(+) diff --git a/extensions-builtin/Lora/ui_extra_networks_lora.py b/extensions-builtin/Lora/ui_extra_networks_lora.py index dac90a86c..df02c663b 100644 --- a/extensions-builtin/Lora/ui_extra_networks_lora.py +++ b/extensions-builtin/Lora/ui_extra_networks_lora.py @@ -68,6 +68,7 @@ class ExtraNetworksPageLora(ui_extra_networks.ExtraNetworksPage): return item def list_items(self): + # instantiate a list to protect against concurrent modification names = list(networks.available_networks) for index, name in enumerate(names): item = self.create_item(name, index) diff --git a/modules/ui_extra_networks_checkpoints.py b/modules/ui_extra_networks_checkpoints.py index 35e958a00..df7efb2e1 100644 --- a/modules/ui_extra_networks_checkpoints.py +++ b/modules/ui_extra_networks_checkpoints.py @@ -33,6 +33,7 @@ class ExtraNetworksPageCheckpoints(ui_extra_networks.ExtraNetworksPage): } def list_items(self): + # instantiate a list to protect against concurrent modification names = list(sd_models.checkpoints_list) for index, name in enumerate(names): item = self.create_item(name, index) diff --git a/modules/ui_extra_networks_hypernets.py b/modules/ui_extra_networks_hypernets.py index 74f7d8472..c96c4fa3b 100644 --- a/modules/ui_extra_networks_hypernets.py +++ b/modules/ui_extra_networks_hypernets.py @@ -34,6 +34,7 @@ class ExtraNetworksPageHypernetworks(ui_extra_networks.ExtraNetworksPage): } def list_items(self): + # instantiate a list to protect against concurrent modification names = list(shared.hypernetworks) for index, name in enumerate(names): item = self.create_item(name, index) diff --git a/modules/ui_extra_networks_textual_inversion.py b/modules/ui_extra_networks_textual_inversion.py index 71c38fabc..1b334fda1 100644 --- a/modules/ui_extra_networks_textual_inversion.py +++ b/modules/ui_extra_networks_textual_inversion.py @@ -31,6 +31,7 @@ class ExtraNetworksPageTextualInversion(ui_extra_networks.ExtraNetworksPage): } def list_items(self): + # instantiate a list to protect against concurrent modification names = list(sd_hijack.model_hijack.embedding_db.word_embeddings) for index, name in enumerate(names): item = self.create_item(name, index) From 8aa13d5dce2789a7d0bd802e6d62453b3c380496 Mon Sep 17 00:00:00 2001 From: Anthony Fu Date: Mon, 16 Oct 2023 14:12:18 +0800 Subject: [PATCH 007/342] Interrupt after current generation --- modules/call_queue.py | 1 + modules/img2img.py | 2 +- modules/processing.py | 2 +- modules/shared_options.py | 1 + modules/shared_state.py | 11 +++++++++-- scripts/loopback.py | 6 +++--- scripts/xyz_grid.py | 2 +- 7 files changed, 17 insertions(+), 8 deletions(-) diff --git a/modules/call_queue.py b/modules/call_queue.py index ddf0d5738..01c6d17f6 100644 --- a/modules/call_queue.py +++ b/modules/call_queue.py @@ -78,6 +78,7 @@ def wrap_gradio_call(func, extra_outputs=None, add_stats=False): shared.state.skipped = False shared.state.interrupted = False + shared.state.interrupted_next = False shared.state.job_count = 0 if not add_stats: diff --git a/modules/img2img.py b/modules/img2img.py index 52cb577a6..31f8c2aaf 100644 --- a/modules/img2img.py +++ b/modules/img2img.py @@ -49,7 +49,7 @@ def process_batch(p, input_dir, output_dir, inpaint_mask_dir, args, to_scale=Fal if state.skipped: state.skipped = False - if state.interrupted: + if state.interrupted or state.interrupted_next: break try: diff --git a/modules/processing.py b/modules/processing.py index 40598f5cf..e7eecd66d 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -819,7 +819,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: if state.skipped: state.skipped = False - if state.interrupted: + if state.interrupted or state.interrupted_next: break sd_models.reload_model_weights() # model can be changed for example by refiner diff --git a/modules/shared_options.py b/modules/shared_options.py index 32bf73532..4638ef068 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -113,6 +113,7 @@ options_templates.update(options_section(('system', "System"), { "disable_mmap_load_safetensors": OptionInfo(False, "Disable memmapping for loading .safetensors files.").info("fixes very slow loading speed in some cases"), "hide_ldm_prints": OptionInfo(True, "Prevent Stability-AI's ldm/sgm modules from printing noise to console."), "dump_stacks_on_signal": OptionInfo(False, "Print stack traces before exiting the program with ctrl+c."), + "interrupt_after_current": OptionInfo(False, "Interrupt generation after current image is finished on batch processing"), })) options_templates.update(options_section(('API', "API"), { diff --git a/modules/shared_state.py b/modules/shared_state.py index a68789cc8..c72c3f63c 100644 --- a/modules/shared_state.py +++ b/modules/shared_state.py @@ -12,6 +12,7 @@ log = logging.getLogger(__name__) class State: skipped = False interrupted = False + interrupted_next = False job = "" job_no = 0 job_count = 0 @@ -76,8 +77,12 @@ class State: log.info("Received skip request") def interrupt(self): - self.interrupted = True - log.info("Received interrupt request") + if shared.opts.interrupt_after_current and self.job_count > 1: + self.interrupted_next = True + log.info("Received interrupt request, interrupt after current job") + else: + self.interrupted = True + log.info("Received interrupt request") def nextjob(self): if shared.opts.live_previews_enable and shared.opts.show_progress_every_n_steps == -1: @@ -91,6 +96,7 @@ class State: obj = { "skipped": self.skipped, "interrupted": self.interrupted, + "interrupted_next": self.interrupted_next, "job": self.job, "job_count": self.job_count, "job_timestamp": self.job_timestamp, @@ -114,6 +120,7 @@ class State: self.id_live_preview = 0 self.skipped = False self.interrupted = False + self.interrupted_next = False self.textinfo = None self.job = job devices.torch_gc() diff --git a/scripts/loopback.py b/scripts/loopback.py index 2d5feaf9b..ad921269a 100644 --- a/scripts/loopback.py +++ b/scripts/loopback.py @@ -95,7 +95,7 @@ class Script(scripts.Script): processed = processing.process_images(p) # Generation cancelled. - if state.interrupted: + if state.interrupted or state.interrupted_next: break if initial_seed is None: @@ -122,8 +122,8 @@ class Script(scripts.Script): p.inpainting_fill = original_inpainting_fill - if state.interrupted: - break + if state.interrupted or state.interrupted_next: + break if len(history) > 1: grid = images.image_grid(history, rows=1) diff --git a/scripts/xyz_grid.py b/scripts/xyz_grid.py index 0dc255bc4..495008ada 100644 --- a/scripts/xyz_grid.py +++ b/scripts/xyz_grid.py @@ -688,7 +688,7 @@ class Script(scripts.Script): grid_infotext = [None] * (1 + len(zs)) def cell(x, y, z, ix, iy, iz): - if shared.state.interrupted: + if shared.state.interrupted or state.interrupted_next: return Processed(p, [], p.seed, "") pc = copy(p) From 3d15e58b0a30f2ef1e731f9e429f4d3cf1c259c5 Mon Sep 17 00:00:00 2001 From: Anthony Fu Date: Mon, 16 Oct 2023 15:00:17 +0800 Subject: [PATCH 008/342] feat: refactor --- modules/shared_state.py | 12 ++++++------ modules/ui.py | 8 +++++++- 2 files changed, 13 insertions(+), 7 deletions(-) diff --git a/modules/shared_state.py b/modules/shared_state.py index c72c3f63c..532fdcd8d 100644 --- a/modules/shared_state.py +++ b/modules/shared_state.py @@ -77,12 +77,12 @@ class State: log.info("Received skip request") def interrupt(self): - if shared.opts.interrupt_after_current and self.job_count > 1: - self.interrupted_next = True - log.info("Received interrupt request, interrupt after current job") - else: - self.interrupted = True - log.info("Received interrupt request") + self.interrupted = True + log.info("Received interrupt request") + + def interrupt_next(self): + self.interrupted_next = True + log.info("Received interrupt request, interrupt after current job") def nextjob(self): if shared.opts.live_previews_enable and shared.opts.show_progress_every_n_steps == -1: diff --git a/modules/ui.py b/modules/ui.py index bcf391997..c30093d75 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -216,8 +216,14 @@ class Toprow: outputs=[], ) + def interrupt_fn(): + if shared.state.job_count > 1 and shared.opts.interrupt_after_current: + shared.state.interrupt_next() + else: + shared.state.interrupt() + self.interrupt.click( - fn=lambda: shared.state.interrupt(), + fn=interrupt_fn, inputs=[], outputs=[], ) From ec718f76b58b183859ed732e11ec748c41a13f76 Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Tue, 17 Oct 2023 23:35:50 -0700 Subject: [PATCH 009/342] wip incorrect OFT implementation --- extensions-builtin/Lora/network_oft.py | 82 ++++++++++++++++++++++++++ extensions-builtin/Lora/networks.py | 5 ++ 2 files changed, 87 insertions(+) create mode 100644 extensions-builtin/Lora/network_oft.py diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py new file mode 100644 index 000000000..9ddb175ce --- /dev/null +++ b/extensions-builtin/Lora/network_oft.py @@ -0,0 +1,82 @@ +import torch +import network + + +class ModuleTypeOFT(network.ModuleType): + def create_module(self, net: network.Network, weights: network.NetworkWeights): + if all(x in weights.w for x in ["oft_blocks"]): + return NetworkModuleOFT(net, weights) + + return None + +# adapted from https://github.com/kohya-ss/sd-scripts/blob/main/networks/oft.py +class NetworkModuleOFT(network.NetworkModule): + def __init__(self, net: network.Network, weights: network.NetworkWeights): + super().__init__(net, weights) + + self.oft_blocks = weights.w["oft_blocks"] + self.alpha = weights.w["alpha"] + + self.dim = self.oft_blocks.shape[0] + self.num_blocks = self.dim + + #if type(self.alpha) == torch.Tensor: + # self.alpha = self.alpha.detach().numpy() + + if "Linear" in self.sd_module.__class__.__name__: + self.out_dim = self.sd_module.out_features + elif "Conv" in self.sd_module.__class__.__name__: + self.out_dim = self.sd_module.out_channels + + self.constraint = self.alpha * self.out_dim + self.block_size = self.out_dim // self.num_blocks + + self.oft_multiplier = self.multiplier() + + # replace forward method of original linear rather than replacing the module + # self.org_forward = self.sd_module.forward + # self.sd_module.forward = self.forward + + def get_weight(self): + block_Q = self.oft_blocks - self.oft_blocks.transpose(1, 2) + norm_Q = torch.norm(block_Q.flatten()) + new_norm_Q = torch.clamp(norm_Q, max=self.constraint) + block_Q = block_Q * ((new_norm_Q + 1e-8) / (norm_Q + 1e-8)) + I = torch.eye(self.block_size, device=self.oft_blocks.device).unsqueeze(0).repeat(self.num_blocks, 1, 1) + block_R = torch.matmul(I + block_Q, (I - block_Q).inverse()) + + block_R_weighted = self.oft_multiplier * block_R + (1 - self.oft_multiplier) * I + R = torch.block_diag(*block_R_weighted) + + return R + + def calc_updown(self, orig_weight): + oft_blocks = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) + block_Q = oft_blocks - oft_blocks.transpose(1, 2) + norm_Q = torch.norm(block_Q.flatten()) + new_norm_Q = torch.clamp(norm_Q, max=self.constraint) + block_Q = block_Q * ((new_norm_Q + 1e-8) / (norm_Q + 1e-8)) + I = torch.eye(self.block_size, device=oft_blocks.device).unsqueeze(0).repeat(self.num_blocks, 1, 1) + block_R = torch.matmul(I + block_Q, (I - block_Q).inverse()) + + block_R_weighted = self.oft_multiplier * block_R + (1 - self.oft_multiplier) * I + R = torch.block_diag(*block_R_weighted) + #R = self.get_weight().to(orig_weight.device, dtype=orig_weight.dtype) + # W = R*W_0 + updown = orig_weight + R + output_shape = [R.size(0), orig_weight.size(1)] + return self.finalize_updown(updown, orig_weight, output_shape) + + # def forward(self, x, y=None): + # x = self.org_forward(x) + # if self.oft_multiplier == 0.0: + # return x + + # R = self.get_weight().to(x.device, dtype=x.dtype) + # if x.dim() == 4: + # x = x.permute(0, 2, 3, 1) + # x = torch.matmul(x, R) + # x = x.permute(0, 3, 1, 2) + # else: + # x = torch.matmul(x, R) + # return x diff --git a/extensions-builtin/Lora/networks.py b/extensions-builtin/Lora/networks.py index 60d8dec4c..bd1f1b756 100644 --- a/extensions-builtin/Lora/networks.py +++ b/extensions-builtin/Lora/networks.py @@ -11,6 +11,7 @@ import network_ia3 import network_lokr import network_full import network_norm +import network_oft import torch from typing import Union @@ -28,6 +29,7 @@ module_types = [ network_full.ModuleTypeFull(), network_norm.ModuleTypeNorm(), network_glora.ModuleTypeGLora(), + network_oft.ModuleTypeOFT(), ] @@ -183,6 +185,9 @@ def load_network(name, network_on_disk): elif sd_module is None and "lora_te1_text_model" in key_network_without_network_parts: key = key_network_without_network_parts.replace("lora_te1_text_model", "0_transformer_text_model") sd_module = shared.sd_model.network_layer_mapping.get(key, None) + elif sd_module is None and "oft_unet" in key_network_without_network_parts: + key = key_network_without_network_parts.replace("oft_unet", "diffusion_model") + sd_module = shared.sd_model.network_layer_mapping.get(key, None) # some SD1 Loras also have correct compvis keys if sd_module is None: From 1c6efdbba774d603c592debaccd6f5ad827bd1b2 Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Wed, 18 Oct 2023 04:16:01 -0700 Subject: [PATCH 010/342] inference working but SLOW --- extensions-builtin/Lora/network_oft.py | 73 +++++++++++++------------- extensions-builtin/Lora/networks.py | 42 +++++++++++++-- 2 files changed, 75 insertions(+), 40 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index 9ddb175ce..f085eca53 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -12,6 +12,7 @@ class ModuleTypeOFT(network.ModuleType): # adapted from https://github.com/kohya-ss/sd-scripts/blob/main/networks/oft.py class NetworkModuleOFT(network.NetworkModule): def __init__(self, net: network.Network, weights: network.NetworkWeights): + super().__init__(net, weights) self.oft_blocks = weights.w["oft_blocks"] @@ -20,24 +21,29 @@ class NetworkModuleOFT(network.NetworkModule): self.dim = self.oft_blocks.shape[0] self.num_blocks = self.dim - #if type(self.alpha) == torch.Tensor: - # self.alpha = self.alpha.detach().numpy() - if "Linear" in self.sd_module.__class__.__name__: self.out_dim = self.sd_module.out_features elif "Conv" in self.sd_module.__class__.__name__: self.out_dim = self.sd_module.out_channels - self.constraint = self.alpha * self.out_dim + self.constraint = self.alpha + #self.constraint = self.alpha * self.out_dim self.block_size = self.out_dim // self.num_blocks - self.oft_multiplier = self.multiplier() + self.org_module: list[torch.Module] = [self.sd_module] - # replace forward method of original linear rather than replacing the module - # self.org_forward = self.sd_module.forward - # self.sd_module.forward = self.forward + self.R = self.get_weight() + + self.apply_to() + + # replace forward method of original linear rather than replacing the module + def apply_to(self): + self.org_forward = self.org_module[0].forward + self.org_module[0].forward = self.forward - def get_weight(self): + def get_weight(self, multiplier=None): + if not multiplier: + multiplier = self.multiplier() block_Q = self.oft_blocks - self.oft_blocks.transpose(1, 2) norm_Q = torch.norm(block_Q.flatten()) new_norm_Q = torch.clamp(norm_Q, max=self.constraint) @@ -45,38 +51,31 @@ class NetworkModuleOFT(network.NetworkModule): I = torch.eye(self.block_size, device=self.oft_blocks.device).unsqueeze(0).repeat(self.num_blocks, 1, 1) block_R = torch.matmul(I + block_Q, (I - block_Q).inverse()) - block_R_weighted = self.oft_multiplier * block_R + (1 - self.oft_multiplier) * I + block_R_weighted = multiplier * block_R + (1 - multiplier) * I R = torch.block_diag(*block_R_weighted) return R def calc_updown(self, orig_weight): - oft_blocks = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) - block_Q = oft_blocks - oft_blocks.transpose(1, 2) - norm_Q = torch.norm(block_Q.flatten()) - new_norm_Q = torch.clamp(norm_Q, max=self.constraint) - block_Q = block_Q * ((new_norm_Q + 1e-8) / (norm_Q + 1e-8)) - I = torch.eye(self.block_size, device=oft_blocks.device).unsqueeze(0).repeat(self.num_blocks, 1, 1) - block_R = torch.matmul(I + block_Q, (I - block_Q).inverse()) - - block_R_weighted = self.oft_multiplier * block_R + (1 - self.oft_multiplier) * I - R = torch.block_diag(*block_R_weighted) - #R = self.get_weight().to(orig_weight.device, dtype=orig_weight.dtype) - # W = R*W_0 - updown = orig_weight + R - output_shape = [R.size(0), orig_weight.size(1)] + R = self.R + if orig_weight.dim() == 4: + weight = torch.einsum("oihw, op -> pihw", orig_weight, R) + else: + weight = torch.einsum("oi, op -> pi", orig_weight, R) + updown = orig_weight @ R + output_shape = [orig_weight.size(0), R.size(1)] + #output_shape = [R.size(0), orig_weight.size(1)] return self.finalize_updown(updown, orig_weight, output_shape) - # def forward(self, x, y=None): - # x = self.org_forward(x) - # if self.oft_multiplier == 0.0: - # return x - - # R = self.get_weight().to(x.device, dtype=x.dtype) - # if x.dim() == 4: - # x = x.permute(0, 2, 3, 1) - # x = torch.matmul(x, R) - # x = x.permute(0, 3, 1, 2) - # else: - # x = torch.matmul(x, R) - # return x + def forward(self, x, y=None): + x = self.org_forward(x) + if self.multiplier() == 0.0: + return x + R = self.get_weight().to(x.device, dtype=x.dtype) + if x.dim() == 4: + x = x.permute(0, 2, 3, 1) + x = torch.matmul(x, R) + x = x.permute(0, 3, 1, 2) + else: + x = torch.matmul(x, R) + return x diff --git a/extensions-builtin/Lora/networks.py b/extensions-builtin/Lora/networks.py index bd1f1b756..e5e73450b 100644 --- a/extensions-builtin/Lora/networks.py +++ b/extensions-builtin/Lora/networks.py @@ -169,6 +169,10 @@ def load_network(name, network_on_disk): else: emb_dict[vec_name] = weight bundle_embeddings[emb_name] = emb_dict + + #if key_network_without_network_parts == "oft_unet": + # print(key_network_without_network_parts) + # pass key = convert_diffusers_name_to_compvis(key_network_without_network_parts, is_sd2) sd_module = shared.sd_model.network_layer_mapping.get(key, None) @@ -185,15 +189,39 @@ def load_network(name, network_on_disk): elif sd_module is None and "lora_te1_text_model" in key_network_without_network_parts: key = key_network_without_network_parts.replace("lora_te1_text_model", "0_transformer_text_model") sd_module = shared.sd_model.network_layer_mapping.get(key, None) - elif sd_module is None and "oft_unet" in key_network_without_network_parts: - key = key_network_without_network_parts.replace("oft_unet", "diffusion_model") - sd_module = shared.sd_model.network_layer_mapping.get(key, None) # some SD1 Loras also have correct compvis keys if sd_module is None: key = key_network_without_network_parts.replace("lora_te1_text_model", "transformer_text_model") sd_module = shared.sd_model.network_layer_mapping.get(key, None) + elif sd_module is None and "oft_unet" in key_network_without_network_parts: + # UNET_TARGET_REPLACE_MODULE_ALL_LINEAR = ["Transformer2DModel"] + # UNET_TARGET_REPLACE_MODULE_CONV2D_3X3 = ["ResnetBlock2D", "Downsample2D", "Upsample2D"] + UNET_TARGET_REPLACE_MODULE_ATTN_ONLY = ["CrossAttention"] + # TODO: Change matchedm odules based on whether all linear, conv, etc + + key = key_network_without_network_parts.replace("oft_unet", "diffusion_model") + sd_module = shared.sd_model.network_layer_mapping.get(key, None) + #key_no_suffix = key.rsplit("_to_", 1)[0] + ## Match all modules of class CrossAttention + #replace_module_list = [] + #for module_type in UNET_TARGET_REPLACE_MODULE_ATTN_ONLY: + # replace_module_list += [module for k, module in shared.sd_model.network_layer_mapping.items() if module_type in module.__class__.__name__] + + #matched_module = replace_module_list.get(key_no_suffix, None) + #if key.endswith('to_q'): + # sd_module = matched_module.to_q or None + #if key.endswith('to_k'): + # sd_module = matched_module.to_k or None + #if key.endswith('to_v'): + # sd_module = matched_module.to_v or None + #if key.endswith('to_out_0'): + # sd_module = matched_module.to_out[0] or None + #if key.endswith('to_out_1'): + # sd_module = matched_module.to_out[1] or None + + if sd_module is None: keys_failed_to_match[key_network] = key continue @@ -214,6 +242,14 @@ def load_network(name, network_on_disk): raise AssertionError(f"Could not find a module type (out of {', '.join([x.__class__.__name__ for x in module_types])}) that would accept those keys: {', '.join(weights.w)}") net.modules[key] = net_module + + # replaces forward method of original Linear + # applied_to_count = 0 + #for key, created_module in net.modules.items(): + # if isinstance(created_module, network_oft.NetworkModuleOFT): + # net_module.apply_to() + #applied_to_count += 1 + # print(f'Applied OFT modules: {applied_to_count}') embeddings = {} for emb_name, data in bundle_embeddings.items(): From 853e21d98eada4db9a9fd1ae8eda90cf763e2818 Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Wed, 18 Oct 2023 04:27:44 -0700 Subject: [PATCH 011/342] faster by using cached R in forward --- extensions-builtin/Lora/network_oft.py | 17 ++++++++++++++--- 1 file changed, 14 insertions(+), 3 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index f085eca53..68efb1db9 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -57,21 +57,32 @@ class NetworkModuleOFT(network.NetworkModule): return R def calc_updown(self, orig_weight): + # this works R = self.R + + # this causes major deepfrying i.e. just doesn't work + # R = self.R.to(orig_weight.device, dtype=orig_weight.dtype) + if orig_weight.dim() == 4: weight = torch.einsum("oihw, op -> pihw", orig_weight, R) else: weight = torch.einsum("oi, op -> pi", orig_weight, R) + updown = orig_weight @ R - output_shape = [orig_weight.size(0), R.size(1)] - #output_shape = [R.size(0), orig_weight.size(1)] + output_shape = self.oft_blocks.shape + + ## this works + # updown = orig_weight @ R + # output_shape = [orig_weight.size(0), R.size(1)] + return self.finalize_updown(updown, orig_weight, output_shape) def forward(self, x, y=None): x = self.org_forward(x) if self.multiplier() == 0.0: return x - R = self.get_weight().to(x.device, dtype=x.dtype) + #R = self.get_weight().to(x.device, dtype=x.dtype) + R = self.R.to(x.device, dtype=x.dtype) if x.dim() == 4: x = x.permute(0, 2, 3, 1) x = torch.matmul(x, R) From eb01d7f0e0fb46285985803296a25715165fb3f9 Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Wed, 18 Oct 2023 04:56:53 -0700 Subject: [PATCH 012/342] faster by calculating R in updown and using cached R in forward --- extensions-builtin/Lora/network_oft.py | 15 ++++++++------- 1 file changed, 8 insertions(+), 7 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index 68efb1db9..fd5b0c0fd 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -58,17 +58,18 @@ class NetworkModuleOFT(network.NetworkModule): def calc_updown(self, orig_weight): # this works - R = self.R + # R = self.R + self.R = self.get_weight(self.multiplier()) - # this causes major deepfrying i.e. just doesn't work + # sending R to device causes major deepfrying i.e. just doesn't work # R = self.R.to(orig_weight.device, dtype=orig_weight.dtype) - if orig_weight.dim() == 4: - weight = torch.einsum("oihw, op -> pihw", orig_weight, R) - else: - weight = torch.einsum("oi, op -> pi", orig_weight, R) + # if orig_weight.dim() == 4: + # weight = torch.einsum("oihw, op -> pihw", orig_weight, R) + # else: + # weight = torch.einsum("oi, op -> pi", orig_weight, R) - updown = orig_weight @ R + updown = orig_weight @ self.R output_shape = self.oft_blocks.shape ## this works From 7c128bbdac0da1767c239174e91af6f327845372 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Thu, 19 Oct 2023 13:56:17 +0800 Subject: [PATCH 013/342] Add fp8 for sd unet --- extensions-builtin/Lora/network.py | 2 +- extensions-builtin/Lora/network_full.py | 4 ++-- extensions-builtin/Lora/network_glora.py | 10 +++++----- extensions-builtin/Lora/network_hada.py | 12 ++++++------ extensions-builtin/Lora/network_ia3.py | 2 +- extensions-builtin/Lora/network_lokr.py | 18 +++++++++--------- extensions-builtin/Lora/network_lora.py | 6 +++--- extensions-builtin/Lora/network_norm.py | 4 ++-- extensions-builtin/Lora/networks.py | 6 +++--- modules/cmd_args.py | 1 + modules/sd_models.py | 3 +++ 11 files changed, 36 insertions(+), 32 deletions(-) diff --git a/extensions-builtin/Lora/network.py b/extensions-builtin/Lora/network.py index 6021fd8de..a62e5eff9 100644 --- a/extensions-builtin/Lora/network.py +++ b/extensions-builtin/Lora/network.py @@ -137,7 +137,7 @@ class NetworkModule: def finalize_updown(self, updown, orig_weight, output_shape, ex_bias=None): if self.bias is not None: updown = updown.reshape(self.bias.shape) - updown += self.bias.to(orig_weight.device, dtype=orig_weight.dtype) + updown += self.bias.to(orig_weight.device, dtype=updown.dtype) updown = updown.reshape(output_shape) if len(output_shape) == 4: diff --git a/extensions-builtin/Lora/network_full.py b/extensions-builtin/Lora/network_full.py index bf6930e96..f221c95f3 100644 --- a/extensions-builtin/Lora/network_full.py +++ b/extensions-builtin/Lora/network_full.py @@ -18,9 +18,9 @@ class NetworkModuleFull(network.NetworkModule): def calc_updown(self, orig_weight): output_shape = self.weight.shape - updown = self.weight.to(orig_weight.device, dtype=orig_weight.dtype) + updown = self.weight.to(orig_weight.device) if self.ex_bias is not None: - ex_bias = self.ex_bias.to(orig_weight.device, dtype=orig_weight.dtype) + ex_bias = self.ex_bias.to(orig_weight.device) else: ex_bias = None diff --git a/extensions-builtin/Lora/network_glora.py b/extensions-builtin/Lora/network_glora.py index 492d48707..efe5c6814 100644 --- a/extensions-builtin/Lora/network_glora.py +++ b/extensions-builtin/Lora/network_glora.py @@ -22,12 +22,12 @@ class NetworkModuleGLora(network.NetworkModule): self.w2b = weights.w["b2.weight"] def calc_updown(self, orig_weight): - w1a = self.w1a.to(orig_weight.device, dtype=orig_weight.dtype) - w1b = self.w1b.to(orig_weight.device, dtype=orig_weight.dtype) - w2a = self.w2a.to(orig_weight.device, dtype=orig_weight.dtype) - w2b = self.w2b.to(orig_weight.device, dtype=orig_weight.dtype) + w1a = self.w1a.to(orig_weight.device) + w1b = self.w1b.to(orig_weight.device) + w2a = self.w2a.to(orig_weight.device) + w2b = self.w2b.to(orig_weight.device) output_shape = [w1a.size(0), w1b.size(1)] - updown = ((w2b @ w1b) + ((orig_weight @ w2a) @ w1a)) + updown = ((w2b @ w1b) + ((orig_weight.to(dtype = w1a.dtype) @ w2a) @ w1a)) return self.finalize_updown(updown, orig_weight, output_shape) diff --git a/extensions-builtin/Lora/network_hada.py b/extensions-builtin/Lora/network_hada.py index 5fcb0695f..d95a0fd18 100644 --- a/extensions-builtin/Lora/network_hada.py +++ b/extensions-builtin/Lora/network_hada.py @@ -27,16 +27,16 @@ class NetworkModuleHada(network.NetworkModule): self.t2 = weights.w.get("hada_t2") def calc_updown(self, orig_weight): - w1a = self.w1a.to(orig_weight.device, dtype=orig_weight.dtype) - w1b = self.w1b.to(orig_weight.device, dtype=orig_weight.dtype) - w2a = self.w2a.to(orig_weight.device, dtype=orig_weight.dtype) - w2b = self.w2b.to(orig_weight.device, dtype=orig_weight.dtype) + w1a = self.w1a.to(orig_weight.device) + w1b = self.w1b.to(orig_weight.device) + w2a = self.w2a.to(orig_weight.device) + w2b = self.w2b.to(orig_weight.device) output_shape = [w1a.size(0), w1b.size(1)] if self.t1 is not None: output_shape = [w1a.size(1), w1b.size(1)] - t1 = self.t1.to(orig_weight.device, dtype=orig_weight.dtype) + t1 = self.t1.to(orig_weight.device) updown1 = lyco_helpers.make_weight_cp(t1, w1a, w1b) output_shape += t1.shape[2:] else: @@ -45,7 +45,7 @@ class NetworkModuleHada(network.NetworkModule): updown1 = lyco_helpers.rebuild_conventional(w1a, w1b, output_shape) if self.t2 is not None: - t2 = self.t2.to(orig_weight.device, dtype=orig_weight.dtype) + t2 = self.t2.to(orig_weight.device) updown2 = lyco_helpers.make_weight_cp(t2, w2a, w2b) else: updown2 = lyco_helpers.rebuild_conventional(w2a, w2b, output_shape) diff --git a/extensions-builtin/Lora/network_ia3.py b/extensions-builtin/Lora/network_ia3.py index 7edc42497..96faeaf3e 100644 --- a/extensions-builtin/Lora/network_ia3.py +++ b/extensions-builtin/Lora/network_ia3.py @@ -17,7 +17,7 @@ class NetworkModuleIa3(network.NetworkModule): self.on_input = weights.w["on_input"].item() def calc_updown(self, orig_weight): - w = self.w.to(orig_weight.device, dtype=orig_weight.dtype) + w = self.w.to(orig_weight.device) output_shape = [w.size(0), orig_weight.size(1)] if self.on_input: diff --git a/extensions-builtin/Lora/network_lokr.py b/extensions-builtin/Lora/network_lokr.py index 340acdab3..fcdaeafd8 100644 --- a/extensions-builtin/Lora/network_lokr.py +++ b/extensions-builtin/Lora/network_lokr.py @@ -37,22 +37,22 @@ class NetworkModuleLokr(network.NetworkModule): def calc_updown(self, orig_weight): if self.w1 is not None: - w1 = self.w1.to(orig_weight.device, dtype=orig_weight.dtype) + w1 = self.w1.to(orig_weight.device) else: - w1a = self.w1a.to(orig_weight.device, dtype=orig_weight.dtype) - w1b = self.w1b.to(orig_weight.device, dtype=orig_weight.dtype) + w1a = self.w1a.to(orig_weight.device) + w1b = self.w1b.to(orig_weight.device) w1 = w1a @ w1b if self.w2 is not None: - w2 = self.w2.to(orig_weight.device, dtype=orig_weight.dtype) + w2 = self.w2.to(orig_weight.device) elif self.t2 is None: - w2a = self.w2a.to(orig_weight.device, dtype=orig_weight.dtype) - w2b = self.w2b.to(orig_weight.device, dtype=orig_weight.dtype) + w2a = self.w2a.to(orig_weight.device) + w2b = self.w2b.to(orig_weight.device) w2 = w2a @ w2b else: - t2 = self.t2.to(orig_weight.device, dtype=orig_weight.dtype) - w2a = self.w2a.to(orig_weight.device, dtype=orig_weight.dtype) - w2b = self.w2b.to(orig_weight.device, dtype=orig_weight.dtype) + t2 = self.t2.to(orig_weight.device) + w2a = self.w2a.to(orig_weight.device) + w2b = self.w2b.to(orig_weight.device) w2 = lyco_helpers.make_weight_cp(t2, w2a, w2b) output_shape = [w1.size(0) * w2.size(0), w1.size(1) * w2.size(1)] diff --git a/extensions-builtin/Lora/network_lora.py b/extensions-builtin/Lora/network_lora.py index 26c0a72c2..4cc402951 100644 --- a/extensions-builtin/Lora/network_lora.py +++ b/extensions-builtin/Lora/network_lora.py @@ -61,13 +61,13 @@ class NetworkModuleLora(network.NetworkModule): return module def calc_updown(self, orig_weight): - up = self.up_model.weight.to(orig_weight.device, dtype=orig_weight.dtype) - down = self.down_model.weight.to(orig_weight.device, dtype=orig_weight.dtype) + up = self.up_model.weight.to(orig_weight.device) + down = self.down_model.weight.to(orig_weight.device) output_shape = [up.size(0), down.size(1)] if self.mid_model is not None: # cp-decomposition - mid = self.mid_model.weight.to(orig_weight.device, dtype=orig_weight.dtype) + mid = self.mid_model.weight.to(orig_weight.device) updown = lyco_helpers.rebuild_cp_decomposition(up, down, mid) output_shape += mid.shape[2:] else: diff --git a/extensions-builtin/Lora/network_norm.py b/extensions-builtin/Lora/network_norm.py index ce4501580..d25afcbb9 100644 --- a/extensions-builtin/Lora/network_norm.py +++ b/extensions-builtin/Lora/network_norm.py @@ -18,10 +18,10 @@ class NetworkModuleNorm(network.NetworkModule): def calc_updown(self, orig_weight): output_shape = self.w_norm.shape - updown = self.w_norm.to(orig_weight.device, dtype=orig_weight.dtype) + updown = self.w_norm.to(orig_weight.device) if self.b_norm is not None: - ex_bias = self.b_norm.to(orig_weight.device, dtype=orig_weight.dtype) + ex_bias = self.b_norm.to(orig_weight.device) else: ex_bias = None diff --git a/extensions-builtin/Lora/networks.py b/extensions-builtin/Lora/networks.py index 60d8dec4c..8ea4ea609 100644 --- a/extensions-builtin/Lora/networks.py +++ b/extensions-builtin/Lora/networks.py @@ -381,12 +381,12 @@ def network_apply_weights(self: Union[torch.nn.Conv2d, torch.nn.Linear, torch.nn # inpainting model. zero pad updown to make channel[1] 4 to 9 updown = torch.nn.functional.pad(updown, (0, 0, 0, 0, 0, 5)) - self.weight += updown + self.weight.copy_((self.weight.to(dtype=updown.dtype) + updown).to(dtype=self.weight.dtype)) if ex_bias is not None and hasattr(self, 'bias'): if self.bias is None: - self.bias = torch.nn.Parameter(ex_bias) + self.bias = torch.nn.Parameter(ex_bias).to(self.weight.dtype) else: - self.bias += ex_bias + self.bias.copy_((self.bias.to(dtype=ex_bias.dtype) + ex_bias).to(dtype=self.bias.dtype)) except RuntimeError as e: logging.debug(f"Network {net.name} layer {network_layer_name}: {e}") extra_network_lora.errors[net.name] = extra_network_lora.errors.get(net.name, 0) + 1 diff --git a/modules/cmd_args.py b/modules/cmd_args.py index 4e602a842..0f14c71e4 100644 --- a/modules/cmd_args.py +++ b/modules/cmd_args.py @@ -118,3 +118,4 @@ parser.add_argument('--timeout-keep-alive', type=int, default=30, help='set time parser.add_argument("--disable-all-extensions", action='store_true', help="prevent all extensions from running regardless of any other settings", default=False) parser.add_argument("--disable-extra-extensions", action='store_true', help="prevent all extensions except built-in from running regardless of any other settings", default=False) parser.add_argument("--skip-load-model-at-start", action='store_true', help="if load a model at web start, only take effect when --nowebui", ) +parser.add_argument("--opt-unet-fp8-storage", action='store_true', help="use fp8 for SD UNet to save vram", default=False) diff --git a/modules/sd_models.py b/modules/sd_models.py index 3b6cdea18..3b8ff8209 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -391,6 +391,9 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer devices.dtype_unet = torch.float16 timer.record("apply half()") + if shared.cmd_opts.opt_unet_fp8_storage: + model.model.diffusion_model = model.model.diffusion_model.to(torch.float8_e4m3fn) + timer.record("apply fp8 unet") devices.unet_needs_upcast = shared.cmd_opts.upcast_sampling and devices.dtype == torch.float16 and devices.dtype_unet == torch.float16 From 5f9ddfa46f28ca2aa9e0bd832f6bbd67069be63e Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Thu, 19 Oct 2023 23:57:22 +0800 Subject: [PATCH 014/342] Add sdxl only arg --- modules/cmd_args.py | 1 + modules/sd_models.py | 3 +++ 2 files changed, 4 insertions(+) diff --git a/modules/cmd_args.py b/modules/cmd_args.py index 0f14c71e4..20bfb2c44 100644 --- a/modules/cmd_args.py +++ b/modules/cmd_args.py @@ -119,3 +119,4 @@ parser.add_argument("--disable-all-extensions", action='store_true', help="preve parser.add_argument("--disable-extra-extensions", action='store_true', help="prevent all extensions except built-in from running regardless of any other settings", default=False) parser.add_argument("--skip-load-model-at-start", action='store_true', help="if load a model at web start, only take effect when --nowebui", ) parser.add_argument("--opt-unet-fp8-storage", action='store_true', help="use fp8 for SD UNet to save vram", default=False) +parser.add_argument("--opt-unet-fp8-storage-xl", action='store_true', help="use fp8 for SD UNet to save vram", default=False) diff --git a/modules/sd_models.py b/modules/sd_models.py index 3b8ff8209..08af128fc 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -394,6 +394,9 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer if shared.cmd_opts.opt_unet_fp8_storage: model.model.diffusion_model = model.model.diffusion_model.to(torch.float8_e4m3fn) timer.record("apply fp8 unet") + elif model.is_sdxl and shared.cmd_opts.opt_unet_fp8_storage_xl: + model.model.diffusion_model = model.model.diffusion_model.to(torch.float8_e4m3fn) + timer.record("apply fp8 unet for sdxl") devices.unet_needs_upcast = shared.cmd_opts.upcast_sampling and devices.dtype == torch.float16 and devices.dtype_unet == torch.float16 From 321680ccd0e0404223fbdf4f26498f7d0317fb75 Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Thu, 19 Oct 2023 12:41:17 -0700 Subject: [PATCH 015/342] refactor: fix constraint, re-use get_weight --- extensions-builtin/Lora/network_oft.py | 40 +++++++++++--------------- 1 file changed, 16 insertions(+), 24 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index fd5b0c0fd..2af1bc4cf 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -9,7 +9,7 @@ class ModuleTypeOFT(network.ModuleType): return None -# adapted from https://github.com/kohya-ss/sd-scripts/blob/main/networks/oft.py +# adapted from kohya's implementation https://github.com/kohya-ss/sd-scripts/blob/main/networks/oft.py class NetworkModuleOFT(network.NetworkModule): def __init__(self, net: network.Network, weights: network.NetworkWeights): @@ -17,7 +17,6 @@ class NetworkModuleOFT(network.NetworkModule): self.oft_blocks = weights.w["oft_blocks"] self.alpha = weights.w["alpha"] - self.dim = self.oft_blocks.shape[0] self.num_blocks = self.dim @@ -26,64 +25,57 @@ class NetworkModuleOFT(network.NetworkModule): elif "Conv" in self.sd_module.__class__.__name__: self.out_dim = self.sd_module.out_channels - self.constraint = self.alpha - #self.constraint = self.alpha * self.out_dim + self.constraint = self.alpha * self.out_dim self.block_size = self.out_dim // self.num_blocks self.org_module: list[torch.Module] = [self.sd_module] - - self.R = self.get_weight() - + self.R = self.get_weight(self.oft_blocks) self.apply_to() # replace forward method of original linear rather than replacing the module + # how do we revert this to unload the weights? def apply_to(self): self.org_forward = self.org_module[0].forward self.org_module[0].forward = self.forward - def get_weight(self, multiplier=None): - if not multiplier: - multiplier = self.multiplier() - block_Q = self.oft_blocks - self.oft_blocks.transpose(1, 2) + def get_weight(self, oft_blocks, multiplier=None): + block_Q = oft_blocks - oft_blocks.transpose(1, 2) norm_Q = torch.norm(block_Q.flatten()) new_norm_Q = torch.clamp(norm_Q, max=self.constraint) block_Q = block_Q * ((new_norm_Q + 1e-8) / (norm_Q + 1e-8)) I = torch.eye(self.block_size, device=self.oft_blocks.device).unsqueeze(0).repeat(self.num_blocks, 1, 1) block_R = torch.matmul(I + block_Q, (I - block_Q).inverse()) - - block_R_weighted = multiplier * block_R + (1 - multiplier) * I - R = torch.block_diag(*block_R_weighted) + #block_R_weighted = multiplier * block_R + (1 - multiplier) * I + #R = torch.block_diag(*block_R_weighted) + R = torch.block_diag(*block_R) return R def calc_updown(self, orig_weight): - # this works - # R = self.R - self.R = self.get_weight(self.multiplier()) + oft_blocks = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) - # sending R to device causes major deepfrying i.e. just doesn't work - # R = self.R.to(orig_weight.device, dtype=orig_weight.dtype) + R = self.get_weight(oft_blocks) + self.R = R # if orig_weight.dim() == 4: # weight = torch.einsum("oihw, op -> pihw", orig_weight, R) # else: # weight = torch.einsum("oi, op -> pi", orig_weight, R) - updown = orig_weight @ self.R + updown = orig_weight @ R output_shape = self.oft_blocks.shape - ## this works - # updown = orig_weight @ R - # output_shape = [orig_weight.size(0), R.size(1)] - return self.finalize_updown(updown, orig_weight, output_shape) def forward(self, x, y=None): x = self.org_forward(x) if self.multiplier() == 0.0: return x + + # calculating R here is excruciatingly slow #R = self.get_weight().to(x.device, dtype=x.dtype) R = self.R.to(x.device, dtype=x.dtype) + if x.dim() == 4: x = x.permute(0, 2, 3, 1) x = torch.matmul(x, R) From d10c4db57ed08234a7aed5f530f269ff78544ab0 Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Thu, 19 Oct 2023 12:52:14 -0700 Subject: [PATCH 016/342] style: formatting --- extensions-builtin/Lora/network_oft.py | 4 +-- extensions-builtin/Lora/networks.py | 35 -------------------------- 2 files changed, 2 insertions(+), 37 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index 2af1bc4cf..0a87958e2 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -37,7 +37,7 @@ class NetworkModuleOFT(network.NetworkModule): def apply_to(self): self.org_forward = self.org_module[0].forward self.org_module[0].forward = self.forward - + def get_weight(self, oft_blocks, multiplier=None): block_Q = oft_blocks - oft_blocks.transpose(1, 2) norm_Q = torch.norm(block_Q.flatten()) @@ -66,7 +66,7 @@ class NetworkModuleOFT(network.NetworkModule): output_shape = self.oft_blocks.shape return self.finalize_updown(updown, orig_weight, output_shape) - + def forward(self, x, y=None): x = self.org_forward(x) if self.multiplier() == 0.0: diff --git a/extensions-builtin/Lora/networks.py b/extensions-builtin/Lora/networks.py index e5e73450b..78a97033d 100644 --- a/extensions-builtin/Lora/networks.py +++ b/extensions-builtin/Lora/networks.py @@ -169,10 +169,6 @@ def load_network(name, network_on_disk): else: emb_dict[vec_name] = weight bundle_embeddings[emb_name] = emb_dict - - #if key_network_without_network_parts == "oft_unet": - # print(key_network_without_network_parts) - # pass key = convert_diffusers_name_to_compvis(key_network_without_network_parts, is_sd2) sd_module = shared.sd_model.network_layer_mapping.get(key, None) @@ -196,31 +192,8 @@ def load_network(name, network_on_disk): sd_module = shared.sd_model.network_layer_mapping.get(key, None) elif sd_module is None and "oft_unet" in key_network_without_network_parts: - # UNET_TARGET_REPLACE_MODULE_ALL_LINEAR = ["Transformer2DModel"] - # UNET_TARGET_REPLACE_MODULE_CONV2D_3X3 = ["ResnetBlock2D", "Downsample2D", "Upsample2D"] - UNET_TARGET_REPLACE_MODULE_ATTN_ONLY = ["CrossAttention"] - # TODO: Change matchedm odules based on whether all linear, conv, etc - key = key_network_without_network_parts.replace("oft_unet", "diffusion_model") sd_module = shared.sd_model.network_layer_mapping.get(key, None) - #key_no_suffix = key.rsplit("_to_", 1)[0] - ## Match all modules of class CrossAttention - #replace_module_list = [] - #for module_type in UNET_TARGET_REPLACE_MODULE_ATTN_ONLY: - # replace_module_list += [module for k, module in shared.sd_model.network_layer_mapping.items() if module_type in module.__class__.__name__] - - #matched_module = replace_module_list.get(key_no_suffix, None) - #if key.endswith('to_q'): - # sd_module = matched_module.to_q or None - #if key.endswith('to_k'): - # sd_module = matched_module.to_k or None - #if key.endswith('to_v'): - # sd_module = matched_module.to_v or None - #if key.endswith('to_out_0'): - # sd_module = matched_module.to_out[0] or None - #if key.endswith('to_out_1'): - # sd_module = matched_module.to_out[1] or None - if sd_module is None: keys_failed_to_match[key_network] = key @@ -242,14 +215,6 @@ def load_network(name, network_on_disk): raise AssertionError(f"Could not find a module type (out of {', '.join([x.__class__.__name__ for x in module_types])}) that would accept those keys: {', '.join(weights.w)}") net.modules[key] = net_module - - # replaces forward method of original Linear - # applied_to_count = 0 - #for key, created_module in net.modules.items(): - # if isinstance(created_module, network_oft.NetworkModuleOFT): - # net_module.apply_to() - #applied_to_count += 1 - # print(f'Applied OFT modules: {applied_to_count}') embeddings = {} for emb_name, data in bundle_embeddings.items(): From 0550659ce6e1c37d1ab05cb8a2cb31d499fa552f Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Thu, 19 Oct 2023 13:13:02 -0700 Subject: [PATCH 017/342] style: fix ambiguous variable name --- extensions-builtin/Lora/network_oft.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index 0a87958e2..4e8382c18 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -43,8 +43,8 @@ class NetworkModuleOFT(network.NetworkModule): norm_Q = torch.norm(block_Q.flatten()) new_norm_Q = torch.clamp(norm_Q, max=self.constraint) block_Q = block_Q * ((new_norm_Q + 1e-8) / (norm_Q + 1e-8)) - I = torch.eye(self.block_size, device=self.oft_blocks.device).unsqueeze(0).repeat(self.num_blocks, 1, 1) - block_R = torch.matmul(I + block_Q, (I - block_Q).inverse()) + m_I = torch.eye(self.block_size, device=self.oft_blocks.device).unsqueeze(0).repeat(self.num_blocks, 1, 1) + block_R = torch.matmul(m_I + block_Q, (m_I - block_Q).inverse()) #block_R_weighted = multiplier * block_R + (1 - multiplier) * I #R = torch.block_diag(*block_R_weighted) R = torch.block_diag(*block_R) From 443ca983ade333721930ea2f18f80b45762e2aea Mon Sep 17 00:00:00 2001 From: avantcontra Date: Sun, 22 Oct 2023 03:21:23 +0800 Subject: [PATCH 018/342] fix bug when using --gfpgan-models-path --- modules/gfpgan_model.py | 25 ++++++++++++++++++++----- 1 file changed, 20 insertions(+), 5 deletions(-) diff --git a/modules/gfpgan_model.py b/modules/gfpgan_model.py index 8e0f13bdc..93567253f 100644 --- a/modules/gfpgan_model.py +++ b/modules/gfpgan_model.py @@ -9,6 +9,7 @@ from modules import paths, shared, devices, modelloader, errors model_dir = "GFPGAN" user_path = None model_path = os.path.join(paths.models_path, model_dir) +model_file_path = None model_url = "https://github.com/TencentARC/GFPGAN/releases/download/v1.3.0/GFPGANv1.4.pth" have_gfpgan = False loaded_gfpgan_model = None @@ -17,24 +18,32 @@ loaded_gfpgan_model = None def gfpgann(): global loaded_gfpgan_model global model_path + global model_file_path if loaded_gfpgan_model is not None: loaded_gfpgan_model.gfpgan.to(devices.device_gfpgan) return loaded_gfpgan_model if gfpgan_constructor is None: return None + + models = modelloader.load_models(model_path, model_url, user_path, ext_filter=['.pth']) - models = modelloader.load_models(model_path, model_url, user_path, ext_filter="GFPGAN") if len(models) == 1 and models[0].startswith("http"): model_file = models[0] elif len(models) != 0: - latest_file = max(models, key=os.path.getctime) + gfp_models = [] + for item in models: + if 'GFPGAN' in os.path.basename(item): + gfp_models.append(item) + latest_file = max(gfp_models, key=os.path.getctime) model_file = latest_file else: print("Unable to load gfpgan model!") return None + if hasattr(facexlib.detection.retinaface, 'device'): facexlib.detection.retinaface.device = devices.device_gfpgan + model_file_path = model_file model = gfpgan_constructor(model_path=model_file, upscale=1, arch='clean', channel_multiplier=2, bg_upsampler=None, device=devices.device_gfpgan) loaded_gfpgan_model = model @@ -77,19 +86,25 @@ def setup_model(dirname): global user_path global have_gfpgan global gfpgan_constructor + global model_file_path + + facexlib_path = model_path + + if dirname is not None: + facexlib_path = dirname load_file_from_url_orig = gfpgan.utils.load_file_from_url facex_load_file_from_url_orig = facexlib.detection.load_file_from_url facex_load_file_from_url_orig2 = facexlib.parsing.load_file_from_url def my_load_file_from_url(**kwargs): - return load_file_from_url_orig(**dict(kwargs, model_dir=model_path)) + return load_file_from_url_orig(**dict(kwargs, model_dir=model_file_path)) def facex_load_file_from_url(**kwargs): - return facex_load_file_from_url_orig(**dict(kwargs, save_dir=model_path, model_dir=None)) + return facex_load_file_from_url_orig(**dict(kwargs, save_dir=facexlib_path, model_dir=None)) def facex_load_file_from_url2(**kwargs): - return facex_load_file_from_url_orig2(**dict(kwargs, save_dir=model_path, model_dir=None)) + return facex_load_file_from_url_orig2(**dict(kwargs, save_dir=facexlib_path, model_dir=None)) gfpgan.utils.load_file_from_url = my_load_file_from_url facexlib.detection.load_file_from_url = facex_load_file_from_url From 236dd55dbe895ba72a64567482ee67ab680c5344 Mon Sep 17 00:00:00 2001 From: avantcontra Date: Sun, 22 Oct 2023 04:32:13 +0800 Subject: [PATCH 019/342] fix Blank line contains whitespace --- modules/gfpgan_model.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/gfpgan_model.py b/modules/gfpgan_model.py index 93567253f..01d668ecd 100644 --- a/modules/gfpgan_model.py +++ b/modules/gfpgan_model.py @@ -25,7 +25,7 @@ def gfpgann(): if gfpgan_constructor is None: return None - + models = modelloader.load_models(model_path, model_url, user_path, ext_filter=['.pth']) if len(models) == 1 and models[0].startswith("http"): From 2d8c894b274d60a3e3563a2ace23c4ebcea9e652 Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Sat, 21 Oct 2023 13:43:31 -0700 Subject: [PATCH 020/342] refactor: use forward hook instead of custom forward --- extensions-builtin/Lora/network_oft.py | 33 +++++++++++++++++++------- 1 file changed, 24 insertions(+), 9 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index 4e8382c18..8e561ab0b 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -36,9 +36,11 @@ class NetworkModuleOFT(network.NetworkModule): # how do we revert this to unload the weights? def apply_to(self): self.org_forward = self.org_module[0].forward - self.org_module[0].forward = self.forward + #self.org_module[0].forward = self.forward + self.org_module[0].register_forward_hook(self.forward_hook) def get_weight(self, oft_blocks, multiplier=None): + self.constraint = self.constraint.to(oft_blocks.device, dtype=oft_blocks.dtype) block_Q = oft_blocks - oft_blocks.transpose(1, 2) norm_Q = torch.norm(block_Q.flatten()) new_norm_Q = torch.clamp(norm_Q, max=self.constraint) @@ -66,14 +68,10 @@ class NetworkModuleOFT(network.NetworkModule): output_shape = self.oft_blocks.shape return self.finalize_updown(updown, orig_weight, output_shape) - - def forward(self, x, y=None): - x = self.org_forward(x) - if self.multiplier() == 0.0: - return x - - # calculating R here is excruciatingly slow - #R = self.get_weight().to(x.device, dtype=x.dtype) + + def forward_hook(self, module, args, output): + #print(f'Forward hook in {self.network_key} called') + x = output R = self.R.to(x.device, dtype=x.dtype) if x.dim() == 4: @@ -83,3 +81,20 @@ class NetworkModuleOFT(network.NetworkModule): else: x = torch.matmul(x, R) return x + + # def forward(self, x, y=None): + # x = self.org_forward(x) + # if self.multiplier() == 0.0: + # return x + + # # calculating R here is excruciatingly slow + # #R = self.get_weight().to(x.device, dtype=x.dtype) + # R = self.R.to(x.device, dtype=x.dtype) + + # if x.dim() == 4: + # x = x.permute(0, 2, 3, 1) + # x = torch.matmul(x, R) + # x = x.permute(0, 3, 1, 2) + # else: + # x = torch.matmul(x, R) + # return x From 768354772853a1d27a9bf7e41bd6a6e4eac7a9c7 Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Sat, 21 Oct 2023 14:42:24 -0700 Subject: [PATCH 021/342] fix: return orig weights during updown, merge weights before forward --- extensions-builtin/Lora/network_oft.py | 94 +++++++++++++++++++------- 1 file changed, 71 insertions(+), 23 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index 8e561ab0b..f5f32c238 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -1,5 +1,6 @@ import torch import network +from modules import devices class ModuleTypeOFT(network.ModuleType): @@ -29,23 +30,56 @@ class NetworkModuleOFT(network.NetworkModule): self.block_size = self.out_dim // self.num_blocks self.org_module: list[torch.Module] = [self.sd_module] + self.org_weight = self.org_module[0].weight.to(self.org_module[0].weight.device, copy=True) + #self.org_weight = self.org_module[0].weight.to(devices.cpu, copy=True) self.R = self.get_weight(self.oft_blocks) + + self.merged_weight = self.merge_weight() self.apply_to() + self.merged = False + + + def merge_weight(self): + org_sd = self.org_module[0].state_dict() + R = self.R.to(self.org_weight.device, dtype=self.org_weight.dtype) + if self.org_weight.dim() == 4: + weight = torch.einsum("oihw, op -> pihw", self.org_weight, R) + else: + weight = torch.einsum("oi, op -> pi", self.org_weight, R) + org_sd['weight'] = weight + # replace weight + #self.org_module[0].load_state_dict(org_sd) + return weight + pass + + def replace_weight(self, new_weight): + org_sd = self.org_module[0].state_dict() + org_sd['weight'] = new_weight + self.org_module[0].load_state_dict(org_sd) + self.merged = True + + def restore_weight(self): + org_sd = self.org_module[0].state_dict() + org_sd['weight'] = self.org_weight + self.org_module[0].load_state_dict(org_sd) + self.merged = False + # replace forward method of original linear rather than replacing the module # how do we revert this to unload the weights? def apply_to(self): self.org_forward = self.org_module[0].forward #self.org_module[0].forward = self.forward + self.org_module[0].register_forward_pre_hook(self.pre_forward_hook) self.org_module[0].register_forward_hook(self.forward_hook) def get_weight(self, oft_blocks, multiplier=None): - self.constraint = self.constraint.to(oft_blocks.device, dtype=oft_blocks.dtype) + constraint = self.constraint.to(oft_blocks.device, dtype=oft_blocks.dtype) block_Q = oft_blocks - oft_blocks.transpose(1, 2) norm_Q = torch.norm(block_Q.flatten()) - new_norm_Q = torch.clamp(norm_Q, max=self.constraint) + new_norm_Q = torch.clamp(norm_Q, max=constraint) block_Q = block_Q * ((new_norm_Q + 1e-8) / (norm_Q + 1e-8)) - m_I = torch.eye(self.block_size, device=self.oft_blocks.device).unsqueeze(0).repeat(self.num_blocks, 1, 1) + m_I = torch.eye(self.block_size, device=oft_blocks.device).unsqueeze(0).repeat(self.num_blocks, 1, 1) block_R = torch.matmul(m_I + block_Q, (m_I - block_Q).inverse()) #block_R_weighted = multiplier * block_R + (1 - multiplier) * I #R = torch.block_diag(*block_R_weighted) @@ -54,33 +88,47 @@ class NetworkModuleOFT(network.NetworkModule): return R def calc_updown(self, orig_weight): - oft_blocks = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) + #oft_blocks = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) - R = self.get_weight(oft_blocks) - self.R = R + #R = self.R.to(orig_weight.device, dtype=orig_weight.dtype) + ##self.R = R - # if orig_weight.dim() == 4: - # weight = torch.einsum("oihw, op -> pihw", orig_weight, R) - # else: - # weight = torch.einsum("oi, op -> pi", orig_weight, R) + #if orig_weight.dim() == 4: + # weight = torch.einsum("oihw, op -> pihw", orig_weight, R) + #else: + # weight = torch.einsum("oi, op -> pi", orig_weight, R) - updown = orig_weight @ R - output_shape = self.oft_blocks.shape + #updown = orig_weight @ R + #updown = weight + updown = torch.zeros_like(orig_weight, device=orig_weight.device, dtype=orig_weight.dtype) + #updown = orig_weight + output_shape = orig_weight.shape + #orig_weight = self.merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) + #output_shape = self.oft_blocks.shape return self.finalize_updown(updown, orig_weight, output_shape) - def forward_hook(self, module, args, output): - #print(f'Forward hook in {self.network_key} called') - x = output - R = self.R.to(x.device, dtype=x.dtype) + def pre_forward_hook(self, module, input): + if not self.merged: + self.replace_weight(self.merged_weight) - if x.dim() == 4: - x = x.permute(0, 2, 3, 1) - x = torch.matmul(x, R) - x = x.permute(0, 3, 1, 2) - else: - x = torch.matmul(x, R) - return x + + def forward_hook(self, module, args, output): + if self.merged: + pass + #self.restore_weight() + #print(f'Forward hook in {self.network_key} called') + + #x = output + #R = self.R.to(x.device, dtype=x.dtype) + + #if x.dim() == 4: + # x = x.permute(0, 2, 3, 1) + # x = torch.matmul(x, R) + # x = x.permute(0, 3, 1, 2) + #else: + # x = torch.matmul(x, R) + #return x # def forward(self, x, y=None): # x = self.org_forward(x) From fce86ab7d75690785f0f5b496f1b3aee922c0ae3 Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Sat, 21 Oct 2023 16:03:54 -0700 Subject: [PATCH 022/342] fix: support multiplier, no forward pass hook --- extensions-builtin/Lora/network_oft.py | 43 ++++++++++++++++++++------ 1 file changed, 33 insertions(+), 10 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index f5f32c238..e0672ba6d 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -32,21 +32,27 @@ class NetworkModuleOFT(network.NetworkModule): self.org_module: list[torch.Module] = [self.sd_module] self.org_weight = self.org_module[0].weight.to(self.org_module[0].weight.device, copy=True) #self.org_weight = self.org_module[0].weight.to(devices.cpu, copy=True) - self.R = self.get_weight(self.oft_blocks) + init_multiplier = self.multiplier() * self.calc_scale() + self.last_multiplier = init_multiplier + self.R = self.get_weight(self.oft_blocks, init_multiplier) self.merged_weight = self.merge_weight() self.apply_to() self.merged = False + # weights_backup = getattr(self.org_module[0], 'network_weights_backup', None) + # if weights_backup is None: + # self.org_module[0].network_weights_backup = self.org_weight + def merge_weight(self): - org_sd = self.org_module[0].state_dict() + #org_sd = self.org_module[0].state_dict() R = self.R.to(self.org_weight.device, dtype=self.org_weight.dtype) if self.org_weight.dim() == 4: weight = torch.einsum("oihw, op -> pihw", self.org_weight, R) else: weight = torch.einsum("oi, op -> pi", self.org_weight, R) - org_sd['weight'] = weight + #org_sd['weight'] = weight # replace weight #self.org_module[0].load_state_dict(org_sd) return weight @@ -74,6 +80,7 @@ class NetworkModuleOFT(network.NetworkModule): self.org_module[0].register_forward_hook(self.forward_hook) def get_weight(self, oft_blocks, multiplier=None): + multiplier = multiplier.to(oft_blocks.device, dtype=oft_blocks.dtype) constraint = self.constraint.to(oft_blocks.device, dtype=oft_blocks.dtype) block_Q = oft_blocks - oft_blocks.transpose(1, 2) norm_Q = torch.norm(block_Q.flatten()) @@ -81,9 +88,9 @@ class NetworkModuleOFT(network.NetworkModule): block_Q = block_Q * ((new_norm_Q + 1e-8) / (norm_Q + 1e-8)) m_I = torch.eye(self.block_size, device=oft_blocks.device).unsqueeze(0).repeat(self.num_blocks, 1, 1) block_R = torch.matmul(m_I + block_Q, (m_I - block_Q).inverse()) - #block_R_weighted = multiplier * block_R + (1 - multiplier) * I - #R = torch.block_diag(*block_R_weighted) - R = torch.block_diag(*block_R) + block_R_weighted = multiplier * block_R + (1 - multiplier) * m_I + R = torch.block_diag(*block_R_weighted) + #R = torch.block_diag(*block_R) return R @@ -93,6 +100,8 @@ class NetworkModuleOFT(network.NetworkModule): #R = self.R.to(orig_weight.device, dtype=orig_weight.dtype) ##self.R = R + #R = self.R.to(orig_weight.device, dtype=orig_weight.dtype) + ##self.R = R #if orig_weight.dim() == 4: # weight = torch.einsum("oihw, op -> pihw", orig_weight, R) #else: @@ -103,19 +112,33 @@ class NetworkModuleOFT(network.NetworkModule): updown = torch.zeros_like(orig_weight, device=orig_weight.device, dtype=orig_weight.dtype) #updown = orig_weight output_shape = orig_weight.shape - #orig_weight = self.merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) + orig_weight = self.merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) #output_shape = self.oft_blocks.shape return self.finalize_updown(updown, orig_weight, output_shape) def pre_forward_hook(self, module, input): - if not self.merged: + multiplier = self.multiplier() * self.calc_scale() + if not multiplier==self.last_multiplier or not self.merged: + + #if multiplier != self.last_multiplier or not self.merged: + self.R = self.get_weight(self.oft_blocks, multiplier) + self.last_multiplier = multiplier + self.merged_weight = self.merge_weight() self.replace_weight(self.merged_weight) + #elif not self.merged: + # self.replace_weight(self.merged_weight) def forward_hook(self, module, args, output): - if self.merged: - pass + pass + #output = output * self.multiplier() * self.calc_scale() + #if len(args) > 0: + # y = args[0] + # output = output + y + #return output + #if self.merged: + # pass #self.restore_weight() #print(f'Forward hook in {self.network_key} called') From 76f5abdbdb739133eff2ccefa36eac62bea3fa08 Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Sat, 21 Oct 2023 16:07:45 -0700 Subject: [PATCH 023/342] style: cleanup oft --- extensions-builtin/Lora/network_oft.py | 82 +++----------------------- 1 file changed, 7 insertions(+), 75 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index e0672ba6d..e462ccb1b 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -1,6 +1,5 @@ import torch import network -from modules import devices class ModuleTypeOFT(network.ModuleType): @@ -31,33 +30,24 @@ class NetworkModuleOFT(network.NetworkModule): self.org_module: list[torch.Module] = [self.sd_module] self.org_weight = self.org_module[0].weight.to(self.org_module[0].weight.device, copy=True) - #self.org_weight = self.org_module[0].weight.to(devices.cpu, copy=True) + init_multiplier = self.multiplier() * self.calc_scale() self.last_multiplier = init_multiplier + self.R = self.get_weight(self.oft_blocks, init_multiplier) self.merged_weight = self.merge_weight() self.apply_to() self.merged = False - # weights_backup = getattr(self.org_module[0], 'network_weights_backup', None) - # if weights_backup is None: - # self.org_module[0].network_weights_backup = self.org_weight - - def merge_weight(self): - #org_sd = self.org_module[0].state_dict() R = self.R.to(self.org_weight.device, dtype=self.org_weight.dtype) if self.org_weight.dim() == 4: weight = torch.einsum("oihw, op -> pihw", self.org_weight, R) else: weight = torch.einsum("oi, op -> pi", self.org_weight, R) - #org_sd['weight'] = weight - # replace weight - #self.org_module[0].load_state_dict(org_sd) return weight - pass - + def replace_weight(self, new_weight): org_sd = self.org_module[0].state_dict() org_sd['weight'] = new_weight @@ -70,9 +60,7 @@ class NetworkModuleOFT(network.NetworkModule): self.org_module[0].load_state_dict(org_sd) self.merged = False - - # replace forward method of original linear rather than replacing the module - # how do we revert this to unload the weights? + # FIXME: hook forward method of original linear, but how do we undo the hook when we are done? def apply_to(self): self.org_forward = self.org_module[0].forward #self.org_module[0].forward = self.forward @@ -90,82 +78,26 @@ class NetworkModuleOFT(network.NetworkModule): block_R = torch.matmul(m_I + block_Q, (m_I - block_Q).inverse()) block_R_weighted = multiplier * block_R + (1 - multiplier) * m_I R = torch.block_diag(*block_R_weighted) - #R = torch.block_diag(*block_R) return R def calc_updown(self, orig_weight): - #oft_blocks = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) - - #R = self.R.to(orig_weight.device, dtype=orig_weight.dtype) - ##self.R = R - - #R = self.R.to(orig_weight.device, dtype=orig_weight.dtype) - ##self.R = R - #if orig_weight.dim() == 4: - # weight = torch.einsum("oihw, op -> pihw", orig_weight, R) - #else: - # weight = torch.einsum("oi, op -> pi", orig_weight, R) - - #updown = orig_weight @ R - #updown = weight updown = torch.zeros_like(orig_weight, device=orig_weight.device, dtype=orig_weight.dtype) - #updown = orig_weight output_shape = orig_weight.shape orig_weight = self.merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) #output_shape = self.oft_blocks.shape return self.finalize_updown(updown, orig_weight, output_shape) - + def pre_forward_hook(self, module, input): multiplier = self.multiplier() * self.calc_scale() - if not multiplier==self.last_multiplier or not self.merged: - #if multiplier != self.last_multiplier or not self.merged: + if not multiplier==self.last_multiplier or not self.merged: self.R = self.get_weight(self.oft_blocks, multiplier) self.last_multiplier = multiplier self.merged_weight = self.merge_weight() self.replace_weight(self.merged_weight) - #elif not self.merged: - # self.replace_weight(self.merged_weight) - + def forward_hook(self, module, args, output): pass - #output = output * self.multiplier() * self.calc_scale() - #if len(args) > 0: - # y = args[0] - # output = output + y - #return output - #if self.merged: - # pass - #self.restore_weight() - #print(f'Forward hook in {self.network_key} called') - - #x = output - #R = self.R.to(x.device, dtype=x.dtype) - - #if x.dim() == 4: - # x = x.permute(0, 2, 3, 1) - # x = torch.matmul(x, R) - # x = x.permute(0, 3, 1, 2) - #else: - # x = torch.matmul(x, R) - #return x - - # def forward(self, x, y=None): - # x = self.org_forward(x) - # if self.multiplier() == 0.0: - # return x - - # # calculating R here is excruciatingly slow - # #R = self.get_weight().to(x.device, dtype=x.dtype) - # R = self.R.to(x.device, dtype=x.dtype) - - # if x.dim() == 4: - # x = x.permute(0, 2, 3, 1) - # x = torch.matmul(x, R) - # x = x.permute(0, 3, 1, 2) - # else: - # x = torch.matmul(x, R) - # return x From de8ee92ed88b855098e273f576a27f4789f0693d Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Sat, 21 Oct 2023 17:37:17 -0700 Subject: [PATCH 024/342] fix: use merge_weight to cache value --- extensions-builtin/Lora/network_oft.py | 57 ++++++++++++++++++-------- 1 file changed, 40 insertions(+), 17 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index e462ccb1b..ebe6740c5 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -29,23 +29,27 @@ class NetworkModuleOFT(network.NetworkModule): self.block_size = self.out_dim // self.num_blocks self.org_module: list[torch.Module] = [self.sd_module] - self.org_weight = self.org_module[0].weight.to(self.org_module[0].weight.device, copy=True) + #self.org_weight = self.org_module[0].weight.to(self.org_module[0].weight.device, copy=True) init_multiplier = self.multiplier() * self.calc_scale() self.last_multiplier = init_multiplier self.R = self.get_weight(self.oft_blocks, init_multiplier) + self.hooks = [] self.merged_weight = self.merge_weight() - self.apply_to() + + #self.apply_to() + self.applied = False self.merged = False def merge_weight(self): - R = self.R.to(self.org_weight.device, dtype=self.org_weight.dtype) - if self.org_weight.dim() == 4: - weight = torch.einsum("oihw, op -> pihw", self.org_weight, R) + org_weight = self.org_module[0].weight + R = self.R.to(org_weight.device, dtype=org_weight.dtype) + if org_weight.dim() == 4: + weight = torch.einsum("oihw, op -> pihw", org_weight, R) else: - weight = torch.einsum("oi, op -> pi", self.org_weight, R) + weight = torch.einsum("oi, op -> pi", org_weight, R) return weight def replace_weight(self, new_weight): @@ -55,17 +59,29 @@ class NetworkModuleOFT(network.NetworkModule): self.merged = True def restore_weight(self): - org_sd = self.org_module[0].state_dict() - org_sd['weight'] = self.org_weight - self.org_module[0].load_state_dict(org_sd) - self.merged = False + pass + #org_sd = self.org_module[0].state_dict() + #org_sd['weight'] = self.org_weight + #self.org_module[0].load_state_dict(org_sd) + #self.merged = False # FIXME: hook forward method of original linear, but how do we undo the hook when we are done? def apply_to(self): - self.org_forward = self.org_module[0].forward - #self.org_module[0].forward = self.forward - self.org_module[0].register_forward_pre_hook(self.pre_forward_hook) - self.org_module[0].register_forward_hook(self.forward_hook) + if not self.applied: + self.org_forward = self.org_module[0].forward + #self.org_module[0].forward = self.forward + prehook = self.org_module[0].register_forward_pre_hook(self.pre_forward_hook) + hook = self.org_module[0].register_forward_hook(self.forward_hook) + self.hooks.append(prehook) + self.hooks.append(hook) + self.applied = True + + def remove_from(self): + if self.applied: + for hook in self.hooks: + hook.remove() + self.hooks = [] + self.applied = False def get_weight(self, oft_blocks, multiplier=None): multiplier = multiplier.to(oft_blocks.device, dtype=oft_blocks.dtype) @@ -82,14 +98,22 @@ class NetworkModuleOFT(network.NetworkModule): return R def calc_updown(self, orig_weight): + if not self.applied: + self.apply_to() + + self.merged_weight = self.merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) + updown = torch.zeros_like(orig_weight, device=orig_weight.device, dtype=orig_weight.dtype) output_shape = orig_weight.shape - orig_weight = self.merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) + orig_weight = self.merged_weight #output_shape = self.oft_blocks.shape return self.finalize_updown(updown, orig_weight, output_shape) def pre_forward_hook(self, module, input): + #if not self.applied: + # self.apply_to() + multiplier = self.multiplier() * self.calc_scale() if not multiplier==self.last_multiplier or not self.merged: @@ -98,6 +122,5 @@ class NetworkModuleOFT(network.NetworkModule): self.merged_weight = self.merge_weight() self.replace_weight(self.merged_weight) - def forward_hook(self, module, args, output): - pass + pass \ No newline at end of file From 4a50c9638c3eac860fb05ae603cd61aabf4cd1a9 Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Sun, 22 Oct 2023 08:54:24 -0700 Subject: [PATCH 025/342] refactor: remove used OFT functions --- extensions-builtin/Lora/network_oft.py | 82 ++++---------------------- 1 file changed, 10 insertions(+), 72 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index ebe6740c5..3034a407e 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -29,98 +29,36 @@ class NetworkModuleOFT(network.NetworkModule): self.block_size = self.out_dim // self.num_blocks self.org_module: list[torch.Module] = [self.sd_module] - #self.org_weight = self.org_module[0].weight.to(self.org_module[0].weight.device, copy=True) - init_multiplier = self.multiplier() * self.calc_scale() - self.last_multiplier = init_multiplier - - self.R = self.get_weight(self.oft_blocks, init_multiplier) - - self.hooks = [] - self.merged_weight = self.merge_weight() - - #self.apply_to() - self.applied = False - self.merged = False - - def merge_weight(self): - org_weight = self.org_module[0].weight - R = self.R.to(org_weight.device, dtype=org_weight.dtype) + def merge_weight(self, R_weight, org_weight): + R_weight = R_weight.to(org_weight.device, dtype=org_weight.dtype) if org_weight.dim() == 4: - weight = torch.einsum("oihw, op -> pihw", org_weight, R) + weight = torch.einsum("oihw, op -> pihw", org_weight, R_weight) else: - weight = torch.einsum("oi, op -> pi", org_weight, R) + weight = torch.einsum("oi, op -> pi", org_weight, R_weight) return weight - def replace_weight(self, new_weight): - org_sd = self.org_module[0].state_dict() - org_sd['weight'] = new_weight - self.org_module[0].load_state_dict(org_sd) - self.merged = True - - def restore_weight(self): - pass - #org_sd = self.org_module[0].state_dict() - #org_sd['weight'] = self.org_weight - #self.org_module[0].load_state_dict(org_sd) - #self.merged = False - - # FIXME: hook forward method of original linear, but how do we undo the hook when we are done? - def apply_to(self): - if not self.applied: - self.org_forward = self.org_module[0].forward - #self.org_module[0].forward = self.forward - prehook = self.org_module[0].register_forward_pre_hook(self.pre_forward_hook) - hook = self.org_module[0].register_forward_hook(self.forward_hook) - self.hooks.append(prehook) - self.hooks.append(hook) - self.applied = True - - def remove_from(self): - if self.applied: - for hook in self.hooks: - hook.remove() - self.hooks = [] - self.applied = False - def get_weight(self, oft_blocks, multiplier=None): - multiplier = multiplier.to(oft_blocks.device, dtype=oft_blocks.dtype) constraint = self.constraint.to(oft_blocks.device, dtype=oft_blocks.dtype) + block_Q = oft_blocks - oft_blocks.transpose(1, 2) norm_Q = torch.norm(block_Q.flatten()) new_norm_Q = torch.clamp(norm_Q, max=constraint) block_Q = block_Q * ((new_norm_Q + 1e-8) / (norm_Q + 1e-8)) m_I = torch.eye(self.block_size, device=oft_blocks.device).unsqueeze(0).repeat(self.num_blocks, 1, 1) block_R = torch.matmul(m_I + block_Q, (m_I - block_Q).inverse()) + block_R_weighted = multiplier * block_R + (1 - multiplier) * m_I R = torch.block_diag(*block_R_weighted) return R def calc_updown(self, orig_weight): - if not self.applied: - self.apply_to() + R = self.get_weight(self.oft_blocks, self.multiplier()) + merged_weight = self.merge_weight(R, orig_weight) - self.merged_weight = self.merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) - - updown = torch.zeros_like(orig_weight, device=orig_weight.device, dtype=orig_weight.dtype) + updown = merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight output_shape = orig_weight.shape - orig_weight = self.merged_weight - #output_shape = self.oft_blocks.shape + orig_weight = orig_weight return self.finalize_updown(updown, orig_weight, output_shape) - - def pre_forward_hook(self, module, input): - #if not self.applied: - # self.apply_to() - - multiplier = self.multiplier() * self.calc_scale() - - if not multiplier==self.last_multiplier or not self.merged: - self.R = self.get_weight(self.oft_blocks, multiplier) - self.last_multiplier = multiplier - self.merged_weight = self.merge_weight() - self.replace_weight(self.merged_weight) - - def forward_hook(self, module, args, output): - pass \ No newline at end of file From 3b8515d2c9abad7f0ccaac0215803716e861ee0e Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Sun, 22 Oct 2023 09:27:48 -0700 Subject: [PATCH 026/342] fix: multiplier applied twice in finalize_updown --- extensions-builtin/Lora/network_oft.py | 23 ++++++++++++++++++++++- 1 file changed, 22 insertions(+), 1 deletion(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index 3034a407e..efbdd296a 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -54,7 +54,8 @@ class NetworkModuleOFT(network.NetworkModule): return R def calc_updown(self, orig_weight): - R = self.get_weight(self.oft_blocks, self.multiplier()) + multiplier = self.multiplier() * self.calc_scale() + R = self.get_weight(self.oft_blocks, multiplier) merged_weight = self.merge_weight(R, orig_weight) updown = merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight @@ -62,3 +63,23 @@ class NetworkModuleOFT(network.NetworkModule): orig_weight = orig_weight return self.finalize_updown(updown, orig_weight, output_shape) + + # override to remove the multiplier/scale factor; it's already multiplied in get_weight + def finalize_updown(self, updown, orig_weight, output_shape, ex_bias=None): + #return super().finalize_updown(updown, orig_weight, output_shape, ex_bias) + + if self.bias is not None: + updown = updown.reshape(self.bias.shape) + updown += self.bias.to(orig_weight.device, dtype=orig_weight.dtype) + updown = updown.reshape(output_shape) + + if len(output_shape) == 4: + updown = updown.reshape(output_shape) + + if orig_weight.size().numel() == updown.size().numel(): + updown = updown.reshape(orig_weight.shape) + + if ex_bias is not None: + ex_bias = ex_bias * self.multiplier() + + return updown, ex_bias From 6523edb8a45d4e09f11f3b4e1d133afa6fb65e53 Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Sun, 22 Oct 2023 09:31:15 -0700 Subject: [PATCH 027/342] style: conform style --- extensions-builtin/Lora/network_oft.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index efbdd296a..e43c9a1df 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -63,7 +63,7 @@ class NetworkModuleOFT(network.NetworkModule): orig_weight = orig_weight return self.finalize_updown(updown, orig_weight, output_shape) - + # override to remove the multiplier/scale factor; it's already multiplied in get_weight def finalize_updown(self, updown, orig_weight, output_shape, ex_bias=None): #return super().finalize_updown(updown, orig_weight, output_shape, ex_bias) From 88b2ef3b04c37ec068fdfea9ba2596645e981b46 Mon Sep 17 00:00:00 2001 From: David Benson Date: Mon, 23 Oct 2023 08:16:26 -0400 Subject: [PATCH 028/342] Update prompts_from_file script to allow concatenating entries with the general prompt. --- scripts/prompts_from_file.py | 17 +++++++++++++++-- 1 file changed, 15 insertions(+), 2 deletions(-) diff --git a/scripts/prompts_from_file.py b/scripts/prompts_from_file.py index 50320d553..1aadf1135 100644 --- a/scripts/prompts_from_file.py +++ b/scripts/prompts_from_file.py @@ -108,6 +108,7 @@ class Script(scripts.Script): def ui(self, is_img2img): checkbox_iterate = gr.Checkbox(label="Iterate seed every line", value=False, elem_id=self.elem_id("checkbox_iterate")) checkbox_iterate_batch = gr.Checkbox(label="Use same random seed for all lines", value=False, elem_id=self.elem_id("checkbox_iterate_batch")) + prompt_position = gr.Radio(["start", "end"], label="Insert prompts at the", elem_id=self.elem_id("prompt_position"), value="start") prompt_txt = gr.Textbox(label="List of prompt inputs", lines=1, elem_id=self.elem_id("prompt_txt")) file = gr.File(label="Upload prompt inputs", type='binary', elem_id=self.elem_id("file")) @@ -118,9 +119,9 @@ class Script(scripts.Script): # We don't shrink back to 1, because that causes the control to ignore [enter], and it may # be unclear to the user that shift-enter is needed. prompt_txt.change(lambda tb: gr.update(lines=7) if ("\n" in tb) else gr.update(lines=2), inputs=[prompt_txt], outputs=[prompt_txt], show_progress=False) - return [checkbox_iterate, checkbox_iterate_batch, prompt_txt] + return [checkbox_iterate, checkbox_iterate_batch, prompt_position, prompt_txt] - def run(self, p, checkbox_iterate, checkbox_iterate_batch, prompt_txt: str): + def run(self, p, checkbox_iterate, checkbox_iterate_batch, prompt_position, prompt_txt: str): lines = [x for x in (x.strip() for x in prompt_txt.splitlines()) if x] p.do_not_save_grid = True @@ -158,6 +159,18 @@ class Script(scripts.Script): for k, v in args.items(): setattr(copy_p, k, v) + if args.get("prompt") and p.prompt: + if prompt_position == "start": + copy_p.prompt = args.get("prompt") + " " + p.prompt + else: + copy_p.prompt = p.prompt + " " + args.get("prompt") + + if args.get("negative_prompt") and p.negative_prompt: + if prompt_position == "start": + copy_p.negative_prompt = args.get("negative_prompt") + " " + p.negative_prompt + else: + copy_p.negative_prompt = p.negative_prompt + " " + args.get("negative_prompt") + proc = process_images(copy_p) images += proc.images From dfc4c27b2402a35a1820ffa549e74bb79873aaaa Mon Sep 17 00:00:00 2001 From: David Benson Date: Mon, 23 Oct 2023 08:26:40 -0400 Subject: [PATCH 029/342] linting issue --- scripts/prompts_from_file.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/scripts/prompts_from_file.py b/scripts/prompts_from_file.py index 1aadf1135..3c09bb976 100644 --- a/scripts/prompts_from_file.py +++ b/scripts/prompts_from_file.py @@ -164,7 +164,7 @@ class Script(scripts.Script): copy_p.prompt = args.get("prompt") + " " + p.prompt else: copy_p.prompt = p.prompt + " " + args.get("prompt") - + if args.get("negative_prompt") and p.negative_prompt: if prompt_position == "start": copy_p.negative_prompt = args.get("negative_prompt") + " " + p.negative_prompt From eaa9f5162fbca2ebcb2682eb861bc7e5510a2b66 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Tue, 24 Oct 2023 01:49:05 +0800 Subject: [PATCH 030/342] Add CPU fp8 support Since norm layer need fp32, I only convert the linear operation layer(conv2d/linear) And TE have some pytorch function not support bf16 amp in CPU. I add a condition to indicate if the autocast is for unet. --- modules/devices.py | 6 +++++- modules/processing.py | 2 +- modules/sd_models.py | 20 ++++++++++++++++---- 3 files changed, 22 insertions(+), 6 deletions(-) diff --git a/modules/devices.py b/modules/devices.py index 1d4eb5635..0cd2b55d2 100644 --- a/modules/devices.py +++ b/modules/devices.py @@ -71,6 +71,7 @@ def enable_tf32(): errors.run(enable_tf32, "Enabling TF32") cpu: torch.device = torch.device("cpu") +fp8: bool = False device: torch.device = None device_interrogate: torch.device = None device_gfpgan: torch.device = None @@ -93,10 +94,13 @@ def cond_cast_float(input): nv_rng = None -def autocast(disable=False): +def autocast(disable=False, unet=False): if disable: return contextlib.nullcontext() + if unet and fp8 and device==cpu: + return torch.autocast("cpu", dtype=torch.bfloat16, enabled=True) + if dtype == torch.float32 or shared.cmd_opts.precision == "full": return contextlib.nullcontext() diff --git a/modules/processing.py b/modules/processing.py index 40598f5cf..2df8a7ea7 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -865,7 +865,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: if p.n_iter > 1: shared.state.job = f"Batch {n+1} out of {p.n_iter}" - with devices.without_autocast() if devices.unet_needs_upcast else devices.autocast(): + with devices.without_autocast() if devices.unet_needs_upcast else devices.autocast(unet=True): samples_ddim = p.sample(conditioning=p.c, unconditional_conditioning=p.uc, seeds=p.seeds, subseeds=p.subseeds, subseed_strength=p.subseed_strength, prompts=p.prompts) if getattr(samples_ddim, 'already_decoded', False): diff --git a/modules/sd_models.py b/modules/sd_models.py index 08af128fc..c5fe57bfe 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -391,12 +391,24 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer devices.dtype_unet = torch.float16 timer.record("apply half()") - if shared.cmd_opts.opt_unet_fp8_storage: + + if shared.cmd_opts.opt_unet_fp8_storage: + enable_fp8 = True + elif model.is_sdxl and shared.cmd_opts.opt_unet_fp8_storage_xl: + enable_fp8 = True + + if enable_fp8: + devices.fp8 = True + if devices.device == devices.cpu: + for module in model.model.diffusion_model.modules(): + if isinstance(module, torch.nn.Conv2d): + module.to(torch.float8_e4m3fn) + elif isinstance(module, torch.nn.Linear): + module.to(torch.float8_e4m3fn) + timer.record("apply fp8 unet for cpu") + else: model.model.diffusion_model = model.model.diffusion_model.to(torch.float8_e4m3fn) timer.record("apply fp8 unet") - elif model.is_sdxl and shared.cmd_opts.opt_unet_fp8_storage_xl: - model.model.diffusion_model = model.model.diffusion_model.to(torch.float8_e4m3fn) - timer.record("apply fp8 unet for sdxl") devices.unet_needs_upcast = shared.cmd_opts.upcast_sampling and devices.dtype == torch.float16 and devices.dtype_unet == torch.float16 From 9c1eba2af3a6f9cd6282b3a367656793cbe70c01 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Tue, 24 Oct 2023 02:11:27 +0800 Subject: [PATCH 031/342] Fix lint --- modules/sd_models.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/sd_models.py b/modules/sd_models.py index c5fe57bfe..44d4038b2 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -396,7 +396,7 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer enable_fp8 = True elif model.is_sdxl and shared.cmd_opts.opt_unet_fp8_storage_xl: enable_fp8 = True - + if enable_fp8: devices.fp8 = True if devices.device == devices.cpu: From 1df6c8bfec4715610d64684b6ad2fa38c76c1df6 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Wed, 25 Oct 2023 11:36:43 +0800 Subject: [PATCH 032/342] fp8 for TE --- modules/sd_models.py | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/modules/sd_models.py b/modules/sd_models.py index 44d4038b2..693952943 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -407,6 +407,13 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer module.to(torch.float8_e4m3fn) timer.record("apply fp8 unet for cpu") else: + if model.is_sdxl: + cond_stage = model.conditioner + else: + cond_stage = model.cond_stage_model + for module in cond_stage.modules(): + if isinstance(module, torch.nn.Linear): + module.to(torch.float8_e4m3fn) model.model.diffusion_model = model.model.diffusion_model.to(torch.float8_e4m3fn) timer.record("apply fp8 unet") From 4830b251366436ee8499c003fe87e46ddb4a4581 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Wed, 25 Oct 2023 11:53:37 +0800 Subject: [PATCH 033/342] Fix alphas_cumprod dtype --- modules/sd_models.py | 1 + 1 file changed, 1 insertion(+) diff --git a/modules/sd_models.py b/modules/sd_models.py index 693952943..236604545 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -416,6 +416,7 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer module.to(torch.float8_e4m3fn) model.model.diffusion_model = model.model.diffusion_model.to(torch.float8_e4m3fn) timer.record("apply fp8 unet") + model.alphas_cumprod = model.alphas_cumprod.to(torch.float32) devices.unet_needs_upcast = shared.cmd_opts.upcast_sampling and devices.dtype == torch.float16 and devices.dtype_unet == torch.float16 From bf5067f50ca32cd4764638702e3cc38bca8bfd8b Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Wed, 25 Oct 2023 12:54:28 +0800 Subject: [PATCH 034/342] Fix alphas cumprod --- modules/sd_models.py | 3 ++- modules/sd_models_xl.py | 2 +- 2 files changed, 3 insertions(+), 2 deletions(-) diff --git a/modules/sd_models.py b/modules/sd_models.py index 236604545..7ed89a9c9 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -396,6 +396,8 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer enable_fp8 = True elif model.is_sdxl and shared.cmd_opts.opt_unet_fp8_storage_xl: enable_fp8 = True + else: + enable_fp8 = False if enable_fp8: devices.fp8 = True @@ -416,7 +418,6 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer module.to(torch.float8_e4m3fn) model.model.diffusion_model = model.model.diffusion_model.to(torch.float8_e4m3fn) timer.record("apply fp8 unet") - model.alphas_cumprod = model.alphas_cumprod.to(torch.float32) devices.unet_needs_upcast = shared.cmd_opts.upcast_sampling and devices.dtype == torch.float16 and devices.dtype_unet == torch.float16 diff --git a/modules/sd_models_xl.py b/modules/sd_models_xl.py index 011233216..11259a369 100644 --- a/modules/sd_models_xl.py +++ b/modules/sd_models_xl.py @@ -93,7 +93,7 @@ def extend_sdxl(model): model.parameterization = "v" if isinstance(model.denoiser.scaling, sgm.modules.diffusionmodules.denoiser_scaling.VScaling) else "eps" discretization = sgm.modules.diffusionmodules.discretizer.LegacyDDPMDiscretization() - model.alphas_cumprod = torch.asarray(discretization.alphas_cumprod, device=devices.device, dtype=dtype) + model.alphas_cumprod = torch.asarray(discretization.alphas_cumprod, device=devices.device, dtype=torch.float32) model.conditioner.wrapped = torch.nn.Module() From dda067f64d3289cee3ffd65767126cb30ae73b13 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Wed, 25 Oct 2023 19:53:22 +0800 Subject: [PATCH 035/342] ignore mps for fp8 --- modules/sd_models.py | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/modules/sd_models.py b/modules/sd_models.py index 7ed89a9c9..ccb6afd2a 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -392,7 +392,9 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer devices.dtype_unet = torch.float16 timer.record("apply half()") - if shared.cmd_opts.opt_unet_fp8_storage: + if devices.get_optimal_device_name() == "mps": + enable_fp8 = False + elif shared.cmd_opts.opt_unet_fp8_storage: enable_fp8 = True elif model.is_sdxl and shared.cmd_opts.opt_unet_fp8_storage_xl: enable_fp8 = True From 0beb131c7ffae6f756a6339206da311232a36970 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Wed, 25 Oct 2023 20:07:37 +0800 Subject: [PATCH 036/342] change torch version --- modules/launch_utils.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/modules/launch_utils.py b/modules/launch_utils.py index 8cdbafa50..636da679e 100644 --- a/modules/launch_utils.py +++ b/modules/launch_utils.py @@ -308,8 +308,8 @@ def requirements_met(requirements_file): def prepare_environment(): - torch_index_url = os.environ.get('TORCH_INDEX_URL', "https://download.pytorch.org/whl/cu118") - torch_command = os.environ.get('TORCH_COMMAND', f"pip install torch==2.0.1 torchvision==0.15.2 --extra-index-url {torch_index_url}") + torch_index_url = os.environ.get('TORCH_INDEX_URL', "https://download.pytorch.org/whl/cu121") + torch_command = os.environ.get('TORCH_COMMAND', f"pip install torch==2.1.0 torchvision==0.16.0 --extra-index-url {torch_index_url}") requirements_file = os.environ.get('REQS_FILE', "requirements_versions.txt") xformers_package = os.environ.get('XFORMERS_PACKAGE', 'xformers==0.0.20') From 5121846d34d74aee9b55d48d35c1559a710051b0 Mon Sep 17 00:00:00 2001 From: Won-Kyu Park Date: Wed, 25 Oct 2023 21:37:55 +0900 Subject: [PATCH 037/342] call state.jobnext() before postproces*() --- modules/processing.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index 40598f5cf..70ad1ebed 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -886,6 +886,8 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: devices.torch_gc() + state.nextjob() + if p.scripts is not None: p.scripts.postprocess_batch(p, x_samples_ddim, batch_number=n) @@ -958,8 +960,6 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: devices.torch_gc() - state.nextjob() - if not infotexts: infotexts.append(Processed(p, []).infotext(p, 0)) From d4d3134f6d2d232c7bcfa80900a362921e644976 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Sat, 28 Oct 2023 15:24:26 +0800 Subject: [PATCH 038/342] ManualCast for 10/16 series gpu --- modules/devices.py | 53 +++++++++++++++++++++++++++++++++++++++---- modules/processing.py | 2 +- modules/sd_models.py | 21 +++++++++-------- 3 files changed, 62 insertions(+), 14 deletions(-) diff --git a/modules/devices.py b/modules/devices.py index 0cd2b55d2..c05f2b35e 100644 --- a/modules/devices.py +++ b/modules/devices.py @@ -16,6 +16,23 @@ def has_mps() -> bool: return mac_specific.has_mps +def cuda_no_autocast(device_id=None) -> bool: + if device_id is None: + device_id = get_cuda_device_id() + return ( + torch.cuda.get_device_capability(device_id) == (7, 5) + and torch.cuda.get_device_name(device_id).startswith("NVIDIA GeForce GTX 16") + ) + + +def get_cuda_device_id(): + return ( + int(shared.cmd_opts.device_id) + if shared.cmd_opts.device_id is not None and shared.cmd_opts.device_id.isdigit() + else 0 + ) or torch.cuda.current_device() + + def get_cuda_device_string(): if shared.cmd_opts.device_id is not None: return f"cuda:{shared.cmd_opts.device_id}" @@ -60,8 +77,7 @@ def enable_tf32(): # enabling benchmark option seems to enable a range of cards to do fp16 when they otherwise can't # see https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/4407 - device_id = (int(shared.cmd_opts.device_id) if shared.cmd_opts.device_id is not None and shared.cmd_opts.device_id.isdigit() else 0) or torch.cuda.current_device() - if torch.cuda.get_device_capability(device_id) == (7, 5) and torch.cuda.get_device_name(device_id).startswith("NVIDIA GeForce GTX 16"): + if cuda_no_autocast(): torch.backends.cudnn.benchmark = True torch.backends.cuda.matmul.allow_tf32 = True @@ -92,15 +108,44 @@ def cond_cast_float(input): nv_rng = None +patch_module_list = [ + torch.nn.Linear, + torch.nn.Conv2d, + torch.nn.MultiheadAttention, + torch.nn.GroupNorm, + torch.nn.LayerNorm, +] + +@contextlib.contextmanager +def manual_autocast(): + def manual_cast_forward(self, *args, **kwargs): + org_dtype = next(self.parameters()).dtype + self.to(dtype) + result = self.org_forward(*args, **kwargs) + self.to(org_dtype) + return result + for module_type in patch_module_list: + org_forward = module_type.forward + module_type.forward = manual_cast_forward + module_type.org_forward = org_forward + try: + yield None + finally: + for module_type in patch_module_list: + module_type.forward = module_type.org_forward -def autocast(disable=False, unet=False): +def autocast(disable=False): + print(fp8, dtype, shared.cmd_opts.precision, device) if disable: return contextlib.nullcontext() - if unet and fp8 and device==cpu: + if fp8 and device==cpu: return torch.autocast("cpu", dtype=torch.bfloat16, enabled=True) + if fp8 and (dtype == torch.float32 or shared.cmd_opts.precision == "full" or cuda_no_autocast()): + return manual_autocast() + if dtype == torch.float32 or shared.cmd_opts.precision == "full": return contextlib.nullcontext() diff --git a/modules/processing.py b/modules/processing.py index 2df8a7ea7..40598f5cf 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -865,7 +865,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: if p.n_iter > 1: shared.state.job = f"Batch {n+1} out of {p.n_iter}" - with devices.without_autocast() if devices.unet_needs_upcast else devices.autocast(unet=True): + with devices.without_autocast() if devices.unet_needs_upcast else devices.autocast(): samples_ddim = p.sample(conditioning=p.c, unconditional_conditioning=p.uc, seeds=p.seeds, subseeds=p.subseeds, subseed_strength=p.subseed_strength, prompts=p.prompts) if getattr(samples_ddim, 'already_decoded', False): diff --git a/modules/sd_models.py b/modules/sd_models.py index ccb6afd2a..31bcb913a 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -403,23 +403,26 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer if enable_fp8: devices.fp8 = True + if model.is_sdxl: + cond_stage = model.conditioner + else: + cond_stage = model.cond_stage_model + + for module in cond_stage.modules(): + if isinstance(module, torch.nn.Linear): + module.to(torch.float8_e4m3fn) + if devices.device == devices.cpu: for module in model.model.diffusion_model.modules(): if isinstance(module, torch.nn.Conv2d): module.to(torch.float8_e4m3fn) elif isinstance(module, torch.nn.Linear): module.to(torch.float8_e4m3fn) - timer.record("apply fp8 unet for cpu") else: - if model.is_sdxl: - cond_stage = model.conditioner - else: - cond_stage = model.cond_stage_model - for module in cond_stage.modules(): - if isinstance(module, torch.nn.Linear): - module.to(torch.float8_e4m3fn) model.model.diffusion_model = model.model.diffusion_model.to(torch.float8_e4m3fn) - timer.record("apply fp8 unet") + timer.record("apply fp8") + else: + devices.fp8 = False devices.unet_needs_upcast = shared.cmd_opts.upcast_sampling and devices.dtype == torch.float16 and devices.dtype_unet == torch.float16 From ddc2a3499b8cd120b4a42358bcd33137ce1d1e75 Mon Sep 17 00:00:00 2001 From: KohakuBlueleaf Date: Sat, 28 Oct 2023 16:52:35 +0800 Subject: [PATCH 039/342] Add MPS manual cast --- modules/devices.py | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/modules/devices.py b/modules/devices.py index c05f2b35e..d7c905c28 100644 --- a/modules/devices.py +++ b/modules/devices.py @@ -121,6 +121,8 @@ def manual_autocast(): def manual_cast_forward(self, *args, **kwargs): org_dtype = next(self.parameters()).dtype self.to(dtype) + args = [arg.to(dtype) if isinstance(arg, torch.Tensor) else arg for arg in args] + kwargs = {k: v.to(dtype) if isinstance(v, torch.Tensor) else v for k, v in kwargs.items()} result = self.org_forward(*args, **kwargs) self.to(org_dtype) return result @@ -136,7 +138,6 @@ def manual_autocast(): def autocast(disable=False): - print(fp8, dtype, shared.cmd_opts.precision, device) if disable: return contextlib.nullcontext() @@ -146,6 +147,9 @@ def autocast(disable=False): if fp8 and (dtype == torch.float32 or shared.cmd_opts.precision == "full" or cuda_no_autocast()): return manual_autocast() + if has_mps() and shared.cmd_opts.precision != "full": + return manual_autocast() + if dtype == torch.float32 or shared.cmd_opts.precision == "full": return contextlib.nullcontext() From fbc5c531b9cfa949d60dae19420d01f8af186b55 Mon Sep 17 00:00:00 2001 From: Meerkov Date: Sun, 29 Oct 2023 15:37:08 -0700 Subject: [PATCH 040/342] Fix #13796 Fix comment error that makes understanding scheduling more confusing. --- modules/prompt_parser.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/prompt_parser.py b/modules/prompt_parser.py index 334efeef3..86b7acb50 100644 --- a/modules/prompt_parser.py +++ b/modules/prompt_parser.py @@ -5,7 +5,7 @@ from collections import namedtuple from typing import List import lark -# a prompt like this: "fantasy landscape with a [mountain:lake:0.25] and [an oak:a christmas tree:0.75][ in foreground::0.6][ in background:0.25] [shoddy:masterful:0.5]" +# a prompt like this: "fantasy landscape with a [mountain:lake:0.25] and [an oak:a christmas tree:0.75][ in foreground::0.6][: in background:0.25] [shoddy:masterful:0.5]" # will be represented with prompt_schedule like this (assuming steps=100): # [25, 'fantasy landscape with a mountain and an oak in foreground shoddy'] # [50, 'fantasy landscape with a lake and an oak in foreground in background shoddy'] From a2fad6ee055f3f4e98e46b6c2d912776fe608214 Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Wed, 1 Nov 2023 22:34:27 -0700 Subject: [PATCH 041/342] test implementation based on kohaku diag-oft implementation --- extensions-builtin/Lora/network_oft.py | 57 +++++++++++++++++--------- 1 file changed, 37 insertions(+), 20 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index e43c9a1df..ff61b3699 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -1,5 +1,6 @@ import torch import network +from einops import rearrange class ModuleTypeOFT(network.ModuleType): @@ -30,35 +31,51 @@ class NetworkModuleOFT(network.NetworkModule): self.org_module: list[torch.Module] = [self.sd_module] - def merge_weight(self, R_weight, org_weight): - R_weight = R_weight.to(org_weight.device, dtype=org_weight.dtype) - if org_weight.dim() == 4: - weight = torch.einsum("oihw, op -> pihw", org_weight, R_weight) - else: - weight = torch.einsum("oi, op -> pi", org_weight, R_weight) - return weight + # def merge_weight(self, R_weight, org_weight): + # R_weight = R_weight.to(org_weight.device, dtype=org_weight.dtype) + # if org_weight.dim() == 4: + # weight = torch.einsum("oihw, op -> pihw", org_weight, R_weight) + # else: + # weight = torch.einsum("oi, op -> pi", org_weight, R_weight) + # weight = torch.einsum( + # "k n m, k n ... -> k m ...", + # self.oft_diag * scale + torch.eye(self.block_size, device=device), + # org_weight + # ) + # return weight def get_weight(self, oft_blocks, multiplier=None): - constraint = self.constraint.to(oft_blocks.device, dtype=oft_blocks.dtype) + # constraint = self.constraint.to(oft_blocks.device, dtype=oft_blocks.dtype) - block_Q = oft_blocks - oft_blocks.transpose(1, 2) - norm_Q = torch.norm(block_Q.flatten()) - new_norm_Q = torch.clamp(norm_Q, max=constraint) - block_Q = block_Q * ((new_norm_Q + 1e-8) / (norm_Q + 1e-8)) - m_I = torch.eye(self.block_size, device=oft_blocks.device).unsqueeze(0).repeat(self.num_blocks, 1, 1) - block_R = torch.matmul(m_I + block_Q, (m_I - block_Q).inverse()) + # block_Q = oft_blocks - oft_blocks.transpose(1, 2) + # norm_Q = torch.norm(block_Q.flatten()) + # new_norm_Q = torch.clamp(norm_Q, max=constraint) + # block_Q = block_Q * ((new_norm_Q + 1e-8) / (norm_Q + 1e-8)) + # m_I = torch.eye(self.block_size, device=oft_blocks.device).unsqueeze(0).repeat(self.num_blocks, 1, 1) + # block_R = torch.matmul(m_I + block_Q, (m_I - block_Q).inverse()) - block_R_weighted = multiplier * block_R + (1 - multiplier) * m_I - R = torch.block_diag(*block_R_weighted) + # block_R_weighted = multiplier * block_R + (1 - multiplier) * m_I + # R = torch.block_diag(*block_R_weighted) + #return R + return self.oft_blocks - return R def calc_updown(self, orig_weight): multiplier = self.multiplier() * self.calc_scale() - R = self.get_weight(self.oft_blocks, multiplier) - merged_weight = self.merge_weight(R, orig_weight) + #R = self.get_weight(self.oft_blocks, multiplier) + R = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) + #merged_weight = self.merge_weight(R, orig_weight) - updown = merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight + orig_weight = rearrange(orig_weight, '(k n) ... -> k n ...', k=self.num_blocks, n=self.block_size) + weight = torch.einsum( + 'k n m, k n ... -> k m ...', + R * multiplier + torch.eye(self.block_size, device=orig_weight.device), + orig_weight + ) + weight = rearrange(weight, 'k m ... -> (k m) ...') + + #updown = merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight + updown = weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight output_shape = orig_weight.shape orig_weight = orig_weight From 65ccd6305fcf72347d5ed68f03095dced865ef6e Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Thu, 2 Nov 2023 00:11:32 -0700 Subject: [PATCH 042/342] detect diag_oft type --- extensions-builtin/Lora/networks.py | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/extensions-builtin/Lora/networks.py b/extensions-builtin/Lora/networks.py index 78a97033d..7f814706a 100644 --- a/extensions-builtin/Lora/networks.py +++ b/extensions-builtin/Lora/networks.py @@ -191,10 +191,17 @@ def load_network(name, network_on_disk): key = key_network_without_network_parts.replace("lora_te1_text_model", "transformer_text_model") sd_module = shared.sd_model.network_layer_mapping.get(key, None) + # kohya_ss OFT module elif sd_module is None and "oft_unet" in key_network_without_network_parts: key = key_network_without_network_parts.replace("oft_unet", "diffusion_model") sd_module = shared.sd_model.network_layer_mapping.get(key, None) + # KohakuBlueLeaf OFT module + if sd_module is None and "oft_diag" in key: + key = key_network_without_network_parts.replace("lora_unet", "diffusion_model") + key = key_network_without_network_parts.replace("lora_te1_text_model", "0_transformer_text_model") + sd_module = shared.sd_model.network_layer_mapping.get(key, None) + if sd_module is None: keys_failed_to_match[key_network] = key continue From d727ddfccdc6d474767be9dc3bf504150e81a8a5 Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Thu, 2 Nov 2023 00:13:11 -0700 Subject: [PATCH 043/342] no idea what i'm doing, trying to support both type of OFT, kblueleaf diag_oft has MultiheadAttn which kohya's doesn't?, attempt create new module based off network_lora.py, errors about tensor dim mismatch --- extensions-builtin/Lora/network_oft.py | 186 +++++++++++++++++++------ 1 file changed, 142 insertions(+), 44 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index ff61b3699..e102eafc1 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -1,11 +1,12 @@ import torch import network from einops import rearrange +from modules import devices class ModuleTypeOFT(network.ModuleType): def create_module(self, net: network.Network, weights: network.NetworkWeights): - if all(x in weights.w for x in ["oft_blocks"]): + if all(x in weights.w for x in ["oft_blocks"]) or all(x in weights.w for x in ["oft_diag"]): return NetworkModuleOFT(net, weights) return None @@ -16,66 +17,117 @@ class NetworkModuleOFT(network.NetworkModule): super().__init__(net, weights) - self.oft_blocks = weights.w["oft_blocks"] - self.alpha = weights.w["alpha"] - self.dim = self.oft_blocks.shape[0] - self.num_blocks = self.dim + self.lin_module = None + # kohya-ss + if "oft_blocks" in weights.w.keys(): + self.is_kohya = True + self.oft_blocks = weights.w["oft_blocks"] + self.alpha = weights.w["alpha"] + self.dim = self.oft_blocks.shape[0] + elif "oft_diag" in weights.w.keys(): + self.is_kohya = False + self.oft_blocks = weights.w["oft_diag"] + # alpha is rank if alpha is 0 or None + if self.alpha is None: + pass + self.dim = self.oft_blocks.shape[0] # FIXME: almost certainly incorrect, assumes tensor is shape [*, m, n] + else: + raise ValueError("oft_blocks or oft_diag must be in weights dict") - if "Linear" in self.sd_module.__class__.__name__: + is_linear = type(self.sd_module) in [torch.nn.Linear, torch.nn.modules.linear.NonDynamicallyQuantizableLinear] + is_conv = type(self.sd_module) in [torch.nn.Conv2d] + is_other_linear = type(self.sd_module) in [ torch.nn.MultiheadAttention] + #if "Linear" in self.sd_module.__class__.__name__ or is_linear: + if is_linear: self.out_dim = self.sd_module.out_features - elif "Conv" in self.sd_module.__class__.__name__: + #elif hasattr(self.sd_module, "embed_dim"): + # self.out_dim = self.sd_module.embed_dim + #else: + # raise ValueError("Linear sd_module must have out_features or embed_dim") + elif is_other_linear: + self.out_dim = self.sd_module.embed_dim + elif is_conv: self.out_dim = self.sd_module.out_channels + else: + raise ValueError("sd_module must be Linear or Conv") - self.constraint = self.alpha * self.out_dim - self.block_size = self.out_dim // self.num_blocks + + if self.is_kohya: + self.num_blocks = self.dim + self.block_size = self.out_dim // self.num_blocks + self.constraint = self.alpha * self.out_dim + #elif is_linear or is_conv: + else: + self.num_blocks, self.block_size = factorization(self.out_dim, self.dim) + self.constraint = None self.org_module: list[torch.Module] = [self.sd_module] - # def merge_weight(self, R_weight, org_weight): - # R_weight = R_weight.to(org_weight.device, dtype=org_weight.dtype) - # if org_weight.dim() == 4: - # weight = torch.einsum("oihw, op -> pihw", org_weight, R_weight) - # else: - # weight = torch.einsum("oi, op -> pi", org_weight, R_weight) - # weight = torch.einsum( - # "k n m, k n ... -> k m ...", - # self.oft_diag * scale + torch.eye(self.block_size, device=device), - # org_weight - # ) - # return weight + # if is_other_linear: + # weight = self.oft_blocks.reshape(self.oft_blocks.shape[0], -1) + # module = torch.nn.Linear(weight.shape[1], weight.shape[0], bias=False) + # with torch.no_grad(): + # if weight.shape != module.weight.shape: + # weight = weight.reshape(module.weight.shape) + # module.weight.copy_(weight) + # module.to(device=devices.cpu, dtype=devices.dtype) + # module.weight.requires_grad_(False) + # self.lin_module = module + #return module + + def merge_weight(self, R_weight, org_weight): + R_weight = R_weight.to(org_weight.device, dtype=org_weight.dtype) + if org_weight.dim() == 4: + weight = torch.einsum("oihw, op -> pihw", org_weight, R_weight) + else: + weight = torch.einsum("oi, op -> pi", org_weight, R_weight) + #weight = torch.einsum( + # "k n m, k n ... -> k m ...", + # self.oft_diag * scale + torch.eye(self.block_size, device=device), + # org_weight + #) + return weight def get_weight(self, oft_blocks, multiplier=None): - # constraint = self.constraint.to(oft_blocks.device, dtype=oft_blocks.dtype) + if self.constraint is not None: + constraint = self.constraint.to(oft_blocks.device, dtype=oft_blocks.dtype) - # block_Q = oft_blocks - oft_blocks.transpose(1, 2) - # norm_Q = torch.norm(block_Q.flatten()) - # new_norm_Q = torch.clamp(norm_Q, max=constraint) - # block_Q = block_Q * ((new_norm_Q + 1e-8) / (norm_Q + 1e-8)) - # m_I = torch.eye(self.block_size, device=oft_blocks.device).unsqueeze(0).repeat(self.num_blocks, 1, 1) - # block_R = torch.matmul(m_I + block_Q, (m_I - block_Q).inverse()) + block_Q = oft_blocks - oft_blocks.transpose(1, 2) + norm_Q = torch.norm(block_Q.flatten()) + if self.constraint is not None: + new_norm_Q = torch.clamp(norm_Q, max=constraint) + else: + new_norm_Q = norm_Q + block_Q = block_Q * ((new_norm_Q + 1e-8) / (norm_Q + 1e-8)) + m_I = torch.eye(self.block_size, device=oft_blocks.device).unsqueeze(0).repeat(self.num_blocks, 1, 1) + block_R = torch.matmul(m_I + block_Q, (m_I - block_Q).inverse()) - # block_R_weighted = multiplier * block_R + (1 - multiplier) * m_I - # R = torch.block_diag(*block_R_weighted) - #return R - return self.oft_blocks + block_R_weighted = multiplier * block_R + (1 - multiplier) * m_I + R = torch.block_diag(*block_R_weighted) + return R + #return self.oft_blocks def calc_updown(self, orig_weight): multiplier = self.multiplier() * self.calc_scale() - #R = self.get_weight(self.oft_blocks, multiplier) - R = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) - #merged_weight = self.merge_weight(R, orig_weight) + R = self.get_weight(self.oft_blocks, multiplier) + #R = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) + merged_weight = self.merge_weight(R, orig_weight) - orig_weight = rearrange(orig_weight, '(k n) ... -> k n ...', k=self.num_blocks, n=self.block_size) - weight = torch.einsum( - 'k n m, k n ... -> k m ...', - R * multiplier + torch.eye(self.block_size, device=orig_weight.device), - orig_weight - ) - weight = rearrange(weight, 'k m ... -> (k m) ...') + #if self.lin_module is not None: + # R = self.lin_module.weight.to(orig_weight.device, dtype=orig_weight.dtype) + # weight = torch.mul(torch.mul(R, multiplier), orig_weight) + #else: + # orig_weight = rearrange(orig_weight, '(k n) ... -> k n ...', k=self.num_blocks, n=self.block_size) + # weight = torch.einsum( + # 'k n m, k n ... -> k m ...', + # R * multiplier + torch.eye(self.block_size, device=orig_weight.device), + # orig_weight + # ) + # weight = rearrange(weight, 'k m ... -> (k m) ...') - #updown = merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight - updown = weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight + updown = merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight + #updown = weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight output_shape = orig_weight.shape orig_weight = orig_weight @@ -100,3 +152,49 @@ class NetworkModuleOFT(network.NetworkModule): ex_bias = ex_bias * self.multiplier() return updown, ex_bias + +# copied from https://github.com/KohakuBlueleaf/LyCORIS/blob/dev/lycoris/modules/lokr.py +def factorization(dimension: int, factor:int=-1) -> tuple[int, int]: + ''' + return a tuple of two value of input dimension decomposed by the number closest to factor + second value is higher or equal than first value. + + In LoRA with Kroneckor Product, first value is a value for weight scale. + secon value is a value for weight. + + Becuase of non-commutative property, A⊗B ≠ B⊗A. Meaning of two matrices is slightly different. + + examples) + factor + -1 2 4 8 16 ... + 127 -> 1, 127 127 -> 1, 127 127 -> 1, 127 127 -> 1, 127 127 -> 1, 127 + 128 -> 8, 16 128 -> 2, 64 128 -> 4, 32 128 -> 8, 16 128 -> 8, 16 + 250 -> 10, 25 250 -> 2, 125 250 -> 2, 125 250 -> 5, 50 250 -> 10, 25 + 360 -> 8, 45 360 -> 2, 180 360 -> 4, 90 360 -> 8, 45 360 -> 12, 30 + 512 -> 16, 32 512 -> 2, 256 512 -> 4, 128 512 -> 8, 64 512 -> 16, 32 + 1024 -> 32, 32 1024 -> 2, 512 1024 -> 4, 256 1024 -> 8, 128 1024 -> 16, 64 + ''' + + if factor > 0 and (dimension % factor) == 0: + m = factor + n = dimension // factor + if m > n: + n, m = m, n + return m, n + if factor < 0: + factor = dimension + m, n = 1, dimension + length = m + n + while m length or new_m>factor: + break + else: + m, n = new_m, new_n + if m > n: + n, m = m, n + return m, n + From 759515316e8ec536f34fad616e8c6a33674a164b Mon Sep 17 00:00:00 2001 From: Emily Zeng Date: Thu, 2 Nov 2023 21:54:48 -0400 Subject: [PATCH 044/342] added accordion settings options --- modules/shared_options.py | 2 + modules/ui.py | 480 +++++++++++++++++++------------------- 2 files changed, 243 insertions(+), 239 deletions(-) diff --git a/modules/shared_options.py b/modules/shared_options.py index 0a82216ff..5b07dd041 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -270,6 +270,8 @@ options_templates.update(options_section(('ui', "User interface"), { "hires_fix_show_sampler": OptionInfo(False, "Hires fix: show hires checkpoint and sampler selection").needs_reload_ui(), "hires_fix_show_prompts": OptionInfo(False, "Hires fix: show hires prompt and negative prompt").needs_reload_ui(), "disable_token_counters": OptionInfo(False, "Disable prompt token counters").needs_reload_ui(), + "txt2img_settings_accordion": OptionInfo(False, "Settings in txt2img hidden under Accordion").needs_reload_ui(), + "img2img_settings_accordion": OptionInfo(False, "Settings in img2img hidden under Accordion").needs_reload_ui(), })) diff --git a/modules/ui.py b/modules/ui.py index bcf391997..d05b9f55d 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -344,85 +344,86 @@ def create_ui(): extra_tabs.__enter__() with gr.Tab("Generation", id="txt2img_generation") as txt2img_generation_tab, ResizeHandleRow(equal_height=False): - with gr.Column(variant='compact', elem_id="txt2img_settings"): - scripts.scripts_txt2img.prepare_ui() + with gr.Accordion("Open for Settings", open=False) if shared.opts.img2img_settings_accordion else gr.Group(): + with gr.Column(variant='compact', elem_id="txt2img_settings"): + scripts.scripts_txt2img.prepare_ui() - for category in ordered_ui_categories(): - if category == "sampler": - steps, sampler_name = create_sampler_and_steps_selection(sd_samplers.visible_sampler_names(), "txt2img") + for category in ordered_ui_categories(): + if category == "sampler": + steps, sampler_name = create_sampler_and_steps_selection(sd_samplers.visible_sampler_names(), "txt2img") - elif category == "dimensions": - with FormRow(): - with gr.Column(elem_id="txt2img_column_size", scale=4): - width = gr.Slider(minimum=64, maximum=2048, step=8, label="Width", value=512, elem_id="txt2img_width") - height = gr.Slider(minimum=64, maximum=2048, step=8, label="Height", value=512, elem_id="txt2img_height") + elif category == "dimensions": + with FormRow(): + with gr.Column(elem_id="txt2img_column_size", scale=4): + width = gr.Slider(minimum=64, maximum=2048, step=8, label="Width", value=512, elem_id="txt2img_width") + height = gr.Slider(minimum=64, maximum=2048, step=8, label="Height", value=512, elem_id="txt2img_height") - with gr.Column(elem_id="txt2img_dimensions_row", scale=1, elem_classes="dimensions-tools"): - res_switch_btn = ToolButton(value=switch_values_symbol, elem_id="txt2img_res_switch_btn", tooltip="Switch width/height") + with gr.Column(elem_id="txt2img_dimensions_row", scale=1, elem_classes="dimensions-tools"): + res_switch_btn = ToolButton(value=switch_values_symbol, elem_id="txt2img_res_switch_btn", tooltip="Switch width/height") - if opts.dimensions_and_batch_together: - with gr.Column(elem_id="txt2img_column_batch"): + if opts.dimensions_and_batch_together: + with gr.Column(elem_id="txt2img_column_batch"): + batch_count = gr.Slider(minimum=1, step=1, label='Batch count', value=1, elem_id="txt2img_batch_count") + batch_size = gr.Slider(minimum=1, maximum=8, step=1, label='Batch size', value=1, elem_id="txt2img_batch_size") + + elif category == "cfg": + with gr.Row(): + cfg_scale = gr.Slider(minimum=1.0, maximum=30.0, step=0.5, label='CFG Scale', value=7.0, elem_id="txt2img_cfg_scale") + + elif category == "checkboxes": + with FormRow(elem_classes="checkboxes-row", variant="compact"): + pass + + elif category == "accordions": + with gr.Row(elem_id="txt2img_accordions", elem_classes="accordions"): + with InputAccordion(False, label="Hires. fix", elem_id="txt2img_hr") as enable_hr: + with enable_hr.extra(): + hr_final_resolution = FormHTML(value="", elem_id="txtimg_hr_finalres", label="Upscaled resolution", interactive=False, min_width=0) + + with FormRow(elem_id="txt2img_hires_fix_row1", variant="compact"): + hr_upscaler = gr.Dropdown(label="Upscaler", elem_id="txt2img_hr_upscaler", choices=[*shared.latent_upscale_modes, *[x.name for x in shared.sd_upscalers]], value=shared.latent_upscale_default_mode) + hr_second_pass_steps = gr.Slider(minimum=0, maximum=150, step=1, label='Hires steps', value=0, elem_id="txt2img_hires_steps") + denoising_strength = gr.Slider(minimum=0.0, maximum=1.0, step=0.01, label='Denoising strength', value=0.7, elem_id="txt2img_denoising_strength") + + with FormRow(elem_id="txt2img_hires_fix_row2", variant="compact"): + hr_scale = gr.Slider(minimum=1.0, maximum=4.0, step=0.05, label="Upscale by", value=2.0, elem_id="txt2img_hr_scale") + hr_resize_x = gr.Slider(minimum=0, maximum=2048, step=8, label="Resize width to", value=0, elem_id="txt2img_hr_resize_x") + hr_resize_y = gr.Slider(minimum=0, maximum=2048, step=8, label="Resize height to", value=0, elem_id="txt2img_hr_resize_y") + + with FormRow(elem_id="txt2img_hires_fix_row3", variant="compact", visible=opts.hires_fix_show_sampler) as hr_sampler_container: + + hr_checkpoint_name = gr.Dropdown(label='Hires checkpoint', elem_id="hr_checkpoint", choices=["Use same checkpoint"] + modules.sd_models.checkpoint_tiles(use_short=True), value="Use same checkpoint") + create_refresh_button(hr_checkpoint_name, modules.sd_models.list_models, lambda: {"choices": ["Use same checkpoint"] + modules.sd_models.checkpoint_tiles(use_short=True)}, "hr_checkpoint_refresh") + + hr_sampler_name = gr.Dropdown(label='Hires sampling method', elem_id="hr_sampler", choices=["Use same sampler"] + sd_samplers.visible_sampler_names(), value="Use same sampler") + + with FormRow(elem_id="txt2img_hires_fix_row4", variant="compact", visible=opts.hires_fix_show_prompts) as hr_prompts_container: + with gr.Column(scale=80): + with gr.Row(): + hr_prompt = gr.Textbox(label="Hires prompt", elem_id="hires_prompt", show_label=False, lines=3, placeholder="Prompt for hires fix pass.\nLeave empty to use the same prompt as in first pass.", elem_classes=["prompt"]) + with gr.Column(scale=80): + with gr.Row(): + hr_negative_prompt = gr.Textbox(label="Hires negative prompt", elem_id="hires_neg_prompt", show_label=False, lines=3, placeholder="Negative prompt for hires fix pass.\nLeave empty to use the same negative prompt as in first pass.", elem_classes=["prompt"]) + + scripts.scripts_txt2img.setup_ui_for_section(category) + + elif category == "batch": + if not opts.dimensions_and_batch_together: + with FormRow(elem_id="txt2img_column_batch"): batch_count = gr.Slider(minimum=1, step=1, label='Batch count', value=1, elem_id="txt2img_batch_count") batch_size = gr.Slider(minimum=1, maximum=8, step=1, label='Batch size', value=1, elem_id="txt2img_batch_size") - elif category == "cfg": - with gr.Row(): - cfg_scale = gr.Slider(minimum=1.0, maximum=30.0, step=0.5, label='CFG Scale', value=7.0, elem_id="txt2img_cfg_scale") + elif category == "override_settings": + with FormRow(elem_id="txt2img_override_settings_row") as row: + override_settings = create_override_settings_dropdown('txt2img', row) - elif category == "checkboxes": - with FormRow(elem_classes="checkboxes-row", variant="compact"): - pass - - elif category == "accordions": - with gr.Row(elem_id="txt2img_accordions", elem_classes="accordions"): - with InputAccordion(False, label="Hires. fix", elem_id="txt2img_hr") as enable_hr: - with enable_hr.extra(): - hr_final_resolution = FormHTML(value="", elem_id="txtimg_hr_finalres", label="Upscaled resolution", interactive=False, min_width=0) - - with FormRow(elem_id="txt2img_hires_fix_row1", variant="compact"): - hr_upscaler = gr.Dropdown(label="Upscaler", elem_id="txt2img_hr_upscaler", choices=[*shared.latent_upscale_modes, *[x.name for x in shared.sd_upscalers]], value=shared.latent_upscale_default_mode) - hr_second_pass_steps = gr.Slider(minimum=0, maximum=150, step=1, label='Hires steps', value=0, elem_id="txt2img_hires_steps") - denoising_strength = gr.Slider(minimum=0.0, maximum=1.0, step=0.01, label='Denoising strength', value=0.7, elem_id="txt2img_denoising_strength") - - with FormRow(elem_id="txt2img_hires_fix_row2", variant="compact"): - hr_scale = gr.Slider(minimum=1.0, maximum=4.0, step=0.05, label="Upscale by", value=2.0, elem_id="txt2img_hr_scale") - hr_resize_x = gr.Slider(minimum=0, maximum=2048, step=8, label="Resize width to", value=0, elem_id="txt2img_hr_resize_x") - hr_resize_y = gr.Slider(minimum=0, maximum=2048, step=8, label="Resize height to", value=0, elem_id="txt2img_hr_resize_y") - - with FormRow(elem_id="txt2img_hires_fix_row3", variant="compact", visible=opts.hires_fix_show_sampler) as hr_sampler_container: - - hr_checkpoint_name = gr.Dropdown(label='Hires checkpoint', elem_id="hr_checkpoint", choices=["Use same checkpoint"] + modules.sd_models.checkpoint_tiles(use_short=True), value="Use same checkpoint") - create_refresh_button(hr_checkpoint_name, modules.sd_models.list_models, lambda: {"choices": ["Use same checkpoint"] + modules.sd_models.checkpoint_tiles(use_short=True)}, "hr_checkpoint_refresh") - - hr_sampler_name = gr.Dropdown(label='Hires sampling method', elem_id="hr_sampler", choices=["Use same sampler"] + sd_samplers.visible_sampler_names(), value="Use same sampler") - - with FormRow(elem_id="txt2img_hires_fix_row4", variant="compact", visible=opts.hires_fix_show_prompts) as hr_prompts_container: - with gr.Column(scale=80): - with gr.Row(): - hr_prompt = gr.Textbox(label="Hires prompt", elem_id="hires_prompt", show_label=False, lines=3, placeholder="Prompt for hires fix pass.\nLeave empty to use the same prompt as in first pass.", elem_classes=["prompt"]) - with gr.Column(scale=80): - with gr.Row(): - hr_negative_prompt = gr.Textbox(label="Hires negative prompt", elem_id="hires_neg_prompt", show_label=False, lines=3, placeholder="Negative prompt for hires fix pass.\nLeave empty to use the same negative prompt as in first pass.", elem_classes=["prompt"]) + elif category == "scripts": + with FormGroup(elem_id="txt2img_script_container"): + custom_inputs = scripts.scripts_txt2img.setup_ui() + if category not in {"accordions"}: scripts.scripts_txt2img.setup_ui_for_section(category) - elif category == "batch": - if not opts.dimensions_and_batch_together: - with FormRow(elem_id="txt2img_column_batch"): - batch_count = gr.Slider(minimum=1, step=1, label='Batch count', value=1, elem_id="txt2img_batch_count") - batch_size = gr.Slider(minimum=1, maximum=8, step=1, label='Batch size', value=1, elem_id="txt2img_batch_size") - - elif category == "override_settings": - with FormRow(elem_id="txt2img_override_settings_row") as row: - override_settings = create_override_settings_dropdown('txt2img', row) - - elif category == "scripts": - with FormGroup(elem_id="txt2img_script_container"): - custom_inputs = scripts.scripts_txt2img.setup_ui() - - if category not in {"accordions"}: - scripts.scripts_txt2img.setup_ui_for_section(category) - hr_resolution_preview_inputs = [enable_hr, width, height, hr_scale, hr_resize_x, hr_resize_y] for component in hr_resolution_preview_inputs: @@ -560,215 +561,216 @@ def create_ui(): extra_tabs.__enter__() with gr.Tab("Generation", id="img2img_generation") as img2img_generation_tab, ResizeHandleRow(equal_height=False): - with gr.Column(variant='compact', elem_id="img2img_settings"): - copy_image_buttons = [] - copy_image_destinations = {} + with gr.Accordion("Open for Settings", open=False) if shared.opts.img2img_settings_accordion else gr.Group(): + with gr.Column(variant='compact', elem_id="img2img_settings"): + copy_image_buttons = [] + copy_image_destinations = {} - def add_copy_image_controls(tab_name, elem): - with gr.Row(variant="compact", elem_id=f"img2img_copy_to_{tab_name}"): - gr.HTML("Copy image to: ", elem_id=f"img2img_label_copy_to_{tab_name}") + def add_copy_image_controls(tab_name, elem): + with gr.Row(variant="compact", elem_id=f"img2img_copy_to_{tab_name}"): + gr.HTML("Copy image to: ", elem_id=f"img2img_label_copy_to_{tab_name}") - for title, name in zip(['img2img', 'sketch', 'inpaint', 'inpaint sketch'], ['img2img', 'sketch', 'inpaint', 'inpaint_sketch']): - if name == tab_name: - gr.Button(title, interactive=False) - copy_image_destinations[name] = elem - continue + for title, name in zip(['img2img', 'sketch', 'inpaint', 'inpaint sketch'], ['img2img', 'sketch', 'inpaint', 'inpaint_sketch']): + if name == tab_name: + gr.Button(title, interactive=False) + copy_image_destinations[name] = elem + continue - button = gr.Button(title) - copy_image_buttons.append((button, name, elem)) + button = gr.Button(title) + copy_image_buttons.append((button, name, elem)) - with gr.Tabs(elem_id="mode_img2img"): - img2img_selected_tab = gr.State(0) + with gr.Tabs(elem_id="mode_img2img"): + img2img_selected_tab = gr.State(0) - with gr.TabItem('img2img', id='img2img', elem_id="img2img_img2img_tab") as tab_img2img: - init_img = gr.Image(label="Image for img2img", elem_id="img2img_image", show_label=False, source="upload", interactive=True, type="pil", tool="editor", image_mode="RGBA", height=opts.img2img_editor_height) - add_copy_image_controls('img2img', init_img) + with gr.TabItem('img2img', id='img2img', elem_id="img2img_img2img_tab") as tab_img2img: + init_img = gr.Image(label="Image for img2img", elem_id="img2img_image", show_label=False, source="upload", interactive=True, type="pil", tool="editor", image_mode="RGBA", height=opts.img2img_editor_height) + add_copy_image_controls('img2img', init_img) - with gr.TabItem('Sketch', id='img2img_sketch', elem_id="img2img_img2img_sketch_tab") as tab_sketch: - sketch = gr.Image(label="Image for img2img", elem_id="img2img_sketch", show_label=False, source="upload", interactive=True, type="pil", tool="color-sketch", image_mode="RGB", height=opts.img2img_editor_height, brush_color=opts.img2img_sketch_default_brush_color) - add_copy_image_controls('sketch', sketch) + with gr.TabItem('Sketch', id='img2img_sketch', elem_id="img2img_img2img_sketch_tab") as tab_sketch: + sketch = gr.Image(label="Image for img2img", elem_id="img2img_sketch", show_label=False, source="upload", interactive=True, type="pil", tool="color-sketch", image_mode="RGB", height=opts.img2img_editor_height, brush_color=opts.img2img_sketch_default_brush_color) + add_copy_image_controls('sketch', sketch) - with gr.TabItem('Inpaint', id='inpaint', elem_id="img2img_inpaint_tab") as tab_inpaint: - init_img_with_mask = gr.Image(label="Image for inpainting with mask", show_label=False, elem_id="img2maskimg", source="upload", interactive=True, type="pil", tool="sketch", image_mode="RGBA", height=opts.img2img_editor_height, brush_color=opts.img2img_inpaint_mask_brush_color) - add_copy_image_controls('inpaint', init_img_with_mask) + with gr.TabItem('Inpaint', id='inpaint', elem_id="img2img_inpaint_tab") as tab_inpaint: + init_img_with_mask = gr.Image(label="Image for inpainting with mask", show_label=False, elem_id="img2maskimg", source="upload", interactive=True, type="pil", tool="sketch", image_mode="RGBA", height=opts.img2img_editor_height, brush_color=opts.img2img_inpaint_mask_brush_color) + add_copy_image_controls('inpaint', init_img_with_mask) - with gr.TabItem('Inpaint sketch', id='inpaint_sketch', elem_id="img2img_inpaint_sketch_tab") as tab_inpaint_color: - inpaint_color_sketch = gr.Image(label="Color sketch inpainting", show_label=False, elem_id="inpaint_sketch", source="upload", interactive=True, type="pil", tool="color-sketch", image_mode="RGB", height=opts.img2img_editor_height, brush_color=opts.img2img_inpaint_sketch_default_brush_color) - inpaint_color_sketch_orig = gr.State(None) - add_copy_image_controls('inpaint_sketch', inpaint_color_sketch) + with gr.TabItem('Inpaint sketch', id='inpaint_sketch', elem_id="img2img_inpaint_sketch_tab") as tab_inpaint_color: + inpaint_color_sketch = gr.Image(label="Color sketch inpainting", show_label=False, elem_id="inpaint_sketch", source="upload", interactive=True, type="pil", tool="color-sketch", image_mode="RGB", height=opts.img2img_editor_height, brush_color=opts.img2img_inpaint_sketch_default_brush_color) + inpaint_color_sketch_orig = gr.State(None) + add_copy_image_controls('inpaint_sketch', inpaint_color_sketch) - def update_orig(image, state): - if image is not None: - same_size = state is not None and state.size == image.size - has_exact_match = np.any(np.all(np.array(image) == np.array(state), axis=-1)) - edited = same_size and has_exact_match - return image if not edited or state is None else state + def update_orig(image, state): + if image is not None: + same_size = state is not None and state.size == image.size + has_exact_match = np.any(np.all(np.array(image) == np.array(state), axis=-1)) + edited = same_size and has_exact_match + return image if not edited or state is None else state - inpaint_color_sketch.change(update_orig, [inpaint_color_sketch, inpaint_color_sketch_orig], inpaint_color_sketch_orig) + inpaint_color_sketch.change(update_orig, [inpaint_color_sketch, inpaint_color_sketch_orig], inpaint_color_sketch_orig) - with gr.TabItem('Inpaint upload', id='inpaint_upload', elem_id="img2img_inpaint_upload_tab") as tab_inpaint_upload: - init_img_inpaint = gr.Image(label="Image for img2img", show_label=False, source="upload", interactive=True, type="pil", elem_id="img_inpaint_base") - init_mask_inpaint = gr.Image(label="Mask", source="upload", interactive=True, type="pil", image_mode="RGBA", elem_id="img_inpaint_mask") + with gr.TabItem('Inpaint upload', id='inpaint_upload', elem_id="img2img_inpaint_upload_tab") as tab_inpaint_upload: + init_img_inpaint = gr.Image(label="Image for img2img", show_label=False, source="upload", interactive=True, type="pil", elem_id="img_inpaint_base") + init_mask_inpaint = gr.Image(label="Mask", source="upload", interactive=True, type="pil", image_mode="RGBA", elem_id="img_inpaint_mask") - with gr.TabItem('Batch', id='batch', elem_id="img2img_batch_tab") as tab_batch: - hidden = '
Disabled when launched with --hide-ui-dir-config.' if shared.cmd_opts.hide_ui_dir_config else '' - gr.HTML( - "

Process images in a directory on the same machine where the server is running." + - "
Use an empty output directory to save pictures normally instead of writing to the output directory." + - f"
Add inpaint batch mask directory to enable inpaint batch processing." - f"{hidden}

" + with gr.TabItem('Batch', id='batch', elem_id="img2img_batch_tab") as tab_batch: + hidden = '
Disabled when launched with --hide-ui-dir-config.' if shared.cmd_opts.hide_ui_dir_config else '' + gr.HTML( + "

Process images in a directory on the same machine where the server is running." + + "
Use an empty output directory to save pictures normally instead of writing to the output directory." + + f"
Add inpaint batch mask directory to enable inpaint batch processing." + f"{hidden}

" + ) + img2img_batch_input_dir = gr.Textbox(label="Input directory", **shared.hide_dirs, elem_id="img2img_batch_input_dir") + img2img_batch_output_dir = gr.Textbox(label="Output directory", **shared.hide_dirs, elem_id="img2img_batch_output_dir") + img2img_batch_inpaint_mask_dir = gr.Textbox(label="Inpaint batch mask directory (required for inpaint batch processing only)", **shared.hide_dirs, elem_id="img2img_batch_inpaint_mask_dir") + with gr.Accordion("PNG info", open=False): + img2img_batch_use_png_info = gr.Checkbox(label="Append png info to prompts", **shared.hide_dirs, elem_id="img2img_batch_use_png_info") + img2img_batch_png_info_dir = gr.Textbox(label="PNG info directory", **shared.hide_dirs, placeholder="Leave empty to use input directory", elem_id="img2img_batch_png_info_dir") + img2img_batch_png_info_props = gr.CheckboxGroup(["Prompt", "Negative prompt", "Seed", "CFG scale", "Sampler", "Steps", "Model hash"], label="Parameters to take from png info", info="Prompts from png info will be appended to prompts set in ui.") + + img2img_tabs = [tab_img2img, tab_sketch, tab_inpaint, tab_inpaint_color, tab_inpaint_upload, tab_batch] + + for i, tab in enumerate(img2img_tabs): + tab.select(fn=lambda tabnum=i: tabnum, inputs=[], outputs=[img2img_selected_tab]) + + def copy_image(img): + if isinstance(img, dict) and 'image' in img: + return img['image'] + + return img + + for button, name, elem in copy_image_buttons: + button.click( + fn=copy_image, + inputs=[elem], + outputs=[copy_image_destinations[name]], + ) + button.click( + fn=lambda: None, + _js=f"switch_to_{name.replace(' ', '_')}", + inputs=[], + outputs=[], ) - img2img_batch_input_dir = gr.Textbox(label="Input directory", **shared.hide_dirs, elem_id="img2img_batch_input_dir") - img2img_batch_output_dir = gr.Textbox(label="Output directory", **shared.hide_dirs, elem_id="img2img_batch_output_dir") - img2img_batch_inpaint_mask_dir = gr.Textbox(label="Inpaint batch mask directory (required for inpaint batch processing only)", **shared.hide_dirs, elem_id="img2img_batch_inpaint_mask_dir") - with gr.Accordion("PNG info", open=False): - img2img_batch_use_png_info = gr.Checkbox(label="Append png info to prompts", **shared.hide_dirs, elem_id="img2img_batch_use_png_info") - img2img_batch_png_info_dir = gr.Textbox(label="PNG info directory", **shared.hide_dirs, placeholder="Leave empty to use input directory", elem_id="img2img_batch_png_info_dir") - img2img_batch_png_info_props = gr.CheckboxGroup(["Prompt", "Negative prompt", "Seed", "CFG scale", "Sampler", "Steps", "Model hash"], label="Parameters to take from png info", info="Prompts from png info will be appended to prompts set in ui.") - img2img_tabs = [tab_img2img, tab_sketch, tab_inpaint, tab_inpaint_color, tab_inpaint_upload, tab_batch] + with FormRow(): + resize_mode = gr.Radio(label="Resize mode", elem_id="resize_mode", choices=["Just resize", "Crop and resize", "Resize and fill", "Just resize (latent upscale)"], type="index", value="Just resize") - for i, tab in enumerate(img2img_tabs): - tab.select(fn=lambda tabnum=i: tabnum, inputs=[], outputs=[img2img_selected_tab]) + scripts.scripts_img2img.prepare_ui() - def copy_image(img): - if isinstance(img, dict) and 'image' in img: - return img['image'] + for category in ordered_ui_categories(): + if category == "sampler": + steps, sampler_name = create_sampler_and_steps_selection(sd_samplers.visible_sampler_names(), "img2img") - return img + elif category == "dimensions": + with FormRow(): + with gr.Column(elem_id="img2img_column_size", scale=4): + selected_scale_tab = gr.State(value=0) - for button, name, elem in copy_image_buttons: - button.click( - fn=copy_image, - inputs=[elem], - outputs=[copy_image_destinations[name]], - ) - button.click( - fn=lambda: None, - _js=f"switch_to_{name.replace(' ', '_')}", - inputs=[], - outputs=[], - ) + with gr.Tabs(): + with gr.Tab(label="Resize to", elem_id="img2img_tab_resize_to") as tab_scale_to: + with FormRow(): + with gr.Column(elem_id="img2img_column_size", scale=4): + width = gr.Slider(minimum=64, maximum=2048, step=8, label="Width", value=512, elem_id="img2img_width") + height = gr.Slider(minimum=64, maximum=2048, step=8, label="Height", value=512, elem_id="img2img_height") + with gr.Column(elem_id="img2img_dimensions_row", scale=1, elem_classes="dimensions-tools"): + res_switch_btn = ToolButton(value=switch_values_symbol, elem_id="img2img_res_switch_btn", tooltip="Switch width/height") + detect_image_size_btn = ToolButton(value=detect_image_size_symbol, elem_id="img2img_detect_image_size_btn", tooltip="Auto detect size from img2img") - with FormRow(): - resize_mode = gr.Radio(label="Resize mode", elem_id="resize_mode", choices=["Just resize", "Crop and resize", "Resize and fill", "Just resize (latent upscale)"], type="index", value="Just resize") + with gr.Tab(label="Resize by", elem_id="img2img_tab_resize_by") as tab_scale_by: + scale_by = gr.Slider(minimum=0.05, maximum=4.0, step=0.05, label="Scale", value=1.0, elem_id="img2img_scale") - scripts.scripts_img2img.prepare_ui() + with FormRow(): + scale_by_html = FormHTML(resize_from_to_html(0, 0, 0.0), elem_id="img2img_scale_resolution_preview") + gr.Slider(label="Unused", elem_id="img2img_unused_scale_by_slider") + button_update_resize_to = gr.Button(visible=False, elem_id="img2img_update_resize_to") - for category in ordered_ui_categories(): - if category == "sampler": - steps, sampler_name = create_sampler_and_steps_selection(sd_samplers.visible_sampler_names(), "img2img") + on_change_args = dict( + fn=resize_from_to_html, + _js="currentImg2imgSourceResolution", + inputs=[dummy_component, dummy_component, scale_by], + outputs=scale_by_html, + show_progress=False, + ) - elif category == "dimensions": - with FormRow(): - with gr.Column(elem_id="img2img_column_size", scale=4): - selected_scale_tab = gr.State(value=0) + scale_by.release(**on_change_args) + button_update_resize_to.click(**on_change_args) - with gr.Tabs(): - with gr.Tab(label="Resize to", elem_id="img2img_tab_resize_to") as tab_scale_to: - with FormRow(): - with gr.Column(elem_id="img2img_column_size", scale=4): - width = gr.Slider(minimum=64, maximum=2048, step=8, label="Width", value=512, elem_id="img2img_width") - height = gr.Slider(minimum=64, maximum=2048, step=8, label="Height", value=512, elem_id="img2img_height") - with gr.Column(elem_id="img2img_dimensions_row", scale=1, elem_classes="dimensions-tools"): - res_switch_btn = ToolButton(value=switch_values_symbol, elem_id="img2img_res_switch_btn", tooltip="Switch width/height") - detect_image_size_btn = ToolButton(value=detect_image_size_symbol, elem_id="img2img_detect_image_size_btn", tooltip="Auto detect size from img2img") + # the code below is meant to update the resolution label after the image in the image selection UI has changed. + # as it is now the event keeps firing continuously for inpaint edits, which ruins the page with constant requests. + # I assume this must be a gradio bug and for now we'll just do it for non-inpaint inputs. + for component in [init_img, sketch]: + component.change(fn=lambda: None, _js="updateImg2imgResizeToTextAfterChangingImage", inputs=[], outputs=[], show_progress=False) - with gr.Tab(label="Resize by", elem_id="img2img_tab_resize_by") as tab_scale_by: - scale_by = gr.Slider(minimum=0.05, maximum=4.0, step=0.05, label="Scale", value=1.0, elem_id="img2img_scale") + tab_scale_to.select(fn=lambda: 0, inputs=[], outputs=[selected_scale_tab]) + tab_scale_by.select(fn=lambda: 1, inputs=[], outputs=[selected_scale_tab]) - with FormRow(): - scale_by_html = FormHTML(resize_from_to_html(0, 0, 0.0), elem_id="img2img_scale_resolution_preview") - gr.Slider(label="Unused", elem_id="img2img_unused_scale_by_slider") - button_update_resize_to = gr.Button(visible=False, elem_id="img2img_update_resize_to") + if opts.dimensions_and_batch_together: + with gr.Column(elem_id="img2img_column_batch"): + batch_count = gr.Slider(minimum=1, step=1, label='Batch count', value=1, elem_id="img2img_batch_count") + batch_size = gr.Slider(minimum=1, maximum=8, step=1, label='Batch size', value=1, elem_id="img2img_batch_size") - on_change_args = dict( - fn=resize_from_to_html, - _js="currentImg2imgSourceResolution", - inputs=[dummy_component, dummy_component, scale_by], - outputs=scale_by_html, - show_progress=False, - ) + elif category == "denoising": + denoising_strength = gr.Slider(minimum=0.0, maximum=1.0, step=0.01, label='Denoising strength', value=0.75, elem_id="img2img_denoising_strength") - scale_by.release(**on_change_args) - button_update_resize_to.click(**on_change_args) + elif category == "cfg": + with gr.Row(): + cfg_scale = gr.Slider(minimum=1.0, maximum=30.0, step=0.5, label='CFG Scale', value=7.0, elem_id="img2img_cfg_scale") + image_cfg_scale = gr.Slider(minimum=0, maximum=3.0, step=0.05, label='Image CFG Scale', value=1.5, elem_id="img2img_image_cfg_scale", visible=False) - # the code below is meant to update the resolution label after the image in the image selection UI has changed. - # as it is now the event keeps firing continuously for inpaint edits, which ruins the page with constant requests. - # I assume this must be a gradio bug and for now we'll just do it for non-inpaint inputs. - for component in [init_img, sketch]: - component.change(fn=lambda: None, _js="updateImg2imgResizeToTextAfterChangingImage", inputs=[], outputs=[], show_progress=False) + elif category == "checkboxes": + with FormRow(elem_classes="checkboxes-row", variant="compact"): + pass - tab_scale_to.select(fn=lambda: 0, inputs=[], outputs=[selected_scale_tab]) - tab_scale_by.select(fn=lambda: 1, inputs=[], outputs=[selected_scale_tab]) + elif category == "accordions": + with gr.Row(elem_id="img2img_accordions", elem_classes="accordions"): + scripts.scripts_img2img.setup_ui_for_section(category) - if opts.dimensions_and_batch_together: - with gr.Column(elem_id="img2img_column_batch"): + elif category == "batch": + if not opts.dimensions_and_batch_together: + with FormRow(elem_id="img2img_column_batch"): batch_count = gr.Slider(minimum=1, step=1, label='Batch count', value=1, elem_id="img2img_batch_count") batch_size = gr.Slider(minimum=1, maximum=8, step=1, label='Batch size', value=1, elem_id="img2img_batch_size") - elif category == "denoising": - denoising_strength = gr.Slider(minimum=0.0, maximum=1.0, step=0.01, label='Denoising strength', value=0.75, elem_id="img2img_denoising_strength") + elif category == "override_settings": + with FormRow(elem_id="img2img_override_settings_row") as row: + override_settings = create_override_settings_dropdown('img2img', row) - elif category == "cfg": - with gr.Row(): - cfg_scale = gr.Slider(minimum=1.0, maximum=30.0, step=0.5, label='CFG Scale', value=7.0, elem_id="img2img_cfg_scale") - image_cfg_scale = gr.Slider(minimum=0, maximum=3.0, step=0.05, label='Image CFG Scale', value=1.5, elem_id="img2img_image_cfg_scale", visible=False) + elif category == "scripts": + with FormGroup(elem_id="img2img_script_container"): + custom_inputs = scripts.scripts_img2img.setup_ui() - elif category == "checkboxes": - with FormRow(elem_classes="checkboxes-row", variant="compact"): - pass + elif category == "inpaint": + with FormGroup(elem_id="inpaint_controls", visible=False) as inpaint_controls: + with FormRow(): + mask_blur = gr.Slider(label='Mask blur', minimum=0, maximum=64, step=1, value=4, elem_id="img2img_mask_blur") + mask_alpha = gr.Slider(label="Mask transparency", visible=False, elem_id="img2img_mask_alpha") - elif category == "accordions": - with gr.Row(elem_id="img2img_accordions", elem_classes="accordions"): + with FormRow(): + inpainting_mask_invert = gr.Radio(label='Mask mode', choices=['Inpaint masked', 'Inpaint not masked'], value='Inpaint masked', type="index", elem_id="img2img_mask_mode") + + with FormRow(): + inpainting_fill = gr.Radio(label='Masked content', choices=['fill', 'original', 'latent noise', 'latent nothing'], value='original', type="index", elem_id="img2img_inpainting_fill") + + with FormRow(): + with gr.Column(): + inpaint_full_res = gr.Radio(label="Inpaint area", choices=["Whole picture", "Only masked"], type="index", value="Whole picture", elem_id="img2img_inpaint_full_res") + + with gr.Column(scale=4): + inpaint_full_res_padding = gr.Slider(label='Only masked padding, pixels', minimum=0, maximum=256, step=4, value=32, elem_id="img2img_inpaint_full_res_padding") + + def select_img2img_tab(tab): + return gr.update(visible=tab in [2, 3, 4]), gr.update(visible=tab == 3), + + for i, elem in enumerate(img2img_tabs): + elem.select( + fn=lambda tab=i: select_img2img_tab(tab), + inputs=[], + outputs=[inpaint_controls, mask_alpha], + ) + + if category not in {"accordions"}: scripts.scripts_img2img.setup_ui_for_section(category) - elif category == "batch": - if not opts.dimensions_and_batch_together: - with FormRow(elem_id="img2img_column_batch"): - batch_count = gr.Slider(minimum=1, step=1, label='Batch count', value=1, elem_id="img2img_batch_count") - batch_size = gr.Slider(minimum=1, maximum=8, step=1, label='Batch size', value=1, elem_id="img2img_batch_size") - - elif category == "override_settings": - with FormRow(elem_id="img2img_override_settings_row") as row: - override_settings = create_override_settings_dropdown('img2img', row) - - elif category == "scripts": - with FormGroup(elem_id="img2img_script_container"): - custom_inputs = scripts.scripts_img2img.setup_ui() - - elif category == "inpaint": - with FormGroup(elem_id="inpaint_controls", visible=False) as inpaint_controls: - with FormRow(): - mask_blur = gr.Slider(label='Mask blur', minimum=0, maximum=64, step=1, value=4, elem_id="img2img_mask_blur") - mask_alpha = gr.Slider(label="Mask transparency", visible=False, elem_id="img2img_mask_alpha") - - with FormRow(): - inpainting_mask_invert = gr.Radio(label='Mask mode', choices=['Inpaint masked', 'Inpaint not masked'], value='Inpaint masked', type="index", elem_id="img2img_mask_mode") - - with FormRow(): - inpainting_fill = gr.Radio(label='Masked content', choices=['fill', 'original', 'latent noise', 'latent nothing'], value='original', type="index", elem_id="img2img_inpainting_fill") - - with FormRow(): - with gr.Column(): - inpaint_full_res = gr.Radio(label="Inpaint area", choices=["Whole picture", "Only masked"], type="index", value="Whole picture", elem_id="img2img_inpaint_full_res") - - with gr.Column(scale=4): - inpaint_full_res_padding = gr.Slider(label='Only masked padding, pixels', minimum=0, maximum=256, step=4, value=32, elem_id="img2img_inpaint_full_res_padding") - - def select_img2img_tab(tab): - return gr.update(visible=tab in [2, 3, 4]), gr.update(visible=tab == 3), - - for i, elem in enumerate(img2img_tabs): - elem.select( - fn=lambda tab=i: select_img2img_tab(tab), - inputs=[], - outputs=[inpaint_controls, mask_alpha], - ) - - if category not in {"accordions"}: - scripts.scripts_img2img.setup_ui_for_section(category) - img2img_gallery, generation_info, html_info, html_log = create_output_panel("img2img", opts.outdir_img2img_samples) img2img_args = dict( From 8052a4971e1be48e1df2535284a7791cd1ad39ae Mon Sep 17 00:00:00 2001 From: missionfloyd Date: Fri, 3 Nov 2023 00:59:19 -0600 Subject: [PATCH 045/342] Fix parenthesis auto selection Fixes #13813 --- javascript/edit-attention.js | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/javascript/edit-attention.js b/javascript/edit-attention.js index 044641006..688c2f112 100644 --- a/javascript/edit-attention.js +++ b/javascript/edit-attention.js @@ -28,7 +28,7 @@ function keyupEditAttention(event) { if (afterParen == -1) return false; let afterOpeningParen = after.indexOf(OPEN); - if (afterOpeningParen != -1 && afterOpeningParen < beforeParen) return false; + if (afterOpeningParen != -1 && afterOpeningParen < afterParen) return false; // Set the selection to the text between the parenthesis const parenContent = text.substring(beforeParen + 1, selectionStart + afterParen); From cc80a09d82afae793800a033a1f525f5dc797cff Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Sat, 4 Nov 2023 00:50:30 +0900 Subject: [PATCH 046/342] Update requirements_versions.txt --- requirements_versions.txt | 1 + 1 file changed, 1 insertion(+) diff --git a/requirements_versions.txt b/requirements_versions.txt index 7d27f2be3..cb7403a9d 100644 --- a/requirements_versions.txt +++ b/requirements_versions.txt @@ -29,3 +29,4 @@ torch torchdiffeq==0.2.3 torchsde==0.2.6 transformers==4.30.2 +httpx==0.24.1 From bda2ecdbf58fd33b4ad3036ed5cc13eef02747ae Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Fri, 3 Nov 2023 19:44:57 +0300 Subject: [PATCH 047/342] Merge pull request #13839 from AUTOMATIC1111/httpx==0.24.1 requirements_versions httpx==0.24.1 --- requirements_versions.txt | 1 + 1 file changed, 1 insertion(+) diff --git a/requirements_versions.txt b/requirements_versions.txt index f8ae1f385..e84bd4270 100644 --- a/requirements_versions.txt +++ b/requirements_versions.txt @@ -29,3 +29,4 @@ torch torchdiffeq==0.2.3 torchsde==0.2.5 transformers==4.30.2 +httpx==0.24.1 From 4afaaf8a020c1df457bcf7250cb1c7f609699fa7 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Fri, 3 Nov 2023 19:50:14 +0300 Subject: [PATCH 048/342] add changelog entry --- CHANGELOG.md | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/CHANGELOG.md b/CHANGELOG.md index 1cd3572c8..2c72359fc 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -1,3 +1,8 @@ +## 1.6.1 + +### Bug Fixes: + * fix an error causing the webui to fail to start ([#13839](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13839)) + ## 1.6.0 ### Features: From fe1967a4c4a02eccfa45b65ee19a5b0773ced31c Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Fri, 3 Nov 2023 17:52:55 -0700 Subject: [PATCH 049/342] skip multihead attn for now --- extensions-builtin/Lora/network_oft.py | 52 ++++++++++++++++++-------- 1 file changed, 36 insertions(+), 16 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index e102eafc1..979a20476 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -18,6 +18,7 @@ class NetworkModuleOFT(network.NetworkModule): super().__init__(net, weights) self.lin_module = None + self.org_module: list[torch.Module] = [self.sd_module] # kohya-ss if "oft_blocks" in weights.w.keys(): self.is_kohya = True @@ -30,7 +31,7 @@ class NetworkModuleOFT(network.NetworkModule): # alpha is rank if alpha is 0 or None if self.alpha is None: pass - self.dim = self.oft_blocks.shape[0] # FIXME: almost certainly incorrect, assumes tensor is shape [*, m, n] + self.dim = self.oft_blocks.shape[1] # FIXME: almost certainly incorrect, assumes tensor is shape [*, m, n] else: raise ValueError("oft_blocks or oft_diag must be in weights dict") @@ -46,6 +47,12 @@ class NetworkModuleOFT(network.NetworkModule): # raise ValueError("Linear sd_module must have out_features or embed_dim") elif is_other_linear: self.out_dim = self.sd_module.embed_dim + #self.org_weight = self.org_module[0].weight +# if hasattr(self.sd_module, "in_proj_weight"): +# self.in_proj_dim = self.sd_module.in_proj_weight.shape[1] +# if hasattr(self.sd_module, "out_proj_weight"): +# self.out_proj_dim = self.sd_module.out_proj_weight.shape[0] +# self.in_proj_dim = self.sd_module.in_proj_weight.shape[1] elif is_conv: self.out_dim = self.sd_module.out_channels else: @@ -58,10 +65,9 @@ class NetworkModuleOFT(network.NetworkModule): self.constraint = self.alpha * self.out_dim #elif is_linear or is_conv: else: - self.num_blocks, self.block_size = factorization(self.out_dim, self.dim) + self.block_size, self.num_blocks = factorization(self.out_dim, self.dim) self.constraint = None - self.org_module: list[torch.Module] = [self.sd_module] # if is_other_linear: # weight = self.oft_blocks.reshape(self.oft_blocks.shape[0], -1) @@ -110,25 +116,39 @@ class NetworkModuleOFT(network.NetworkModule): def calc_updown(self, orig_weight): multiplier = self.multiplier() * self.calc_scale() - R = self.get_weight(self.oft_blocks, multiplier) - #R = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) - merged_weight = self.merge_weight(R, orig_weight) + is_other_linear = type(self.sd_module) in [ torch.nn.MultiheadAttention] + if self.is_kohya and not is_other_linear: + R = self.get_weight(self.oft_blocks, multiplier) + #R = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) + merged_weight = self.merge_weight(R, orig_weight) + elif not self.is_kohya and not is_other_linear: + if is_other_linear and orig_weight.shape[0] != orig_weight.shape[1]: + orig_weight=orig_weight.permute(1, 0) + R = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) + merged_weight = rearrange(orig_weight, '(k n) ... -> k n ...', k=self.num_blocks, n=self.block_size) + #orig_weight = rearrange(orig_weight, '(k n) ... -> k n ...', k=self.block_size, n=self.num_blocks) + merged_weight = torch.einsum( + 'k n m, k n ... -> k m ...', + R * multiplier + torch.eye(self.block_size, device=orig_weight.device), + merged_weight + ) + merged_weight = rearrange(merged_weight, 'k m ... -> (k m) ...') + if is_other_linear and orig_weight.shape[0] != orig_weight.shape[1]: + orig_weight=orig_weight.permute(1, 0) + #merged_weight=merged_weight.permute(1, 0) + updown = merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight + #updown = weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight + output_shape = orig_weight.shape + else: + # skip for now + updown = torch.zeros([orig_weight.shape[1], orig_weight.shape[1]], device=orig_weight.device, dtype=orig_weight.dtype) + output_shape = (orig_weight.shape[1], orig_weight.shape[1]) #if self.lin_module is not None: # R = self.lin_module.weight.to(orig_weight.device, dtype=orig_weight.dtype) # weight = torch.mul(torch.mul(R, multiplier), orig_weight) #else: - # orig_weight = rearrange(orig_weight, '(k n) ... -> k n ...', k=self.num_blocks, n=self.block_size) - # weight = torch.einsum( - # 'k n m, k n ... -> k m ...', - # R * multiplier + torch.eye(self.block_size, device=orig_weight.device), - # orig_weight - # ) - # weight = rearrange(weight, 'k m ... -> (k m) ...') - updown = merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight - #updown = weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight - output_shape = orig_weight.shape orig_weight = orig_weight return self.finalize_updown(updown, orig_weight, output_shape) From f6c8201e5663ca2182a66c8eca63ce4801d52849 Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Fri, 3 Nov 2023 19:35:15 -0700 Subject: [PATCH 050/342] refactor: move factorization to lyco_helpers, separate calc_updown for kohya and kb --- extensions-builtin/Lora/lyco_helpers.py | 47 +++++++++ extensions-builtin/Lora/network_oft.py | 131 ++++++------------------ 2 files changed, 77 insertions(+), 101 deletions(-) diff --git a/extensions-builtin/Lora/lyco_helpers.py b/extensions-builtin/Lora/lyco_helpers.py index 279b34bc9..1679a0ce6 100644 --- a/extensions-builtin/Lora/lyco_helpers.py +++ b/extensions-builtin/Lora/lyco_helpers.py @@ -19,3 +19,50 @@ def rebuild_cp_decomposition(up, down, mid): up = up.reshape(up.size(0), -1) down = down.reshape(down.size(0), -1) return torch.einsum('n m k l, i n, m j -> i j k l', mid, up, down) + + +# copied from https://github.com/KohakuBlueleaf/LyCORIS/blob/dev/lycoris/modules/lokr.py +def factorization(dimension: int, factor:int=-1) -> tuple[int, int]: + ''' + return a tuple of two value of input dimension decomposed by the number closest to factor + second value is higher or equal than first value. + + In LoRA with Kroneckor Product, first value is a value for weight scale. + secon value is a value for weight. + + Becuase of non-commutative property, A⊗B ≠ B⊗A. Meaning of two matrices is slightly different. + + examples) + factor + -1 2 4 8 16 ... + 127 -> 1, 127 127 -> 1, 127 127 -> 1, 127 127 -> 1, 127 127 -> 1, 127 + 128 -> 8, 16 128 -> 2, 64 128 -> 4, 32 128 -> 8, 16 128 -> 8, 16 + 250 -> 10, 25 250 -> 2, 125 250 -> 2, 125 250 -> 5, 50 250 -> 10, 25 + 360 -> 8, 45 360 -> 2, 180 360 -> 4, 90 360 -> 8, 45 360 -> 12, 30 + 512 -> 16, 32 512 -> 2, 256 512 -> 4, 128 512 -> 8, 64 512 -> 16, 32 + 1024 -> 32, 32 1024 -> 2, 512 1024 -> 4, 256 1024 -> 8, 128 1024 -> 16, 64 + ''' + + if factor > 0 and (dimension % factor) == 0: + m = factor + n = dimension // factor + if m > n: + n, m = m, n + return m, n + if factor < 0: + factor = dimension + m, n = 1, dimension + length = m + n + while m length or new_m>factor: + break + else: + m, n = new_m, new_n + if m > n: + n, m = m, n + return m, n + diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index 979a20476..2be67fe53 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -1,7 +1,7 @@ import torch import network +from lyco_helpers import factorization from einops import rearrange -from modules import devices class ModuleTypeOFT(network.ModuleType): @@ -11,7 +11,8 @@ class ModuleTypeOFT(network.ModuleType): return None -# adapted from kohya's implementation https://github.com/kohya-ss/sd-scripts/blob/main/networks/oft.py +# adapted from kohya-ss' implementation https://github.com/kohya-ss/sd-scripts/blob/main/networks/oft.py +# and KohakuBlueleaf's implementation https://github.com/KohakuBlueleaf/LyCORIS/blob/dev/lycoris/modules/diag_oft.py class NetworkModuleOFT(network.NetworkModule): def __init__(self, net: network.Network, weights: network.NetworkWeights): @@ -19,6 +20,7 @@ class NetworkModuleOFT(network.NetworkModule): self.lin_module = None self.org_module: list[torch.Module] = [self.sd_module] + # kohya-ss if "oft_blocks" in weights.w.keys(): self.is_kohya = True @@ -37,61 +39,31 @@ class NetworkModuleOFT(network.NetworkModule): is_linear = type(self.sd_module) in [torch.nn.Linear, torch.nn.modules.linear.NonDynamicallyQuantizableLinear] is_conv = type(self.sd_module) in [torch.nn.Conv2d] - is_other_linear = type(self.sd_module) in [ torch.nn.MultiheadAttention] - #if "Linear" in self.sd_module.__class__.__name__ or is_linear: + is_other_linear = type(self.sd_module) in [torch.nn.MultiheadAttention] + if is_linear: self.out_dim = self.sd_module.out_features - #elif hasattr(self.sd_module, "embed_dim"): - # self.out_dim = self.sd_module.embed_dim - #else: - # raise ValueError("Linear sd_module must have out_features or embed_dim") elif is_other_linear: self.out_dim = self.sd_module.embed_dim - #self.org_weight = self.org_module[0].weight -# if hasattr(self.sd_module, "in_proj_weight"): -# self.in_proj_dim = self.sd_module.in_proj_weight.shape[1] -# if hasattr(self.sd_module, "out_proj_weight"): -# self.out_proj_dim = self.sd_module.out_proj_weight.shape[0] -# self.in_proj_dim = self.sd_module.in_proj_weight.shape[1] elif is_conv: self.out_dim = self.sd_module.out_channels else: raise ValueError("sd_module must be Linear or Conv") - if self.is_kohya: self.num_blocks = self.dim self.block_size = self.out_dim // self.num_blocks self.constraint = self.alpha * self.out_dim - #elif is_linear or is_conv: else: self.block_size, self.num_blocks = factorization(self.out_dim, self.dim) self.constraint = None - - # if is_other_linear: - # weight = self.oft_blocks.reshape(self.oft_blocks.shape[0], -1) - # module = torch.nn.Linear(weight.shape[1], weight.shape[0], bias=False) - # with torch.no_grad(): - # if weight.shape != module.weight.shape: - # weight = weight.reshape(module.weight.shape) - # module.weight.copy_(weight) - # module.to(device=devices.cpu, dtype=devices.dtype) - # module.weight.requires_grad_(False) - # self.lin_module = module - #return module - def merge_weight(self, R_weight, org_weight): R_weight = R_weight.to(org_weight.device, dtype=org_weight.dtype) if org_weight.dim() == 4: weight = torch.einsum("oihw, op -> pihw", org_weight, R_weight) else: weight = torch.einsum("oi, op -> pi", org_weight, R_weight) - #weight = torch.einsum( - # "k n m, k n ... -> k m ...", - # self.oft_diag * scale + torch.eye(self.block_size, device=device), - # org_weight - #) return weight def get_weight(self, oft_blocks, multiplier=None): @@ -111,48 +83,51 @@ class NetworkModuleOFT(network.NetworkModule): block_R_weighted = multiplier * block_R + (1 - multiplier) * m_I R = torch.block_diag(*block_R_weighted) return R - #return self.oft_blocks + def calc_updown_kohya(self, orig_weight, multiplier): + R = self.get_weight(self.oft_blocks, multiplier) + merged_weight = self.merge_weight(R, orig_weight) - def calc_updown(self, orig_weight): - multiplier = self.multiplier() * self.calc_scale() - is_other_linear = type(self.sd_module) in [ torch.nn.MultiheadAttention] - if self.is_kohya and not is_other_linear: - R = self.get_weight(self.oft_blocks, multiplier) - #R = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) - merged_weight = self.merge_weight(R, orig_weight) - elif not self.is_kohya and not is_other_linear: + updown = merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight + output_shape = orig_weight.shape + orig_weight = orig_weight + return self.finalize_updown(updown, orig_weight, output_shape) + + def calc_updown_kb(self, orig_weight, multiplier): + is_other_linear = type(self.sd_module) in [torch.nn.MultiheadAttention] + + if not is_other_linear: if is_other_linear and orig_weight.shape[0] != orig_weight.shape[1]: orig_weight=orig_weight.permute(1, 0) + R = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) merged_weight = rearrange(orig_weight, '(k n) ... -> k n ...', k=self.num_blocks, n=self.block_size) - #orig_weight = rearrange(orig_weight, '(k n) ... -> k n ...', k=self.block_size, n=self.num_blocks) merged_weight = torch.einsum( 'k n m, k n ... -> k m ...', R * multiplier + torch.eye(self.block_size, device=orig_weight.device), - merged_weight + merged_weight ) merged_weight = rearrange(merged_weight, 'k m ... -> (k m) ...') + if is_other_linear and orig_weight.shape[0] != orig_weight.shape[1]: orig_weight=orig_weight.permute(1, 0) - #merged_weight=merged_weight.permute(1, 0) + updown = merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight - #updown = weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight output_shape = orig_weight.shape else: - # skip for now + # FIXME: skip MultiheadAttention for now updown = torch.zeros([orig_weight.shape[1], orig_weight.shape[1]], device=orig_weight.device, dtype=orig_weight.dtype) output_shape = (orig_weight.shape[1], orig_weight.shape[1]) - #if self.lin_module is not None: - # R = self.lin_module.weight.to(orig_weight.device, dtype=orig_weight.dtype) - # weight = torch.mul(torch.mul(R, multiplier), orig_weight) - #else: - - orig_weight = orig_weight - return self.finalize_updown(updown, orig_weight, output_shape) + def calc_updown(self, orig_weight): + multiplier = self.multiplier() * self.calc_scale() + if self.is_kohya: + return self.calc_updown_kohya(orig_weight, multiplier) + else: + return self.calc_updown_kb(orig_weight, multiplier) + # override to remove the multiplier/scale factor; it's already multiplied in get_weight def finalize_updown(self, updown, orig_weight, output_shape, ex_bias=None): #return super().finalize_updown(updown, orig_weight, output_shape, ex_bias) @@ -172,49 +147,3 @@ class NetworkModuleOFT(network.NetworkModule): ex_bias = ex_bias * self.multiplier() return updown, ex_bias - -# copied from https://github.com/KohakuBlueleaf/LyCORIS/blob/dev/lycoris/modules/lokr.py -def factorization(dimension: int, factor:int=-1) -> tuple[int, int]: - ''' - return a tuple of two value of input dimension decomposed by the number closest to factor - second value is higher or equal than first value. - - In LoRA with Kroneckor Product, first value is a value for weight scale. - secon value is a value for weight. - - Becuase of non-commutative property, A⊗B ≠ B⊗A. Meaning of two matrices is slightly different. - - examples) - factor - -1 2 4 8 16 ... - 127 -> 1, 127 127 -> 1, 127 127 -> 1, 127 127 -> 1, 127 127 -> 1, 127 - 128 -> 8, 16 128 -> 2, 64 128 -> 4, 32 128 -> 8, 16 128 -> 8, 16 - 250 -> 10, 25 250 -> 2, 125 250 -> 2, 125 250 -> 5, 50 250 -> 10, 25 - 360 -> 8, 45 360 -> 2, 180 360 -> 4, 90 360 -> 8, 45 360 -> 12, 30 - 512 -> 16, 32 512 -> 2, 256 512 -> 4, 128 512 -> 8, 64 512 -> 16, 32 - 1024 -> 32, 32 1024 -> 2, 512 1024 -> 4, 256 1024 -> 8, 128 1024 -> 16, 64 - ''' - - if factor > 0 and (dimension % factor) == 0: - m = factor - n = dimension // factor - if m > n: - n, m = m, n - return m, n - if factor < 0: - factor = dimension - m, n = 1, dimension - length = m + n - while m length or new_m>factor: - break - else: - m, n = new_m, new_n - if m > n: - n, m = m, n - return m, n - From 329c8bacce706811776e1c1c6a0d39b46886a268 Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Sat, 4 Nov 2023 14:54:36 -0700 Subject: [PATCH 051/342] refactor: use same updown for both kohya OFT and LyCORIS diag-oft --- extensions-builtin/Lora/network_oft.py | 91 +++++++++++++++++++++----- 1 file changed, 74 insertions(+), 17 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index 2be67fe53..e4aa082b7 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -2,6 +2,7 @@ import torch import network from lyco_helpers import factorization from einops import rearrange +from modules import devices class ModuleTypeOFT(network.ModuleType): @@ -24,12 +25,14 @@ class NetworkModuleOFT(network.NetworkModule): # kohya-ss if "oft_blocks" in weights.w.keys(): self.is_kohya = True - self.oft_blocks = weights.w["oft_blocks"] + self.oft_blocks = weights.w["oft_blocks"] # (num_blocks, block_size, block_size) self.alpha = weights.w["alpha"] - self.dim = self.oft_blocks.shape[0] + self.dim = self.oft_blocks.shape[0] # lora dim + #self.oft_blocks = rearrange(self.oft_blocks, 'k m ... -> (k m) ...') elif "oft_diag" in weights.w.keys(): self.is_kohya = False - self.oft_blocks = weights.w["oft_diag"] + self.oft_blocks = weights.w["oft_diag"] # (num_blocks, block_size, block_size) + # alpha is rank if alpha is 0 or None if self.alpha is None: pass @@ -51,12 +54,57 @@ class NetworkModuleOFT(network.NetworkModule): raise ValueError("sd_module must be Linear or Conv") if self.is_kohya: - self.num_blocks = self.dim - self.block_size = self.out_dim // self.num_blocks + #self.num_blocks = self.dim + #self.block_size = self.out_dim // self.num_blocks + #self.block_size = self.dim + #self.num_blocks = self.out_dim // self.block_size self.constraint = self.alpha * self.out_dim + self.num_blocks, self.block_size = factorization(self.out_dim, self.dim) else: - self.block_size, self.num_blocks = factorization(self.out_dim, self.dim) self.constraint = None + self.block_size, self.num_blocks = factorization(self.out_dim, self.dim) + + if is_other_linear: + self.lin_module = self.create_module(weights.w, "oft_diag", none_ok=True) + + + def create_module(self, weights, key, none_ok=False): + weight = weights.get(key) + + if weight is None and none_ok: + return None + + is_linear = type(self.sd_module) in [torch.nn.Linear, torch.nn.modules.linear.NonDynamicallyQuantizableLinear, torch.nn.MultiheadAttention] + is_conv = type(self.sd_module) in [torch.nn.Conv2d] + + if is_linear: + weight = weight.reshape(weight.shape[0], -1) + module = torch.nn.Linear(weight.shape[1], weight.shape[0], bias=False) + elif is_conv and key == "lora_down.weight" or key == "dyn_up": + if len(weight.shape) == 2: + weight = weight.reshape(weight.shape[0], -1, 1, 1) + + if weight.shape[2] != 1 or weight.shape[3] != 1: + module = torch.nn.Conv2d(weight.shape[1], weight.shape[0], self.sd_module.kernel_size, self.sd_module.stride, self.sd_module.padding, bias=False) + else: + module = torch.nn.Conv2d(weight.shape[1], weight.shape[0], (1, 1), bias=False) + elif is_conv and key == "lora_mid.weight": + module = torch.nn.Conv2d(weight.shape[1], weight.shape[0], self.sd_module.kernel_size, self.sd_module.stride, self.sd_module.padding, bias=False) + elif is_conv and key == "lora_up.weight" or key == "dyn_down": + module = torch.nn.Conv2d(weight.shape[1], weight.shape[0], (1, 1), bias=False) + else: + raise AssertionError(f'Lora layer {self.network_key} matched a layer with unsupported type: {type(self.sd_module).__name__}') + + with torch.no_grad(): + if weight.shape != module.weight.shape: + weight = weight.reshape(module.weight.shape) + module.weight.copy_(weight) + + module.to(device=devices.cpu, dtype=devices.dtype) + module.weight.requires_grad_(False) + + return module + def merge_weight(self, R_weight, org_weight): R_weight = R_weight.to(org_weight.device, dtype=org_weight.dtype) @@ -77,7 +125,8 @@ class NetworkModuleOFT(network.NetworkModule): else: new_norm_Q = norm_Q block_Q = block_Q * ((new_norm_Q + 1e-8) / (norm_Q + 1e-8)) - m_I = torch.eye(self.block_size, device=oft_blocks.device).unsqueeze(0).repeat(self.num_blocks, 1, 1) + m_I = torch.eye(self.num_blocks, device=oft_blocks.device).unsqueeze(0).repeat(self.block_size, 1, 1) + #m_I = torch.eye(self.block_size, device=oft_blocks.device).unsqueeze(0).repeat(self.num_blocks, 1, 1) block_R = torch.matmul(m_I + block_Q, (m_I - block_Q).inverse()) block_R_weighted = multiplier * block_R + (1 - multiplier) * m_I @@ -97,25 +146,33 @@ class NetworkModuleOFT(network.NetworkModule): is_other_linear = type(self.sd_module) in [torch.nn.MultiheadAttention] if not is_other_linear: - if is_other_linear and orig_weight.shape[0] != orig_weight.shape[1]: - orig_weight=orig_weight.permute(1, 0) + #if is_other_linear and orig_weight.shape[0] != orig_weight.shape[1]: + # orig_weight=orig_weight.permute(1, 0) + + oft_blocks = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) + + # without this line the results are significantly worse / less accurate + oft_blocks = oft_blocks - oft_blocks.transpose(1, 2) + + R = oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) + R = R * multiplier + torch.eye(self.block_size, device=orig_weight.device) - R = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) merged_weight = rearrange(orig_weight, '(k n) ... -> k n ...', k=self.num_blocks, n=self.block_size) merged_weight = torch.einsum( 'k n m, k n ... -> k m ...', - R * multiplier + torch.eye(self.block_size, device=orig_weight.device), + R, merged_weight ) merged_weight = rearrange(merged_weight, 'k m ... -> (k m) ...') - if is_other_linear and orig_weight.shape[0] != orig_weight.shape[1]: - orig_weight=orig_weight.permute(1, 0) + #if is_other_linear and orig_weight.shape[0] != orig_weight.shape[1]: + # orig_weight=orig_weight.permute(1, 0) updown = merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight output_shape = orig_weight.shape else: # FIXME: skip MultiheadAttention for now + #up = self.lin_module.weight.to(orig_weight.device, dtype=orig_weight.dtype) updown = torch.zeros([orig_weight.shape[1], orig_weight.shape[1]], device=orig_weight.device, dtype=orig_weight.dtype) output_shape = (orig_weight.shape[1], orig_weight.shape[1]) @@ -123,10 +180,10 @@ class NetworkModuleOFT(network.NetworkModule): def calc_updown(self, orig_weight): multiplier = self.multiplier() * self.calc_scale() - if self.is_kohya: - return self.calc_updown_kohya(orig_weight, multiplier) - else: - return self.calc_updown_kb(orig_weight, multiplier) + #if self.is_kohya: + # return self.calc_updown_kohya(orig_weight, multiplier) + #else: + return self.calc_updown_kb(orig_weight, multiplier) # override to remove the multiplier/scale factor; it's already multiplied in get_weight def finalize_updown(self, updown, orig_weight, output_shape, ex_bias=None): From bbf00a96afb2215f13cc72a7908225ae300c423d Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Sat, 4 Nov 2023 14:56:47 -0700 Subject: [PATCH 052/342] refactor: remove unused function --- extensions-builtin/Lora/network_oft.py | 47 -------------------------- 1 file changed, 47 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index e4aa082b7..93402bb28 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -2,7 +2,6 @@ import torch import network from lyco_helpers import factorization from einops import rearrange -from modules import devices class ModuleTypeOFT(network.ModuleType): @@ -54,58 +53,12 @@ class NetworkModuleOFT(network.NetworkModule): raise ValueError("sd_module must be Linear or Conv") if self.is_kohya: - #self.num_blocks = self.dim - #self.block_size = self.out_dim // self.num_blocks - #self.block_size = self.dim - #self.num_blocks = self.out_dim // self.block_size self.constraint = self.alpha * self.out_dim self.num_blocks, self.block_size = factorization(self.out_dim, self.dim) else: self.constraint = None self.block_size, self.num_blocks = factorization(self.out_dim, self.dim) - if is_other_linear: - self.lin_module = self.create_module(weights.w, "oft_diag", none_ok=True) - - - def create_module(self, weights, key, none_ok=False): - weight = weights.get(key) - - if weight is None and none_ok: - return None - - is_linear = type(self.sd_module) in [torch.nn.Linear, torch.nn.modules.linear.NonDynamicallyQuantizableLinear, torch.nn.MultiheadAttention] - is_conv = type(self.sd_module) in [torch.nn.Conv2d] - - if is_linear: - weight = weight.reshape(weight.shape[0], -1) - module = torch.nn.Linear(weight.shape[1], weight.shape[0], bias=False) - elif is_conv and key == "lora_down.weight" or key == "dyn_up": - if len(weight.shape) == 2: - weight = weight.reshape(weight.shape[0], -1, 1, 1) - - if weight.shape[2] != 1 or weight.shape[3] != 1: - module = torch.nn.Conv2d(weight.shape[1], weight.shape[0], self.sd_module.kernel_size, self.sd_module.stride, self.sd_module.padding, bias=False) - else: - module = torch.nn.Conv2d(weight.shape[1], weight.shape[0], (1, 1), bias=False) - elif is_conv and key == "lora_mid.weight": - module = torch.nn.Conv2d(weight.shape[1], weight.shape[0], self.sd_module.kernel_size, self.sd_module.stride, self.sd_module.padding, bias=False) - elif is_conv and key == "lora_up.weight" or key == "dyn_down": - module = torch.nn.Conv2d(weight.shape[1], weight.shape[0], (1, 1), bias=False) - else: - raise AssertionError(f'Lora layer {self.network_key} matched a layer with unsupported type: {type(self.sd_module).__name__}') - - with torch.no_grad(): - if weight.shape != module.weight.shape: - weight = weight.reshape(module.weight.shape) - module.weight.copy_(weight) - - module.to(device=devices.cpu, dtype=devices.dtype) - module.weight.requires_grad_(False) - - return module - - def merge_weight(self, R_weight, org_weight): R_weight = R_weight.to(org_weight.device, dtype=org_weight.dtype) if org_weight.dim() == 4: From 2b06cefe66684ed2648d3221efbc36aeaae99a2f Mon Sep 17 00:00:00 2001 From: gibiee <37574274+gibiee@users.noreply.github.com> Date: Sun, 5 Nov 2023 11:37:23 +0900 Subject: [PATCH 053/342] correct a typo modify "defaul" to "default" --- modules/cmd_args.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/cmd_args.py b/modules/cmd_args.py index 4e602a842..a9fb9bfa3 100644 --- a/modules/cmd_args.py +++ b/modules/cmd_args.py @@ -107,7 +107,7 @@ parser.add_argument("--tls-certfile", type=str, help="Partially enables TLS, req parser.add_argument("--disable-tls-verify", action="store_false", help="When passed, enables the use of self-signed certificates.", default=None) parser.add_argument("--server-name", type=str, help="Sets hostname of server", default=None) parser.add_argument("--gradio-queue", action='store_true', help="does not do anything", default=True) -parser.add_argument("--no-gradio-queue", action='store_true', help="Disables gradio queue; causes the webpage to use http requests instead of websockets; was the defaul in earlier versions") +parser.add_argument("--no-gradio-queue", action='store_true', help="Disables gradio queue; causes the webpage to use http requests instead of websockets; was the default in earlier versions") parser.add_argument("--skip-version-check", action='store_true', help="Do not check versions of torch and xformers") parser.add_argument("--no-hashing", action='store_true', help="disable sha256 hashing of checkpoints to help loading performance", default=False) parser.add_argument("--no-download-sd-model", action='store_true', help="don't download SD1.5 model even if no model is found in --ckpt-dir", default=False) From 6b8c661c49796bba093ca8a8301e81d28afb9832 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sun, 5 Nov 2023 08:55:54 +0300 Subject: [PATCH 054/342] add a visible checkbox to input accordion --- javascript/inputAccordion.js | 87 ++++++++++++++++++++++++------------ style.css | 5 +++ 2 files changed, 63 insertions(+), 29 deletions(-) diff --git a/javascript/inputAccordion.js b/javascript/inputAccordion.js index f2839852e..8fc01230a 100644 --- a/javascript/inputAccordion.js +++ b/javascript/inputAccordion.js @@ -1,37 +1,66 @@ -var observerAccordionOpen = new MutationObserver(function(mutations) { - mutations.forEach(function(mutationRecord) { - var elem = mutationRecord.target; - var open = elem.classList.contains('open'); - - var accordion = elem.parentNode; - accordion.classList.toggle('input-accordion-open', open); - - var checkbox = gradioApp().querySelector('#' + accordion.id + "-checkbox input"); - checkbox.checked = open; - updateInput(checkbox); - - var extra = gradioApp().querySelector('#' + accordion.id + "-extra"); - if (extra) { - extra.style.display = open ? "" : "none"; - } - }); -}); - function inputAccordionChecked(id, checked) { - var label = gradioApp().querySelector('#' + id + " .label-wrap"); - if (label.classList.contains('open') != checked) { - label.click(); + var accordion = gradioApp().getElementById(id); + accordion.visibleCheckbox.checked = checked; + accordion.onVisibleCheckboxChange(); +} + +function setupAccordion(accordion){ + var labelWrap = accordion.querySelector('.label-wrap'); + var gradioCheckbox = gradioApp().querySelector('#' + accordion.id + "-checkbox input"); + var extra = gradioApp().querySelector('#' + accordion.id + "-extra"); + var span = labelWrap.querySelector('span'); + var linked = true; + + var isOpen = function(){ return labelWrap.classList.contains('open'); } + + var observerAccordionOpen = new MutationObserver(function(mutations) { + mutations.forEach(function(mutationRecord) { + accordion.classList.toggle('input-accordion-open', isOpen()); + + if(linked){ + accordion.visibleCheckbox.checked = isOpen(); + accordion.onVisibleCheckboxChange(); + } + }); + }); + observerAccordionOpen.observe(labelWrap, {attributes: true, attributeFilter: ['class']}); + + if (extra) { + labelWrap.insertBefore(extra, labelWrap.lastElementChild); } + + accordion.onChecked = function(checked){ + if (isOpen() != checked) { + labelWrap.click(); + } + } + + var visibleCheckbox = document.createElement('INPUT'); + visibleCheckbox.type = 'checkbox'; + visibleCheckbox.checked = isOpen(); + visibleCheckbox.id = accordion.id + "-visible-checkbox"; + visibleCheckbox.className = gradioCheckbox.className + " input-accordion-checkbox"; + span.insertBefore(visibleCheckbox, span.firstChild); + + accordion.visibleCheckbox = visibleCheckbox; + accordion.onVisibleCheckboxChange = function(){ + if(linked && isOpen() != visibleCheckbox.checked) { + labelWrap.click(); + } + + gradioCheckbox.checked = visibleCheckbox.checked; + updateInput(gradioCheckbox); + }; + + visibleCheckbox.addEventListener('click', function(event){ + linked = false; + event.stopPropagation(); + }); + visibleCheckbox.addEventListener('input', accordion.onVisibleCheckboxChange); } onUiLoaded(function() { for (var accordion of gradioApp().querySelectorAll('.input-accordion')) { - var labelWrap = accordion.querySelector('.label-wrap'); - observerAccordionOpen.observe(labelWrap, {attributes: true, attributeFilter: ['class']}); - - var extra = gradioApp().querySelector('#' + accordion.id + "-extra"); - if (extra) { - labelWrap.insertBefore(extra, labelWrap.lastElementChild); - } + setupAccordion(accordion); } }); diff --git a/style.css b/style.css index 115626cd9..9a1181e8b 100644 --- a/style.css +++ b/style.css @@ -204,6 +204,11 @@ div.block.gradio-accordion { padding: 8px 8px; } +input[type="checkbox"].input-accordion-checkbox{ + vertical-align: sub; + margin-right: 0.5em; +} + /* txt2img/img2img specific */ From 16ab17429016a1154b9aa83244cdbfc7ba463d72 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sun, 5 Nov 2023 09:20:05 +0300 Subject: [PATCH 055/342] eslint --- javascript/inputAccordion.js | 20 +++++++++++--------- 1 file changed, 11 insertions(+), 9 deletions(-) diff --git a/javascript/inputAccordion.js b/javascript/inputAccordion.js index 8fc01230a..7570309aa 100644 --- a/javascript/inputAccordion.js +++ b/javascript/inputAccordion.js @@ -4,20 +4,22 @@ function inputAccordionChecked(id, checked) { accordion.onVisibleCheckboxChange(); } -function setupAccordion(accordion){ +function setupAccordion(accordion) { var labelWrap = accordion.querySelector('.label-wrap'); var gradioCheckbox = gradioApp().querySelector('#' + accordion.id + "-checkbox input"); var extra = gradioApp().querySelector('#' + accordion.id + "-extra"); var span = labelWrap.querySelector('span'); var linked = true; - var isOpen = function(){ return labelWrap.classList.contains('open'); } + var isOpen = function() { + return labelWrap.classList.contains('open'); + }; var observerAccordionOpen = new MutationObserver(function(mutations) { mutations.forEach(function(mutationRecord) { accordion.classList.toggle('input-accordion-open', isOpen()); - if(linked){ + if (linked) { accordion.visibleCheckbox.checked = isOpen(); accordion.onVisibleCheckboxChange(); } @@ -29,22 +31,22 @@ function setupAccordion(accordion){ labelWrap.insertBefore(extra, labelWrap.lastElementChild); } - accordion.onChecked = function(checked){ + accordion.onChecked = function(checked) { if (isOpen() != checked) { labelWrap.click(); } - } + }; var visibleCheckbox = document.createElement('INPUT'); visibleCheckbox.type = 'checkbox'; visibleCheckbox.checked = isOpen(); - visibleCheckbox.id = accordion.id + "-visible-checkbox"; + visibleCheckbox.id = accordion.id + "-visible-checkbox"; visibleCheckbox.className = gradioCheckbox.className + " input-accordion-checkbox"; span.insertBefore(visibleCheckbox, span.firstChild); accordion.visibleCheckbox = visibleCheckbox; - accordion.onVisibleCheckboxChange = function(){ - if(linked && isOpen() != visibleCheckbox.checked) { + accordion.onVisibleCheckboxChange = function() { + if (linked && isOpen() != visibleCheckbox.checked) { labelWrap.click(); } @@ -52,7 +54,7 @@ function setupAccordion(accordion){ updateInput(gradioCheckbox); }; - visibleCheckbox.addEventListener('click', function(event){ + visibleCheckbox.addEventListener('click', function(event) { linked = false; event.stopPropagation(); }); From d9499f4301018ebd2977685d098381aa4111d2ae Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sun, 5 Nov 2023 10:12:50 +0300 Subject: [PATCH 056/342] properly apply sort order for extra network cards when selected from dropdown allow selection of default sort order in settings remove 'Default' sort order, replace with 'Name' --- javascript/extraNetworks.js | 27 ++++++++++++++++----------- modules/shared_options.py | 2 ++ modules/ui_extra_networks.py | 6 ++++-- 3 files changed, 22 insertions(+), 13 deletions(-) diff --git a/javascript/extraNetworks.js b/javascript/extraNetworks.js index ac26718f6..a4d1d9d98 100644 --- a/javascript/extraNetworks.js +++ b/javascript/extraNetworks.js @@ -27,7 +27,6 @@ function setupExtraNetworksForTab(tabname) { var showDirsDiv = gradioApp().getElementById(tabname + '_extra_show_dirs'); var showDirs = gradioApp().querySelector('#' + tabname + '_extra_show_dirs input'); - sort.dataset.sortkey = 'sortDefault'; tabs.appendChild(searchDiv); tabs.appendChild(sort); tabs.appendChild(sortOrder); @@ -49,20 +48,23 @@ function setupExtraNetworksForTab(tabname) { elem.style.display = visible ? "" : "none"; }); + + applySort(); }; var applySort = function() { + var cards = gradioApp().querySelectorAll('#' + tabname + '_extra_tabs div.card'); + var reverse = sortOrder.classList.contains("sortReverse"); - var sortKey = sort.querySelector("input").value.toLowerCase().replace("sort", "").replaceAll(" ", "_").replace(/_+$/, "").trim(); - sortKey = sortKey ? "sort" + sortKey.charAt(0).toUpperCase() + sortKey.slice(1) : ""; - var sortKeyStore = sortKey ? sortKey + (reverse ? "Reverse" : "") : ""; - if (!sortKey || sortKeyStore == sort.dataset.sortkey) { + var sortKey = sort.querySelector("input").value.toLowerCase().replace("sort", "").replaceAll(" ", "_").replace(/_+$/, "").trim() || "name"; + sortKey = "sort" + sortKey.charAt(0).toUpperCase() + sortKey.slice(1); + var sortKeyStore = sortKey + "-" + (reverse ? "Descending" : "Ascending") + "-" + cards.length; + + if (sortKeyStore == sort.dataset.sortkey) { return; } - sort.dataset.sortkey = sortKeyStore; - var cards = gradioApp().querySelectorAll('#' + tabname + '_extra_tabs div.card'); cards.forEach(function(card) { card.originalParentElement = card.parentElement; }); @@ -88,15 +90,13 @@ function setupExtraNetworksForTab(tabname) { }; search.addEventListener("input", applyFilter); - applyFilter(); - ["change", "blur", "click"].forEach(function(evt) { - sort.querySelector("input").addEventListener(evt, applySort); - }); sortOrder.addEventListener("click", function() { sortOrder.classList.toggle("sortReverse"); applySort(); }); + applyFilter(); + extraNetworksApplySort[tabname] = applySort; extraNetworksApplyFilter[tabname] = applyFilter; var showDirsUpdate = function() { @@ -113,7 +113,12 @@ function applyExtraNetworkFilter(tabname) { setTimeout(extraNetworksApplyFilter[tabname], 1); } +function applyExtraNetworkSort(tabname) { + setTimeout(extraNetworksApplySort[tabname], 1); +} + var extraNetworksApplyFilter = {}; +var extraNetworksApplySort = {}; var activePromptTextarea = {}; function setupExtraNetworks() { diff --git a/modules/shared_options.py b/modules/shared_options.py index 0a82216ff..6543e440d 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -234,6 +234,8 @@ options_templates.update(options_section(('extra_networks', "Extra Networks"), { "extra_networks_card_height": OptionInfo(0, "Card height for Extra Networks").info("in pixels"), "extra_networks_card_text_scale": OptionInfo(1.0, "Card text scale", gr.Slider, {"minimum": 0.0, "maximum": 2.0, "step": 0.01}).info("1 = original size"), "extra_networks_card_show_desc": OptionInfo(True, "Show description on card"), + "extra_networks_card_order_field": OptionInfo("Name", "Default order field for Extra Networks cards", gr.Dropdown, {"choices": ['Name', 'Date Created', 'Date Modified']}).needs_reload_ui(), + "extra_networks_card_order": OptionInfo("Ascending", "Default order for Extra Networks cards", gr.Dropdown, {"choices": ['Ascending', 'Descending']}).needs_reload_ui(), "extra_networks_add_text_separator": OptionInfo(" ", "Extra networks separator").info("extra text to add before <...> when adding extra network to prompt"), "ui_extra_networks_tab_reorder": OptionInfo("", "Extra networks tab order").needs_reload_ui(), "textual_inversion_print_at_load": OptionInfo(False, "Print a list of Textual Inversion embeddings when loading model"), diff --git a/modules/ui_extra_networks.py b/modules/ui_extra_networks.py index 59d6ecc61..fc729917b 100644 --- a/modules/ui_extra_networks.py +++ b/modules/ui_extra_networks.py @@ -381,8 +381,8 @@ def create_ui(interface: gr.Blocks, unrelated_tabs, tabname): related_tabs.append(tab) edit_search = gr.Textbox('', show_label=False, elem_id=tabname+"_extra_search", elem_classes="search", placeholder="Search...", visible=False, interactive=True) - dropdown_sort = gr.Dropdown(choices=['Default Sort', 'Date Created', 'Date Modified', 'Name'], value='Default Sort', elem_id=tabname+"_extra_sort", elem_classes="sort", multiselect=False, visible=False, show_label=False, interactive=True, label=tabname+"_extra_sort_order") - button_sortorder = ToolButton(switch_values_symbol, elem_id=tabname+"_extra_sortorder", elem_classes="sortorder", visible=False, tooltip="Invert sort order") + dropdown_sort = gr.Dropdown(choices=['Name', 'Date Created', 'Date Modified', ], value=shared.opts.extra_networks_card_order_field, elem_id=tabname+"_extra_sort", elem_classes="sort", multiselect=False, visible=False, show_label=False, interactive=True, label=tabname+"_extra_sort_order") + button_sortorder = ToolButton(switch_values_symbol, elem_id=tabname+"_extra_sortorder", elem_classes=["sortorder"] + ([] if shared.opts.extra_networks_card_order == "Ascending" else ["sortReverse"]), visible=False, tooltip="Invert sort order") button_refresh = gr.Button('Refresh', elem_id=tabname+"_extra_refresh", visible=False) checkbox_show_dirs = gr.Checkbox(True, label='Show dirs', elem_id=tabname+"_extra_show_dirs", elem_classes="show-dirs", visible=False) @@ -395,6 +395,8 @@ def create_ui(interface: gr.Blocks, unrelated_tabs, tabname): for tab in related_tabs: tab.select(fn=lambda: [gr.update(visible=True) for _ in range(5)], inputs=[], outputs=[edit_search, dropdown_sort, button_sortorder, button_refresh, checkbox_show_dirs], show_progress=False) + dropdown_sort.change(fn=lambda: None, _js="function(){ applyExtraNetworkSort('" + tabname + "'); }") + def pages_html(): if not ui.pages_contents: return refresh() From ff1609f91ea0e9a90ba7b6ecc6d794c39c1f8c8f Mon Sep 17 00:00:00 2001 From: Ritesh Gangnani Date: Sun, 5 Nov 2023 19:13:49 +0530 Subject: [PATCH 057/342] Add SSD-1B as a supported model --- modules/sd_hijack.py | 11 +++++++++++ modules/sd_models.py | 8 ++++++-- modules/sd_models_types.py | 5 ++++- 3 files changed, 21 insertions(+), 3 deletions(-) diff --git a/modules/sd_hijack.py b/modules/sd_hijack.py index 592f00551..d19f853ec 100644 --- a/modules/sd_hijack.py +++ b/modules/sd_hijack.py @@ -180,6 +180,17 @@ class StableDiffusionModelHijack: except Exception as e: errors.display(e, "applying cross attention optimization") undo_optimizations() + + def conv_ssd(self, m): + delattr(m.model.diffusion_model.middle_block, '1') + delattr(m.model.diffusion_model.middle_block, '2') + for i in ['9','8','7','6','5','4']: + delattr(m.model.diffusion_model.input_blocks[7][1].transformer_blocks,i) + delattr(m.model.diffusion_model.input_blocks[8][1].transformer_blocks,i) + delattr(m.model.diffusion_model.output_blocks[0][1].transformer_blocks,i) + delattr(m.model.diffusion_model.output_blocks[1][1].transformer_blocks,i) + delattr(m.model.diffusion_model.output_blocks[4][1].transformer_blocks,'1') + delattr(m.model.diffusion_model.output_blocks[5][1].transformer_blocks,'1') def hijack(self, m): conditioner = getattr(m, 'conditioner', None) diff --git a/modules/sd_models.py b/modules/sd_models.py index 930d0bee5..ef96d29db 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -346,10 +346,14 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer model.is_sdxl = hasattr(model, 'conditioner') model.is_sd2 = not model.is_sdxl and hasattr(model.cond_stage_model, 'model') model.is_sd1 = not model.is_sdxl and not model.is_sd2 - + model.is_ssd = model.is_sdxl and 'model.diffusion_model.middle_block.1.transformer_blocks.0.attn1.to_q.weight' not in state_dict.keys() + if model.is_sdxl: sd_models_xl.extend_sdxl(model) - + + if model.is_ssd: + sd_hijack.model_hijack.conv_ssd(model) + model.load_state_dict(state_dict, strict=False) timer.record("apply weights to model") diff --git a/modules/sd_models_types.py b/modules/sd_models_types.py index 5ffd2f4f9..1f28942a4 100644 --- a/modules/sd_models_types.py +++ b/modules/sd_models_types.py @@ -22,7 +22,10 @@ class WebuiSdModel(LatentDiffusion): """structure with additional information about the file with model's weights""" is_sdxl: bool - """True if the model's architecture is SDXL""" + """True if the model's architecture is SDXL or SSD""" + + is_ssd: bool + """True if the model is SSD""" is_sd2: bool """True if the model's architecture is SD 2.x""" From 44db35fb1ad5d07837e890a0fd3c00addfb0402c Mon Sep 17 00:00:00 2001 From: Ritesh Gangnani Date: Sun, 5 Nov 2023 19:15:38 +0530 Subject: [PATCH 058/342] Added memory clearance after deletion --- modules/sd_hijack.py | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/modules/sd_hijack.py b/modules/sd_hijack.py index d19f853ec..059ffe8f0 100644 --- a/modules/sd_hijack.py +++ b/modules/sd_hijack.py @@ -1,3 +1,5 @@ +import gc + import torch from torch.nn.functional import silu from types import MethodType @@ -190,7 +192,9 @@ class StableDiffusionModelHijack: delattr(m.model.diffusion_model.output_blocks[0][1].transformer_blocks,i) delattr(m.model.diffusion_model.output_blocks[1][1].transformer_blocks,i) delattr(m.model.diffusion_model.output_blocks[4][1].transformer_blocks,'1') - delattr(m.model.diffusion_model.output_blocks[5][1].transformer_blocks,'1') + delattr(m.model.diffusion_model.output_blocks[5][1].transformer_blocks,'1') + torch.cuda.empty_cache() + gc.collect() def hijack(self, m): conditioner = getattr(m, 'conditioner', None) From 44c5097375ae4cf40300c09473bb46cf6c5d6cb7 Mon Sep 17 00:00:00 2001 From: Ritesh Gangnani Date: Sun, 5 Nov 2023 20:31:57 +0530 Subject: [PATCH 059/342] Use devices.torch_gc() instead of empty_cache() --- modules/sd_hijack.py | 5 +---- modules/sd_models.py | 1 - 2 files changed, 1 insertion(+), 5 deletions(-) diff --git a/modules/sd_hijack.py b/modules/sd_hijack.py index 059ffe8f0..0a7e5135d 100644 --- a/modules/sd_hijack.py +++ b/modules/sd_hijack.py @@ -1,5 +1,3 @@ -import gc - import torch from torch.nn.functional import silu from types import MethodType @@ -193,8 +191,7 @@ class StableDiffusionModelHijack: delattr(m.model.diffusion_model.output_blocks[1][1].transformer_blocks,i) delattr(m.model.diffusion_model.output_blocks[4][1].transformer_blocks,'1') delattr(m.model.diffusion_model.output_blocks[5][1].transformer_blocks,'1') - torch.cuda.empty_cache() - gc.collect() + devices.torch_gc() def hijack(self, m): conditioner = getattr(m, 'conditioner', None) diff --git a/modules/sd_models.py b/modules/sd_models.py index ef96d29db..2242c3637 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -347,7 +347,6 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer model.is_sd2 = not model.is_sdxl and hasattr(model.cond_stage_model, 'model') model.is_sd1 = not model.is_sdxl and not model.is_sd2 model.is_ssd = model.is_sdxl and 'model.diffusion_model.middle_block.1.transformer_blocks.0.attn1.to_q.weight' not in state_dict.keys() - if model.is_sdxl: sd_models_xl.extend_sdxl(model) From 4d4a9e733219f8c065a4ab6c5ab42836db7330fe Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sun, 5 Nov 2023 19:19:55 +0300 Subject: [PATCH 060/342] added compact prompt option --- .../mobile/javascript/mobile.js | 2 + javascript/extraNetworks.js | 33 +++ modules/shared_items.py | 2 + modules/shared_options.py | 1 + modules/ui.py | 247 +++++++----------- modules/ui_common.py | 15 +- modules/ui_extra_networks.py | 18 +- modules/ui_extra_networks_checkpoints.py | 2 + modules/ui_toprow.py | 141 ++++++++++ style.css | 23 +- 10 files changed, 315 insertions(+), 169 deletions(-) create mode 100644 modules/ui_toprow.py diff --git a/extensions-builtin/mobile/javascript/mobile.js b/extensions-builtin/mobile/javascript/mobile.js index 652f07ac7..bff1acedf 100644 --- a/extensions-builtin/mobile/javascript/mobile.js +++ b/extensions-builtin/mobile/javascript/mobile.js @@ -12,6 +12,8 @@ function isMobile() { } function reportWindowSize() { + if (gradioApp().querySelector('.toprow-compact-tools')) return; // not applicable for compact prompt layout + var currentlyMobile = isMobile(); if (currentlyMobile == isSetupForMobile) return; isSetupForMobile = currentlyMobile; diff --git a/javascript/extraNetworks.js b/javascript/extraNetworks.js index a4d1d9d98..a1bf29a8c 100644 --- a/javascript/extraNetworks.js +++ b/javascript/extraNetworks.js @@ -26,6 +26,8 @@ function setupExtraNetworksForTab(tabname) { var refresh = gradioApp().getElementById(tabname + '_extra_refresh'); var showDirsDiv = gradioApp().getElementById(tabname + '_extra_show_dirs'); var showDirs = gradioApp().querySelector('#' + tabname + '_extra_show_dirs input'); + var promptContainer = gradioApp().querySelector('.prompt-container-compact#' + tabname + '_prompt_container'); + var negativePrompt = gradioApp().querySelector('#' + tabname + '_neg_prompt'); tabs.appendChild(searchDiv); tabs.appendChild(sort); @@ -109,6 +111,37 @@ function setupExtraNetworksForTab(tabname) { showDirsUpdate(); } +function extraNetworksMovePromptToTab(tabname, id, showPrompt, showNegativePrompt) { + if (!gradioApp().querySelector('.toprow-compact-tools')) return; // only applicable for compact prompt layout + + var promptContainer = gradioApp().getElementById(tabname + '_prompt_container'); + var prompt = gradioApp().getElementById(tabname + '_prompt_row'); + var negPrompt = gradioApp().getElementById(tabname + '_neg_prompt_row'); + var elem = id ? gradioApp().getElementById(id) : null; + + if (showNegativePrompt && elem) { + elem.insertBefore(negPrompt, elem.firstChild); + } else { + promptContainer.insertBefore(negPrompt, promptContainer.firstChild); + } + + if (showPrompt && elem) { + elem.insertBefore(prompt, elem.firstChild); + } else { + promptContainer.insertBefore(prompt, promptContainer.firstChild); + } +} + + +function extraNetworksUrelatedTabSelected(tabname) { // called from python when user selects an unrelated tab (generate) + extraNetworksMovePromptToTab(tabname, '', false, false); +} + +function extraNetworksTabSelected(tabname, id, showPrompt, showNegativePrompt) { // called from python when user selects an extra networks tab + extraNetworksMovePromptToTab(tabname, id, showPrompt, showNegativePrompt); + +} + function applyExtraNetworkFilter(tabname) { setTimeout(extraNetworksApplyFilter[tabname], 1); } diff --git a/modules/shared_items.py b/modules/shared_items.py index b1459f8c4..5024b4268 100644 --- a/modules/shared_items.py +++ b/modules/shared_items.py @@ -67,6 +67,8 @@ def reload_hypernetworks(): ui_reorder_categories_builtin_items = [ + "prompt", + "image", "inpaint", "sampler", "accordions", diff --git a/modules/shared_options.py b/modules/shared_options.py index 6543e440d..4e3d75416 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -272,6 +272,7 @@ options_templates.update(options_section(('ui', "User interface"), { "hires_fix_show_sampler": OptionInfo(False, "Hires fix: show hires checkpoint and sampler selection").needs_reload_ui(), "hires_fix_show_prompts": OptionInfo(False, "Hires fix: show hires prompt and negative prompt").needs_reload_ui(), "disable_token_counters": OptionInfo(False, "Disable prompt token counters").needs_reload_ui(), + "compact_prompt_box": OptionInfo(True, "Compact prompt layout").info("puts prompt and negative prompt inside the Generate tab, leaving more vertical space for the image on the right").needs_reload_ui(), })) diff --git a/modules/ui.py b/modules/ui.py index bcf391997..2454eb36b 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -12,7 +12,7 @@ from PIL import Image, PngImagePlugin # noqa: F401 from modules.call_queue import wrap_gradio_gpu_call, wrap_queued_call, wrap_gradio_call from modules import gradio_extensons # noqa: F401 -from modules import sd_hijack, sd_models, script_callbacks, ui_extensions, deepbooru, extra_networks, ui_common, ui_postprocessing, progress, ui_loadsave, shared_items, ui_settings, timer, sysinfo, ui_checkpoint_merger, ui_prompt_styles, scripts, sd_samplers, processing, ui_extra_networks +from modules import sd_hijack, sd_models, script_callbacks, ui_extensions, deepbooru, extra_networks, ui_common, ui_postprocessing, progress, ui_loadsave, shared_items, ui_settings, timer, sysinfo, ui_checkpoint_merger, scripts, sd_samplers, processing, ui_extra_networks, ui_toprow from modules.ui_components import FormRow, FormGroup, ToolButton, FormHTML, InputAccordion, ResizeHandleRow from modules.paths import script_path from modules.ui_common import create_refresh_button @@ -25,7 +25,6 @@ import modules.hypernetworks.ui as hypernetworks_ui import modules.textual_inversion.ui as textual_inversion_ui import modules.textual_inversion.textual_inversion as textual_inversion import modules.shared as shared -import modules.images from modules import prompt_parser from modules.sd_hijack import model_hijack from modules.generation_parameters_copypaste import image_from_url_text @@ -177,79 +176,6 @@ def update_negative_prompt_token_counter(text, steps): return update_token_counter(text, steps, is_positive=False) -class Toprow: - """Creates a top row UI with prompts, generate button, styles, extra little buttons for things, and enables some functionality related to their operation""" - - def __init__(self, is_img2img): - id_part = "img2img" if is_img2img else "txt2img" - self.id_part = id_part - - with gr.Row(elem_id=f"{id_part}_toprow", variant="compact"): - with gr.Column(elem_id=f"{id_part}_prompt_container", scale=6): - with gr.Row(): - with gr.Column(scale=80): - with gr.Row(): - self.prompt = gr.Textbox(label="Prompt", elem_id=f"{id_part}_prompt", show_label=False, lines=3, placeholder="Prompt (press Ctrl+Enter or Alt+Enter to generate)", elem_classes=["prompt"]) - self.prompt_img = gr.File(label="", elem_id=f"{id_part}_prompt_image", file_count="single", type="binary", visible=False) - - with gr.Row(): - with gr.Column(scale=80): - with gr.Row(): - self.negative_prompt = gr.Textbox(label="Negative prompt", elem_id=f"{id_part}_neg_prompt", show_label=False, lines=3, placeholder="Negative prompt (press Ctrl+Enter or Alt+Enter to generate)", elem_classes=["prompt"]) - - self.button_interrogate = None - self.button_deepbooru = None - if is_img2img: - with gr.Column(scale=1, elem_classes="interrogate-col"): - self.button_interrogate = gr.Button('Interrogate\nCLIP', elem_id="interrogate") - self.button_deepbooru = gr.Button('Interrogate\nDeepBooru', elem_id="deepbooru") - - with gr.Column(scale=1, elem_id=f"{id_part}_actions_column"): - with gr.Row(elem_id=f"{id_part}_generate_box", elem_classes="generate-box"): - self.interrupt = gr.Button('Interrupt', elem_id=f"{id_part}_interrupt", elem_classes="generate-box-interrupt") - self.skip = gr.Button('Skip', elem_id=f"{id_part}_skip", elem_classes="generate-box-skip") - self.submit = gr.Button('Generate', elem_id=f"{id_part}_generate", variant='primary') - - self.skip.click( - fn=lambda: shared.state.skip(), - inputs=[], - outputs=[], - ) - - self.interrupt.click( - fn=lambda: shared.state.interrupt(), - inputs=[], - outputs=[], - ) - - with gr.Row(elem_id=f"{id_part}_tools"): - self.paste = ToolButton(value=paste_symbol, elem_id="paste", tooltip="Read generation parameters from prompt or last generation if prompt is empty into user interface.") - self.clear_prompt_button = ToolButton(value=clear_prompt_symbol, elem_id=f"{id_part}_clear_prompt", tooltip="Clear prompt") - self.apply_styles = ToolButton(value=ui_prompt_styles.styles_materialize_symbol, elem_id=f"{id_part}_style_apply", tooltip="Apply all selected styles to prompts.") - self.restore_progress_button = ToolButton(value=restore_progress_symbol, elem_id=f"{id_part}_restore_progress", visible=False, tooltip="Restore progress") - - self.token_counter = gr.HTML(value="0/75", elem_id=f"{id_part}_token_counter", elem_classes=["token-counter"]) - self.token_button = gr.Button(visible=False, elem_id=f"{id_part}_token_button") - self.negative_token_counter = gr.HTML(value="0/75", elem_id=f"{id_part}_negative_token_counter", elem_classes=["token-counter"]) - self.negative_token_button = gr.Button(visible=False, elem_id=f"{id_part}_negative_token_button") - - self.clear_prompt_button.click( - fn=lambda *x: x, - _js="confirm_clear_prompt", - inputs=[self.prompt, self.negative_prompt], - outputs=[self.prompt, self.negative_prompt], - ) - - self.ui_styles = ui_prompt_styles.UiPromptStyles(id_part, self.prompt, self.negative_prompt) - self.ui_styles.setup_apply_button(self.apply_styles) - - self.prompt_img.change( - fn=modules.images.image_data, - inputs=[self.prompt_img], - outputs=[self.prompt, self.prompt_img], - show_progress=False, - ) - def setup_progressbar(*args, **kwargs): pass @@ -288,8 +214,8 @@ def apply_setting(key, value): return getattr(opts, key) -def create_output_panel(tabname, outdir): - return ui_common.create_output_panel(tabname, outdir) +def create_output_panel(tabname, outdir, toprow=None): + return ui_common.create_output_panel(tabname, outdir, toprow) def create_sampler_and_steps_selection(choices, tabname): @@ -336,7 +262,7 @@ def create_ui(): scripts.scripts_txt2img.initialize_scripts(is_img2img=False) with gr.Blocks(analytics_enabled=False) as txt2img_interface: - toprow = Toprow(is_img2img=False) + toprow = ui_toprow.Toprow(is_img2img=False, is_compact=shared.opts.compact_prompt_box) dummy_component = gr.Label(visible=False) @@ -348,6 +274,9 @@ def create_ui(): scripts.scripts_txt2img.prepare_ui() for category in ordered_ui_categories(): + if category == "prompt": + toprow.create_inline_toprow_prompts() + if category == "sampler": steps, sampler_name = create_sampler_and_steps_selection(sd_samplers.visible_sampler_names(), "txt2img") @@ -442,7 +371,7 @@ def create_ui(): show_progress=False, ) - txt2img_gallery, generation_info, html_info, html_log = create_output_panel("txt2img", opts.outdir_txt2img_samples) + txt2img_gallery, generation_info, html_info, html_log = create_output_panel("txt2img", opts.outdir_txt2img_samples, toprow) txt2img_args = dict( fn=wrap_gradio_gpu_call(modules.txt2img.txt2img, extra_outputs=[None, '', '']), @@ -554,7 +483,7 @@ def create_ui(): scripts.scripts_img2img.initialize_scripts(is_img2img=True) with gr.Blocks(analytics_enabled=False) as img2img_interface: - toprow = Toprow(is_img2img=True) + toprow = ui_toprow.Toprow(is_img2img=True, is_compact=shared.opts.compact_prompt_box) extra_tabs = gr.Tabs(elem_id="img2img_extra_tabs") extra_tabs.__enter__() @@ -577,85 +506,89 @@ def create_ui(): button = gr.Button(title) copy_image_buttons.append((button, name, elem)) - with gr.Tabs(elem_id="mode_img2img"): - img2img_selected_tab = gr.State(0) - - with gr.TabItem('img2img', id='img2img', elem_id="img2img_img2img_tab") as tab_img2img: - init_img = gr.Image(label="Image for img2img", elem_id="img2img_image", show_label=False, source="upload", interactive=True, type="pil", tool="editor", image_mode="RGBA", height=opts.img2img_editor_height) - add_copy_image_controls('img2img', init_img) - - with gr.TabItem('Sketch', id='img2img_sketch', elem_id="img2img_img2img_sketch_tab") as tab_sketch: - sketch = gr.Image(label="Image for img2img", elem_id="img2img_sketch", show_label=False, source="upload", interactive=True, type="pil", tool="color-sketch", image_mode="RGB", height=opts.img2img_editor_height, brush_color=opts.img2img_sketch_default_brush_color) - add_copy_image_controls('sketch', sketch) - - with gr.TabItem('Inpaint', id='inpaint', elem_id="img2img_inpaint_tab") as tab_inpaint: - init_img_with_mask = gr.Image(label="Image for inpainting with mask", show_label=False, elem_id="img2maskimg", source="upload", interactive=True, type="pil", tool="sketch", image_mode="RGBA", height=opts.img2img_editor_height, brush_color=opts.img2img_inpaint_mask_brush_color) - add_copy_image_controls('inpaint', init_img_with_mask) - - with gr.TabItem('Inpaint sketch', id='inpaint_sketch', elem_id="img2img_inpaint_sketch_tab") as tab_inpaint_color: - inpaint_color_sketch = gr.Image(label="Color sketch inpainting", show_label=False, elem_id="inpaint_sketch", source="upload", interactive=True, type="pil", tool="color-sketch", image_mode="RGB", height=opts.img2img_editor_height, brush_color=opts.img2img_inpaint_sketch_default_brush_color) - inpaint_color_sketch_orig = gr.State(None) - add_copy_image_controls('inpaint_sketch', inpaint_color_sketch) - - def update_orig(image, state): - if image is not None: - same_size = state is not None and state.size == image.size - has_exact_match = np.any(np.all(np.array(image) == np.array(state), axis=-1)) - edited = same_size and has_exact_match - return image if not edited or state is None else state - - inpaint_color_sketch.change(update_orig, [inpaint_color_sketch, inpaint_color_sketch_orig], inpaint_color_sketch_orig) - - with gr.TabItem('Inpaint upload', id='inpaint_upload', elem_id="img2img_inpaint_upload_tab") as tab_inpaint_upload: - init_img_inpaint = gr.Image(label="Image for img2img", show_label=False, source="upload", interactive=True, type="pil", elem_id="img_inpaint_base") - init_mask_inpaint = gr.Image(label="Mask", source="upload", interactive=True, type="pil", image_mode="RGBA", elem_id="img_inpaint_mask") - - with gr.TabItem('Batch', id='batch', elem_id="img2img_batch_tab") as tab_batch: - hidden = '
Disabled when launched with --hide-ui-dir-config.' if shared.cmd_opts.hide_ui_dir_config else '' - gr.HTML( - "

Process images in a directory on the same machine where the server is running." + - "
Use an empty output directory to save pictures normally instead of writing to the output directory." + - f"
Add inpaint batch mask directory to enable inpaint batch processing." - f"{hidden}

" - ) - img2img_batch_input_dir = gr.Textbox(label="Input directory", **shared.hide_dirs, elem_id="img2img_batch_input_dir") - img2img_batch_output_dir = gr.Textbox(label="Output directory", **shared.hide_dirs, elem_id="img2img_batch_output_dir") - img2img_batch_inpaint_mask_dir = gr.Textbox(label="Inpaint batch mask directory (required for inpaint batch processing only)", **shared.hide_dirs, elem_id="img2img_batch_inpaint_mask_dir") - with gr.Accordion("PNG info", open=False): - img2img_batch_use_png_info = gr.Checkbox(label="Append png info to prompts", **shared.hide_dirs, elem_id="img2img_batch_use_png_info") - img2img_batch_png_info_dir = gr.Textbox(label="PNG info directory", **shared.hide_dirs, placeholder="Leave empty to use input directory", elem_id="img2img_batch_png_info_dir") - img2img_batch_png_info_props = gr.CheckboxGroup(["Prompt", "Negative prompt", "Seed", "CFG scale", "Sampler", "Steps", "Model hash"], label="Parameters to take from png info", info="Prompts from png info will be appended to prompts set in ui.") - - img2img_tabs = [tab_img2img, tab_sketch, tab_inpaint, tab_inpaint_color, tab_inpaint_upload, tab_batch] - - for i, tab in enumerate(img2img_tabs): - tab.select(fn=lambda tabnum=i: tabnum, inputs=[], outputs=[img2img_selected_tab]) - - def copy_image(img): - if isinstance(img, dict) and 'image' in img: - return img['image'] - - return img - - for button, name, elem in copy_image_buttons: - button.click( - fn=copy_image, - inputs=[elem], - outputs=[copy_image_destinations[name]], - ) - button.click( - fn=lambda: None, - _js=f"switch_to_{name.replace(' ', '_')}", - inputs=[], - outputs=[], - ) - - with FormRow(): - resize_mode = gr.Radio(label="Resize mode", elem_id="resize_mode", choices=["Just resize", "Crop and resize", "Resize and fill", "Just resize (latent upscale)"], type="index", value="Just resize") - scripts.scripts_img2img.prepare_ui() for category in ordered_ui_categories(): + if category == "prompt": + toprow.create_inline_toprow_prompts() + + if category == "image": + with gr.Tabs(elem_id="mode_img2img"): + img2img_selected_tab = gr.State(0) + + with gr.TabItem('img2img', id='img2img', elem_id="img2img_img2img_tab") as tab_img2img: + init_img = gr.Image(label="Image for img2img", elem_id="img2img_image", show_label=False, source="upload", interactive=True, type="pil", tool="editor", image_mode="RGBA", height=opts.img2img_editor_height) + add_copy_image_controls('img2img', init_img) + + with gr.TabItem('Sketch', id='img2img_sketch', elem_id="img2img_img2img_sketch_tab") as tab_sketch: + sketch = gr.Image(label="Image for img2img", elem_id="img2img_sketch", show_label=False, source="upload", interactive=True, type="pil", tool="color-sketch", image_mode="RGB", height=opts.img2img_editor_height, brush_color=opts.img2img_sketch_default_brush_color) + add_copy_image_controls('sketch', sketch) + + with gr.TabItem('Inpaint', id='inpaint', elem_id="img2img_inpaint_tab") as tab_inpaint: + init_img_with_mask = gr.Image(label="Image for inpainting with mask", show_label=False, elem_id="img2maskimg", source="upload", interactive=True, type="pil", tool="sketch", image_mode="RGBA", height=opts.img2img_editor_height, brush_color=opts.img2img_inpaint_mask_brush_color) + add_copy_image_controls('inpaint', init_img_with_mask) + + with gr.TabItem('Inpaint sketch', id='inpaint_sketch', elem_id="img2img_inpaint_sketch_tab") as tab_inpaint_color: + inpaint_color_sketch = gr.Image(label="Color sketch inpainting", show_label=False, elem_id="inpaint_sketch", source="upload", interactive=True, type="pil", tool="color-sketch", image_mode="RGB", height=opts.img2img_editor_height, brush_color=opts.img2img_inpaint_sketch_default_brush_color) + inpaint_color_sketch_orig = gr.State(None) + add_copy_image_controls('inpaint_sketch', inpaint_color_sketch) + + def update_orig(image, state): + if image is not None: + same_size = state is not None and state.size == image.size + has_exact_match = np.any(np.all(np.array(image) == np.array(state), axis=-1)) + edited = same_size and has_exact_match + return image if not edited or state is None else state + + inpaint_color_sketch.change(update_orig, [inpaint_color_sketch, inpaint_color_sketch_orig], inpaint_color_sketch_orig) + + with gr.TabItem('Inpaint upload', id='inpaint_upload', elem_id="img2img_inpaint_upload_tab") as tab_inpaint_upload: + init_img_inpaint = gr.Image(label="Image for img2img", show_label=False, source="upload", interactive=True, type="pil", elem_id="img_inpaint_base") + init_mask_inpaint = gr.Image(label="Mask", source="upload", interactive=True, type="pil", image_mode="RGBA", elem_id="img_inpaint_mask") + + with gr.TabItem('Batch', id='batch', elem_id="img2img_batch_tab") as tab_batch: + hidden = '
Disabled when launched with --hide-ui-dir-config.' if shared.cmd_opts.hide_ui_dir_config else '' + gr.HTML( + "

Process images in a directory on the same machine where the server is running." + + "
Use an empty output directory to save pictures normally instead of writing to the output directory." + + f"
Add inpaint batch mask directory to enable inpaint batch processing." + f"{hidden}

" + ) + img2img_batch_input_dir = gr.Textbox(label="Input directory", **shared.hide_dirs, elem_id="img2img_batch_input_dir") + img2img_batch_output_dir = gr.Textbox(label="Output directory", **shared.hide_dirs, elem_id="img2img_batch_output_dir") + img2img_batch_inpaint_mask_dir = gr.Textbox(label="Inpaint batch mask directory (required for inpaint batch processing only)", **shared.hide_dirs, elem_id="img2img_batch_inpaint_mask_dir") + with gr.Accordion("PNG info", open=False): + img2img_batch_use_png_info = gr.Checkbox(label="Append png info to prompts", **shared.hide_dirs, elem_id="img2img_batch_use_png_info") + img2img_batch_png_info_dir = gr.Textbox(label="PNG info directory", **shared.hide_dirs, placeholder="Leave empty to use input directory", elem_id="img2img_batch_png_info_dir") + img2img_batch_png_info_props = gr.CheckboxGroup(["Prompt", "Negative prompt", "Seed", "CFG scale", "Sampler", "Steps", "Model hash"], label="Parameters to take from png info", info="Prompts from png info will be appended to prompts set in ui.") + + img2img_tabs = [tab_img2img, tab_sketch, tab_inpaint, tab_inpaint_color, tab_inpaint_upload, tab_batch] + + for i, tab in enumerate(img2img_tabs): + tab.select(fn=lambda tabnum=i: tabnum, inputs=[], outputs=[img2img_selected_tab]) + + def copy_image(img): + if isinstance(img, dict) and 'image' in img: + return img['image'] + + return img + + for button, name, elem in copy_image_buttons: + button.click( + fn=copy_image, + inputs=[elem], + outputs=[copy_image_destinations[name]], + ) + button.click( + fn=lambda: None, + _js=f"switch_to_{name.replace(' ', '_')}", + inputs=[], + outputs=[], + ) + + with FormRow(): + resize_mode = gr.Radio(label="Resize mode", elem_id="resize_mode", choices=["Just resize", "Crop and resize", "Resize and fill", "Just resize (latent upscale)"], type="index", value="Just resize") + if category == "sampler": steps, sampler_name = create_sampler_and_steps_selection(sd_samplers.visible_sampler_names(), "img2img") @@ -769,7 +702,7 @@ def create_ui(): if category not in {"accordions"}: scripts.scripts_img2img.setup_ui_for_section(category) - img2img_gallery, generation_info, html_info, html_log = create_output_panel("img2img", opts.outdir_img2img_samples) + img2img_gallery, generation_info, html_info, html_log = create_output_panel("img2img", opts.outdir_img2img_samples, toprow) img2img_args = dict( fn=wrap_gradio_gpu_call(modules.img2img.img2img, extra_outputs=[None, '', '']), diff --git a/modules/ui_common.py b/modules/ui_common.py index 84a7d7f27..032ec4af7 100644 --- a/modules/ui_common.py +++ b/modules/ui_common.py @@ -104,7 +104,7 @@ def save_files(js_data, images, do_make_zip, index): return gr.File.update(value=fullfns, visible=True), plaintext_to_html(f"Saved: {filenames[0]}") -def create_output_panel(tabname, outdir): +def create_output_panel(tabname, outdir, toprow=None): def open_folder(f): if not os.path.exists(f): @@ -130,12 +130,15 @@ Requested path was: {f} else: sp.Popen(["xdg-open", path]) - with gr.Column(variant='panel', elem_id=f"{tabname}_results"): - with gr.Group(elem_id=f"{tabname}_gallery_container"): - result_gallery = gr.Gallery(label='Output', show_label=False, elem_id=f"{tabname}_gallery", columns=4, preview=True, height=shared.opts.gallery_height or None) + with gr.Column(elem_id=f"{tabname}_results"): + if toprow: + toprow.create_inline_toprow_image() - generation_info = None - with gr.Column(): + with gr.Column(variant='panel', elem_id=f"{tabname}_results_panel"): + with gr.Group(elem_id=f"{tabname}_gallery_container"): + result_gallery = gr.Gallery(label='Output', show_label=False, elem_id=f"{tabname}_gallery", columns=4, preview=True, height=shared.opts.gallery_height or None) + + generation_info = None with gr.Row(elem_id=f"image_buttons_{tabname}", elem_classes="image-buttons"): open_folder_button = ToolButton(folder_symbol, elem_id=f'{tabname}_open_folder', visible=not shared.cmd_opts.hide_ui_dir_config, tooltip="Open images output directory.") diff --git a/modules/ui_extra_networks.py b/modules/ui_extra_networks.py index fc729917b..7907cd63f 100644 --- a/modules/ui_extra_networks.py +++ b/modules/ui_extra_networks.py @@ -103,6 +103,7 @@ class ExtraNetworksPage: self.name = title.lower() self.id_page = self.name.replace(" ", "_") self.card_page = shared.html("extra-networks-card.html") + self.allow_prompt = True self.allow_negative_prompt = False self.metadata = {} self.items = {} @@ -367,7 +368,7 @@ def create_ui(interface: gr.Blocks, unrelated_tabs, tabname): related_tabs = [] for page in ui.stored_extra_pages: - with gr.Tab(page.title, id=page.id_page) as tab: + with gr.Tab(page.title, elem_id=f"{tabname}_{page.id_page}", elem_classes=["extra-page"]) as tab: elem_id = f"{tabname}_{page.id_page}_cards_html" page_elem = gr.HTML('Loading...', elem_id=elem_id) ui.pages.append(page_elem) @@ -389,11 +390,18 @@ def create_ui(interface: gr.Blocks, unrelated_tabs, tabname): ui.button_save_preview = gr.Button('Save preview', elem_id=tabname+"_save_preview", visible=False) ui.preview_target_filename = gr.Textbox('Preview save filename', elem_id=tabname+"_preview_filename", visible=False) - for tab in unrelated_tabs: - tab.select(fn=lambda: [gr.update(visible=False) for _ in range(5)], inputs=[], outputs=[edit_search, dropdown_sort, button_sortorder, button_refresh, checkbox_show_dirs], show_progress=False) + tab_controls = [edit_search, dropdown_sort, button_sortorder, button_refresh, checkbox_show_dirs] - for tab in related_tabs: - tab.select(fn=lambda: [gr.update(visible=True) for _ in range(5)], inputs=[], outputs=[edit_search, dropdown_sort, button_sortorder, button_refresh, checkbox_show_dirs], show_progress=False) + for tab in unrelated_tabs: + tab.select(fn=lambda: [gr.update(visible=False) for _ in tab_controls], _js='function(){ extraNetworksUrelatedTabSelected("' + tabname + '"); }', inputs=[], outputs=tab_controls, show_progress=False) + + for page, tab in zip(ui.stored_extra_pages, related_tabs): + allow_prompt = "true" if page.allow_prompt else "false" + allow_negative_prompt = "true" if page.allow_negative_prompt else "false" + + jscode = 'extraNetworksTabSelected("' + tabname + '", "' + f"{tabname}_{page.id_page}" + '", ' + allow_prompt + ', ' + allow_negative_prompt + ');' + + tab.select(fn=lambda: [gr.update(visible=True) for _ in tab_controls], _js='function(){ ' + jscode + ' }', inputs=[], outputs=tab_controls, show_progress=False) dropdown_sort.change(fn=lambda: None, _js="function(){ applyExtraNetworkSort('" + tabname + "'); }") diff --git a/modules/ui_extra_networks_checkpoints.py b/modules/ui_extra_networks_checkpoints.py index ca6c26076..2fc0ed43d 100644 --- a/modules/ui_extra_networks_checkpoints.py +++ b/modules/ui_extra_networks_checkpoints.py @@ -10,6 +10,8 @@ class ExtraNetworksPageCheckpoints(ui_extra_networks.ExtraNetworksPage): def __init__(self): super().__init__('Checkpoints') + self.allow_prompt = False + def refresh(self): shared.refresh_checkpoints() diff --git a/modules/ui_toprow.py b/modules/ui_toprow.py new file mode 100644 index 000000000..985b5a2dd --- /dev/null +++ b/modules/ui_toprow.py @@ -0,0 +1,141 @@ +import gradio as gr + +from modules import shared, ui_prompt_styles +import modules.images + +from modules.ui_components import ToolButton + + +class Toprow: + """Creates a top row UI with prompts, generate button, styles, extra little buttons for things, and enables some functionality related to their operation""" + + prompt = None + prompt_img = None + negative_prompt = None + + button_interrogate = None + button_deepbooru = None + + interrupt = None + skip = None + submit = None + + paste = None + clear_prompt_button = None + apply_styles = None + restore_progress_button = None + + token_counter = None + token_button = None + negative_token_counter = None + negative_token_button = None + + ui_styles = None + + submit_box = None + + def __init__(self, is_img2img, is_compact=False): + id_part = "img2img" if is_img2img else "txt2img" + self.id_part = id_part + self.is_img2img = is_img2img + self.is_compact = is_compact + + if not is_compact: + with gr.Row(elem_id=f"{id_part}_toprow", variant="compact"): + self.create_classic_toprow() + else: + self.create_submit_box() + + def create_classic_toprow(self): + self.create_prompts() + + with gr.Column(scale=1, elem_id=f"{self.id_part}_actions_column"): + self.create_submit_box() + + self.create_tools_row() + + self.create_styles_ui() + + def create_inline_toprow_prompts(self): + if not self.is_compact: + return + + self.create_prompts() + + with gr.Row(elem_classes=["toprow-compact-stylerow"]): + with gr.Column(elem_classes=["toprow-compact-tools"]): + self.create_tools_row() + with gr.Column(): + self.create_styles_ui() + + def create_inline_toprow_image(self): + if not self.is_compact: + return + + self.submit_box.render() + + def create_prompts(self): + with gr.Column(elem_id=f"{self.id_part}_prompt_container", elem_classes=["prompt-container-compact"] if self.is_compact else [], scale=6): + with gr.Row(elem_id=f"{self.id_part}_prompt_row", elem_classes=["prompt-row"]): + self.prompt = gr.Textbox(label="Prompt", elem_id=f"{self.id_part}_prompt", show_label=False, lines=3, placeholder="Prompt (press Ctrl+Enter or Alt+Enter to generate)", elem_classes=["prompt"]) + self.prompt_img = gr.File(label="", elem_id=f"{self.id_part}_prompt_image", file_count="single", type="binary", visible=False) + + with gr.Row(elem_id=f"{self.id_part}_neg_prompt_row", elem_classes=["prompt-row"]): + self.negative_prompt = gr.Textbox(label="Negative prompt", elem_id=f"{self.id_part}_neg_prompt", show_label=False, lines=3, placeholder="Negative prompt (press Ctrl+Enter or Alt+Enter to generate)", elem_classes=["prompt"]) + + self.prompt_img.change( + fn=modules.images.image_data, + inputs=[self.prompt_img], + outputs=[self.prompt, self.prompt_img], + show_progress=False, + ) + + def create_submit_box(self): + with gr.Row(elem_id=f"{self.id_part}_generate_box", elem_classes=["generate-box"] + (["generate-box-compact"] if self.is_compact else []), render=not self.is_compact) as submit_box: + self.submit_box = submit_box + + self.interrupt = gr.Button('Interrupt', elem_id=f"{self.id_part}_interrupt", elem_classes="generate-box-interrupt") + self.skip = gr.Button('Skip', elem_id=f"{self.id_part}_skip", elem_classes="generate-box-skip") + self.submit = gr.Button('Generate', elem_id=f"{self.id_part}_generate", variant='primary') + + self.skip.click( + fn=lambda: shared.state.skip(), + inputs=[], + outputs=[], + ) + + self.interrupt.click( + fn=lambda: shared.state.interrupt(), + inputs=[], + outputs=[], + ) + + def create_tools_row(self): + with gr.Row(elem_id=f"{self.id_part}_tools"): + from modules.ui import paste_symbol, clear_prompt_symbol, restore_progress_symbol + + self.paste = ToolButton(value=paste_symbol, elem_id="paste", tooltip="Read generation parameters from prompt or last generation if prompt is empty into user interface.") + self.clear_prompt_button = ToolButton(value=clear_prompt_symbol, elem_id=f"{self.id_part}_clear_prompt", tooltip="Clear prompt") + self.apply_styles = ToolButton(value=ui_prompt_styles.styles_materialize_symbol, elem_id=f"{self.id_part}_style_apply", tooltip="Apply all selected styles to prompts.") + + if self.is_img2img: + self.button_interrogate = ToolButton('📎', tooltip='Interrogate CLIP - use CLIP neural network to create a text describing the image, and put it into the prompt field', elem_id="interrogate") + self.button_deepbooru = ToolButton('📦', tooltip='Interrogate DeepBooru - use DeepBooru neural network to create a text describing the image, and put it into the prompt field', elem_id="deepbooru") + + self.restore_progress_button = ToolButton(value=restore_progress_symbol, elem_id=f"{self.id_part}_restore_progress", visible=False, tooltip="Restore progress") + + self.token_counter = gr.HTML(value="0/75", elem_id=f"{self.id_part}_token_counter", elem_classes=["token-counter"]) + self.token_button = gr.Button(visible=False, elem_id=f"{self.id_part}_token_button") + self.negative_token_counter = gr.HTML(value="0/75", elem_id=f"{self.id_part}_negative_token_counter", elem_classes=["token-counter"]) + self.negative_token_button = gr.Button(visible=False, elem_id=f"{self.id_part}_negative_token_button") + + self.clear_prompt_button.click( + fn=lambda *x: x, + _js="confirm_clear_prompt", + inputs=[self.prompt, self.negative_prompt], + outputs=[self.prompt, self.negative_prompt], + ) + + def create_styles_ui(self): + self.ui_styles = ui_prompt_styles.UiPromptStyles(self.id_part, self.prompt, self.negative_prompt) + self.ui_styles.setup_apply_button(self.apply_styles) diff --git a/style.css b/style.css index 9a1181e8b..731620226 100644 --- a/style.css +++ b/style.css @@ -296,6 +296,13 @@ input[type="checkbox"].input-accordion-checkbox{ min-height: 4.5em; } +#txt2img_generate, #img2img_generate { + min-height: 4.5em; +} +.generate-box-compact #txt2img_generate, .generate-box-compact #img2img_generate { + min-height: 3em; +} + @media screen and (min-width: 2500px) { #txt2img_gallery, #img2img_gallery { min-height: 768px; @@ -403,6 +410,15 @@ div#extras_scale_to_tab div.form{ min-width: 0.5em; } +div.toprow-compact-stylerow{ + margin: 0.5em 0; +} + +div.toprow-compact-tools{ + min-width: fit-content !important; + max-width: fit-content; +} + /* settings */ #quicksettings { align-items: end; @@ -525,7 +541,8 @@ table.popup-table .link{ height: 20px; background: #b4c0cc; border-radius: 3px !important; - top: -20px; + top: -14px; + left: 0px; width: 100%; } @@ -823,6 +840,10 @@ footer { /* extra networks UI */ +.extra-page .prompt{ + margin: 0 0 0.5em 0; +} + .extra-network-cards{ height: calc(100vh - 24rem); overflow: clip scroll; From c3699d4fd185d5a7285c5519f9bb4b6fec236d9f Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sun, 5 Nov 2023 19:23:48 +0300 Subject: [PATCH 061/342] compact prompt option disabled by default --- modules/shared_options.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/shared_options.py b/modules/shared_options.py index 4e3d75416..a9964fcbb 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -272,7 +272,7 @@ options_templates.update(options_section(('ui', "User interface"), { "hires_fix_show_sampler": OptionInfo(False, "Hires fix: show hires checkpoint and sampler selection").needs_reload_ui(), "hires_fix_show_prompts": OptionInfo(False, "Hires fix: show hires prompt and negative prompt").needs_reload_ui(), "disable_token_counters": OptionInfo(False, "Disable prompt token counters").needs_reload_ui(), - "compact_prompt_box": OptionInfo(True, "Compact prompt layout").info("puts prompt and negative prompt inside the Generate tab, leaving more vertical space for the image on the right").needs_reload_ui(), + "compact_prompt_box": OptionInfo(False, "Compact prompt layout").info("puts prompt and negative prompt inside the Generate tab, leaving more vertical space for the image on the right").needs_reload_ui(), })) From 80d639a440929e9effe4620ce74333de283e7efc Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sun, 5 Nov 2023 19:32:21 +0300 Subject: [PATCH 062/342] linter --- modules/sd_hijack.py | 2 +- modules/sd_models.py | 4 ++-- modules/sd_models_types.py | 2 +- 3 files changed, 4 insertions(+), 4 deletions(-) diff --git a/modules/sd_hijack.py b/modules/sd_hijack.py index 4fff418d7..c6d17764e 100644 --- a/modules/sd_hijack.py +++ b/modules/sd_hijack.py @@ -183,7 +183,7 @@ class StableDiffusionModelHijack: except Exception as e: errors.display(e, "applying cross attention optimization") undo_optimizations() - + def conv_ssd(self, m): delattr(m.model.diffusion_model.middle_block, '1') delattr(m.model.diffusion_model.middle_block, '2') diff --git a/modules/sd_models.py b/modules/sd_models.py index d76dc5803..1036a3b13 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -355,10 +355,10 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer model.is_ssd = model.is_sdxl and 'model.diffusion_model.middle_block.1.transformer_blocks.0.attn1.to_q.weight' not in state_dict.keys() if model.is_sdxl: sd_models_xl.extend_sdxl(model) - + if model.is_ssd: sd_hijack.model_hijack.conv_ssd(model) - + if shared.opts.sd_checkpoint_cache > 0: # cache newly loaded model checkpoints_loaded[checkpoint_info] = state_dict.copy() diff --git a/modules/sd_models_types.py b/modules/sd_models_types.py index 1f28942a4..f911fbb68 100644 --- a/modules/sd_models_types.py +++ b/modules/sd_models_types.py @@ -23,7 +23,7 @@ class WebuiSdModel(LatentDiffusion): is_sdxl: bool """True if the model's architecture is SDXL or SSD""" - + is_ssd: bool """True if the model is SSD""" From 6ad666e4794a57dd65790dd6a259d5d4330d45ed Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sun, 5 Nov 2023 19:46:20 +0300 Subject: [PATCH 063/342] more changes for #13865: fix formatting, rename the function, add comment and add a readme entry --- README.md | 1 + modules/sd_hijack.py | 24 +++++++++++++----------- modules/sd_models.py | 2 +- 3 files changed, 15 insertions(+), 12 deletions(-) diff --git a/README.md b/README.md index c7a4e363f..25ba070e7 100644 --- a/README.md +++ b/README.md @@ -91,6 +91,7 @@ A browser interface based on Gradio library for Stable Diffusion. - Eased resolution restriction: generated image's dimensions must be a multiple of 8 rather than 64 - Now with a license! - Reorder elements in the UI from settings screen +- [Segmind Stable Diffusion](https://huggingface.co/segmind/SSD-1B) support ## Installation and Running Make sure the required [dependencies](https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Dependencies) are met and follow the instructions available for: diff --git a/modules/sd_hijack.py b/modules/sd_hijack.py index c6d17764e..fba23c38b 100644 --- a/modules/sd_hijack.py +++ b/modules/sd_hijack.py @@ -184,17 +184,19 @@ class StableDiffusionModelHijack: errors.display(e, "applying cross attention optimization") undo_optimizations() - def conv_ssd(self, m): - delattr(m.model.diffusion_model.middle_block, '1') - delattr(m.model.diffusion_model.middle_block, '2') - for i in ['9','8','7','6','5','4']: - delattr(m.model.diffusion_model.input_blocks[7][1].transformer_blocks,i) - delattr(m.model.diffusion_model.input_blocks[8][1].transformer_blocks,i) - delattr(m.model.diffusion_model.output_blocks[0][1].transformer_blocks,i) - delattr(m.model.diffusion_model.output_blocks[1][1].transformer_blocks,i) - delattr(m.model.diffusion_model.output_blocks[4][1].transformer_blocks,'1') - delattr(m.model.diffusion_model.output_blocks[5][1].transformer_blocks,'1') - devices.torch_gc() + def convert_sdxl_to_ssd(self, m): + """Converts an SDXL model to a Segmind Stable Diffusion model (see https://huggingface.co/segmind/SSD-1B)""" + + delattr(m.model.diffusion_model.middle_block, '1') + delattr(m.model.diffusion_model.middle_block, '2') + for i in ['9', '8', '7', '6', '5', '4']: + delattr(m.model.diffusion_model.input_blocks[7][1].transformer_blocks, i) + delattr(m.model.diffusion_model.input_blocks[8][1].transformer_blocks, i) + delattr(m.model.diffusion_model.output_blocks[0][1].transformer_blocks, i) + delattr(m.model.diffusion_model.output_blocks[1][1].transformer_blocks, i) + delattr(m.model.diffusion_model.output_blocks[4][1].transformer_blocks, '1') + delattr(m.model.diffusion_model.output_blocks[5][1].transformer_blocks, '1') + devices.torch_gc() def hijack(self, m): conditioner = getattr(m, 'conditioner', None) diff --git a/modules/sd_models.py b/modules/sd_models.py index 1036a3b13..841402e86 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -357,7 +357,7 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer sd_models_xl.extend_sdxl(model) if model.is_ssd: - sd_hijack.model_hijack.conv_ssd(model) + sd_hijack.model_hijack.convert_sdxl_to_ssd(model) if shared.opts.sd_checkpoint_cache > 0: # cache newly loaded model From 656437e0a50212778746c67785d23b0ea14a8837 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Mon, 6 Nov 2023 10:32:21 +0300 Subject: [PATCH 064/342] fix img2img_tabs error --- modules/ui.py | 20 ++++++++++---------- 1 file changed, 10 insertions(+), 10 deletions(-) diff --git a/modules/ui.py b/modules/ui.py index 2454eb36b..accdb4572 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -689,19 +689,19 @@ def create_ui(): with gr.Column(scale=4): inpaint_full_res_padding = gr.Slider(label='Only masked padding, pixels', minimum=0, maximum=256, step=4, value=32, elem_id="img2img_inpaint_full_res_padding") - def select_img2img_tab(tab): - return gr.update(visible=tab in [2, 3, 4]), gr.update(visible=tab == 3), - - for i, elem in enumerate(img2img_tabs): - elem.select( - fn=lambda tab=i: select_img2img_tab(tab), - inputs=[], - outputs=[inpaint_controls, mask_alpha], - ) - if category not in {"accordions"}: scripts.scripts_img2img.setup_ui_for_section(category) + def select_img2img_tab(tab): + return gr.update(visible=tab in [2, 3, 4]), gr.update(visible=tab == 3), + + for i, elem in enumerate(img2img_tabs): + elem.select( + fn=lambda tab=i: select_img2img_tab(tab), + inputs=[], + outputs=[inpaint_controls, mask_alpha], + ) + img2img_gallery, generation_info, html_info, html_log = create_output_panel("img2img", opts.outdir_img2img_samples, toprow) img2img_args = dict( From 9c1c0da026cb7ef091a0f3fa24b14ae8634f6de5 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Mon, 6 Nov 2023 11:17:36 +0300 Subject: [PATCH 065/342] fix exception related to the pix2pix --- modules/sd_hijack.py | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/modules/sd_hijack.py b/modules/sd_hijack.py index fba23c38b..0157e19f0 100644 --- a/modules/sd_hijack.py +++ b/modules/sd_hijack.py @@ -256,8 +256,12 @@ class StableDiffusionModelHijack: self.layers = flatten(m) + import modules.models.diffusion.ddpm_edit + if isinstance(m, ldm.models.diffusion.ddpm.LatentDiffusion): sd_unet.original_forward = ldm_original_forward + elif isinstance(m, modules.models.diffusion.ddpm_edit.LatentDiffusion): + sd_unet.original_forward = ldm_original_forward elif isinstance(m, sgm.models.diffusion.DiffusionEngine): sd_unet.original_forward = sgm_original_forward else: From 9ba991cad8a15a99f71f5b2ec5feff7dd9d270d7 Mon Sep 17 00:00:00 2001 From: GerryDE Date: Tue, 7 Nov 2023 03:09:08 +0100 Subject: [PATCH 066/342] Add option to set notification sound volume --- javascript/notification.js | 6 +++++- modules/shared_options.py | 1 + 2 files changed, 6 insertions(+), 1 deletion(-) diff --git a/javascript/notification.js b/javascript/notification.js index 6d7995612..3ee972ae1 100644 --- a/javascript/notification.js +++ b/javascript/notification.js @@ -26,7 +26,11 @@ onAfterUiUpdate(function() { lastHeadImg = headImg; // play notification sound if available - gradioApp().querySelector('#audio_notification audio')?.play(); + const notificationAudio = gradioApp().querySelector('#audio_notification audio'); + if (notificationAudio) { + notificationAudio.volume = opts.notification_volume / 100.0 || 1.0; + notificationAudio.play(); + } if (document.hasFocus()) return; diff --git a/modules/shared_options.py b/modules/shared_options.py index a9964fcbb..d40db5306 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -64,6 +64,7 @@ options_templates.update(options_section(('saving-images', "Saving images/grids" "save_incomplete_images": OptionInfo(False, "Save incomplete images").info("save images that has been interrupted in mid-generation; even if not saved, they will still show up in webui output."), "notification_audio": OptionInfo(True, "Play notification sound after image generation").info("notification.mp3 should be present in the root directory").needs_reload_ui(), + "notification_volume": OptionInfo(100, "Notification sound volume", gr.Slider, {"minimum": 0, "maximum": 100, "step": 1}).info("in %"), })) options_templates.update(options_section(('saving-paths', "Paths for saving"), { From 5e80d9ee99c5899e5e2b130408ffb65a0585a62a Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Tue, 7 Nov 2023 11:33:16 +0300 Subject: [PATCH 067/342] fix pix2pix producing bad results --- modules/processing.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/processing.py b/modules/processing.py index 70ad1ebed..b0e240a46 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -296,7 +296,7 @@ class StableDiffusionProcessing: return conditioning def edit_image_conditioning(self, source_image): - conditioning_image = images_tensor_to_samples(source_image*0.5+0.5, approximation_indexes.get(opts.sd_vae_encode_method)) + conditioning_image = shared.sd_model.encode_first_stage(source_image).mode() return conditioning_image From a625a7bb817cbf6a97d2030dc3a8015a046bd388 Mon Sep 17 00:00:00 2001 From: Emily Zeng Date: Thu, 9 Nov 2023 13:15:06 -0500 Subject: [PATCH 068/342] moved nested with to single line to remove extra tabs --- modules/ui.py | 573 +++++++++++++++++++++++++------------------------- 1 file changed, 286 insertions(+), 287 deletions(-) diff --git a/modules/ui.py b/modules/ui.py index 4a3e60d13..0faccbd34 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -270,89 +270,88 @@ def create_ui(): extra_tabs.__enter__() with gr.Tab("Generation", id="txt2img_generation") as txt2img_generation_tab, ResizeHandleRow(equal_height=False): - with gr.Accordion("Open for Settings", open=False) if shared.opts.img2img_settings_accordion else gr.Group(): - with gr.Column(variant='compact', elem_id="txt2img_settings"): - scripts.scripts_txt2img.prepare_ui() + with gr.Accordion("Open for Settings", open=False), gr.Column(variant='compact', elem_id="txt2img_settings") if shared.opts.img2img_settings_accordion else gr.Column(variant='compact', elem_id="txt2img_settings"): + scripts.scripts_txt2img.prepare_ui() - for category in ordered_ui_categories(): - if category == "prompt": - toprow.create_inline_toprow_prompts() + for category in ordered_ui_categories(): + if category == "prompt": + toprow.create_inline_toprow_prompts() - if category == "sampler": - steps, sampler_name = create_sampler_and_steps_selection(sd_samplers.visible_sampler_names(), "txt2img") + if category == "sampler": + steps, sampler_name = create_sampler_and_steps_selection(sd_samplers.visible_sampler_names(), "txt2img") - elif category == "dimensions": - with FormRow(): - with gr.Column(elem_id="txt2img_column_size", scale=4): - width = gr.Slider(minimum=64, maximum=2048, step=8, label="Width", value=512, elem_id="txt2img_width") - height = gr.Slider(minimum=64, maximum=2048, step=8, label="Height", value=512, elem_id="txt2img_height") + elif category == "dimensions": + with FormRow(): + with gr.Column(elem_id="txt2img_column_size", scale=4): + width = gr.Slider(minimum=64, maximum=2048, step=8, label="Width", value=512, elem_id="txt2img_width") + height = gr.Slider(minimum=64, maximum=2048, step=8, label="Height", value=512, elem_id="txt2img_height") - with gr.Column(elem_id="txt2img_dimensions_row", scale=1, elem_classes="dimensions-tools"): - res_switch_btn = ToolButton(value=switch_values_symbol, elem_id="txt2img_res_switch_btn", tooltip="Switch width/height") + with gr.Column(elem_id="txt2img_dimensions_row", scale=1, elem_classes="dimensions-tools"): + res_switch_btn = ToolButton(value=switch_values_symbol, elem_id="txt2img_res_switch_btn", tooltip="Switch width/height") - if opts.dimensions_and_batch_together: - with gr.Column(elem_id="txt2img_column_batch"): - batch_count = gr.Slider(minimum=1, step=1, label='Batch count', value=1, elem_id="txt2img_batch_count") - batch_size = gr.Slider(minimum=1, maximum=8, step=1, label='Batch size', value=1, elem_id="txt2img_batch_size") - - elif category == "cfg": - with gr.Row(): - cfg_scale = gr.Slider(minimum=1.0, maximum=30.0, step=0.5, label='CFG Scale', value=7.0, elem_id="txt2img_cfg_scale") - - elif category == "checkboxes": - with FormRow(elem_classes="checkboxes-row", variant="compact"): - pass - - elif category == "accordions": - with gr.Row(elem_id="txt2img_accordions", elem_classes="accordions"): - with InputAccordion(False, label="Hires. fix", elem_id="txt2img_hr") as enable_hr: - with enable_hr.extra(): - hr_final_resolution = FormHTML(value="", elem_id="txtimg_hr_finalres", label="Upscaled resolution", interactive=False, min_width=0) - - with FormRow(elem_id="txt2img_hires_fix_row1", variant="compact"): - hr_upscaler = gr.Dropdown(label="Upscaler", elem_id="txt2img_hr_upscaler", choices=[*shared.latent_upscale_modes, *[x.name for x in shared.sd_upscalers]], value=shared.latent_upscale_default_mode) - hr_second_pass_steps = gr.Slider(minimum=0, maximum=150, step=1, label='Hires steps', value=0, elem_id="txt2img_hires_steps") - denoising_strength = gr.Slider(minimum=0.0, maximum=1.0, step=0.01, label='Denoising strength', value=0.7, elem_id="txt2img_denoising_strength") - - with FormRow(elem_id="txt2img_hires_fix_row2", variant="compact"): - hr_scale = gr.Slider(minimum=1.0, maximum=4.0, step=0.05, label="Upscale by", value=2.0, elem_id="txt2img_hr_scale") - hr_resize_x = gr.Slider(minimum=0, maximum=2048, step=8, label="Resize width to", value=0, elem_id="txt2img_hr_resize_x") - hr_resize_y = gr.Slider(minimum=0, maximum=2048, step=8, label="Resize height to", value=0, elem_id="txt2img_hr_resize_y") - - with FormRow(elem_id="txt2img_hires_fix_row3", variant="compact", visible=opts.hires_fix_show_sampler) as hr_sampler_container: - - hr_checkpoint_name = gr.Dropdown(label='Hires checkpoint', elem_id="hr_checkpoint", choices=["Use same checkpoint"] + modules.sd_models.checkpoint_tiles(use_short=True), value="Use same checkpoint") - create_refresh_button(hr_checkpoint_name, modules.sd_models.list_models, lambda: {"choices": ["Use same checkpoint"] + modules.sd_models.checkpoint_tiles(use_short=True)}, "hr_checkpoint_refresh") - - hr_sampler_name = gr.Dropdown(label='Hires sampling method', elem_id="hr_sampler", choices=["Use same sampler"] + sd_samplers.visible_sampler_names(), value="Use same sampler") - - with FormRow(elem_id="txt2img_hires_fix_row4", variant="compact", visible=opts.hires_fix_show_prompts) as hr_prompts_container: - with gr.Column(scale=80): - with gr.Row(): - hr_prompt = gr.Textbox(label="Hires prompt", elem_id="hires_prompt", show_label=False, lines=3, placeholder="Prompt for hires fix pass.\nLeave empty to use the same prompt as in first pass.", elem_classes=["prompt"]) - with gr.Column(scale=80): - with gr.Row(): - hr_negative_prompt = gr.Textbox(label="Hires negative prompt", elem_id="hires_neg_prompt", show_label=False, lines=3, placeholder="Negative prompt for hires fix pass.\nLeave empty to use the same negative prompt as in first pass.", elem_classes=["prompt"]) - - scripts.scripts_txt2img.setup_ui_for_section(category) - - elif category == "batch": - if not opts.dimensions_and_batch_together: - with FormRow(elem_id="txt2img_column_batch"): + if opts.dimensions_and_batch_together: + with gr.Column(elem_id="txt2img_column_batch"): batch_count = gr.Slider(minimum=1, step=1, label='Batch count', value=1, elem_id="txt2img_batch_count") batch_size = gr.Slider(minimum=1, maximum=8, step=1, label='Batch size', value=1, elem_id="txt2img_batch_size") - elif category == "override_settings": - with FormRow(elem_id="txt2img_override_settings_row") as row: - override_settings = create_override_settings_dropdown('txt2img', row) + elif category == "cfg": + with gr.Row(): + cfg_scale = gr.Slider(minimum=1.0, maximum=30.0, step=0.5, label='CFG Scale', value=7.0, elem_id="txt2img_cfg_scale") - elif category == "scripts": - with FormGroup(elem_id="txt2img_script_container"): - custom_inputs = scripts.scripts_txt2img.setup_ui() + elif category == "checkboxes": + with FormRow(elem_classes="checkboxes-row", variant="compact"): + pass + + elif category == "accordions": + with gr.Row(elem_id="txt2img_accordions", elem_classes="accordions"): + with InputAccordion(False, label="Hires. fix", elem_id="txt2img_hr") as enable_hr: + with enable_hr.extra(): + hr_final_resolution = FormHTML(value="", elem_id="txtimg_hr_finalres", label="Upscaled resolution", interactive=False, min_width=0) + + with FormRow(elem_id="txt2img_hires_fix_row1", variant="compact"): + hr_upscaler = gr.Dropdown(label="Upscaler", elem_id="txt2img_hr_upscaler", choices=[*shared.latent_upscale_modes, *[x.name for x in shared.sd_upscalers]], value=shared.latent_upscale_default_mode) + hr_second_pass_steps = gr.Slider(minimum=0, maximum=150, step=1, label='Hires steps', value=0, elem_id="txt2img_hires_steps") + denoising_strength = gr.Slider(minimum=0.0, maximum=1.0, step=0.01, label='Denoising strength', value=0.7, elem_id="txt2img_denoising_strength") + + with FormRow(elem_id="txt2img_hires_fix_row2", variant="compact"): + hr_scale = gr.Slider(minimum=1.0, maximum=4.0, step=0.05, label="Upscale by", value=2.0, elem_id="txt2img_hr_scale") + hr_resize_x = gr.Slider(minimum=0, maximum=2048, step=8, label="Resize width to", value=0, elem_id="txt2img_hr_resize_x") + hr_resize_y = gr.Slider(minimum=0, maximum=2048, step=8, label="Resize height to", value=0, elem_id="txt2img_hr_resize_y") + + with FormRow(elem_id="txt2img_hires_fix_row3", variant="compact", visible=opts.hires_fix_show_sampler) as hr_sampler_container: + + hr_checkpoint_name = gr.Dropdown(label='Hires checkpoint', elem_id="hr_checkpoint", choices=["Use same checkpoint"] + modules.sd_models.checkpoint_tiles(use_short=True), value="Use same checkpoint") + create_refresh_button(hr_checkpoint_name, modules.sd_models.list_models, lambda: {"choices": ["Use same checkpoint"] + modules.sd_models.checkpoint_tiles(use_short=True)}, "hr_checkpoint_refresh") + + hr_sampler_name = gr.Dropdown(label='Hires sampling method', elem_id="hr_sampler", choices=["Use same sampler"] + sd_samplers.visible_sampler_names(), value="Use same sampler") + + with FormRow(elem_id="txt2img_hires_fix_row4", variant="compact", visible=opts.hires_fix_show_prompts) as hr_prompts_container: + with gr.Column(scale=80): + with gr.Row(): + hr_prompt = gr.Textbox(label="Hires prompt", elem_id="hires_prompt", show_label=False, lines=3, placeholder="Prompt for hires fix pass.\nLeave empty to use the same prompt as in first pass.", elem_classes=["prompt"]) + with gr.Column(scale=80): + with gr.Row(): + hr_negative_prompt = gr.Textbox(label="Hires negative prompt", elem_id="hires_neg_prompt", show_label=False, lines=3, placeholder="Negative prompt for hires fix pass.\nLeave empty to use the same negative prompt as in first pass.", elem_classes=["prompt"]) - if category not in {"accordions"}: scripts.scripts_txt2img.setup_ui_for_section(category) + elif category == "batch": + if not opts.dimensions_and_batch_together: + with FormRow(elem_id="txt2img_column_batch"): + batch_count = gr.Slider(minimum=1, step=1, label='Batch count', value=1, elem_id="txt2img_batch_count") + batch_size = gr.Slider(minimum=1, maximum=8, step=1, label='Batch size', value=1, elem_id="txt2img_batch_size") + + elif category == "override_settings": + with FormRow(elem_id="txt2img_override_settings_row") as row: + override_settings = create_override_settings_dropdown('txt2img', row) + + elif category == "scripts": + with FormGroup(elem_id="txt2img_script_container"): + custom_inputs = scripts.scripts_txt2img.setup_ui() + + if category not in {"accordions"}: + scripts.scripts_txt2img.setup_ui_for_section(category) + hr_resolution_preview_inputs = [enable_hr, width, height, hr_scale, hr_resize_x, hr_resize_y] for component in hr_resolution_preview_inputs: @@ -490,258 +489,258 @@ def create_ui(): extra_tabs.__enter__() with gr.Tab("Generation", id="img2img_generation") as img2img_generation_tab, ResizeHandleRow(equal_height=False): - with gr.Accordion("Open for Settings", open=False) if shared.opts.img2img_settings_accordion else gr.Group(): - with gr.Column(variant='compact', elem_id="img2img_settings"): - copy_image_buttons = [] - copy_image_destinations = {} + with gr.Accordion("Open for Settings", open=False), gr.Column(variant='compact', elem_id="img2img_settings") if shared.opts.img2img_settings_accordion else gr.Column(variant='compact', elem_id="img2img_settings"): + copy_image_buttons = [] + copy_image_destinations = {} - def add_copy_image_controls(tab_name, elem): - with gr.Row(variant="compact", elem_id=f"img2img_copy_to_{tab_name}"): - gr.HTML("Copy image to: ", elem_id=f"img2img_label_copy_to_{tab_name}") + def add_copy_image_controls(tab_name, elem): + with gr.Row(variant="compact", elem_id=f"img2img_copy_to_{tab_name}"): + gr.HTML("Copy image to: ", elem_id=f"img2img_label_copy_to_{tab_name}") - for title, name in zip(['img2img', 'sketch', 'inpaint', 'inpaint sketch'], ['img2img', 'sketch', 'inpaint', 'inpaint_sketch']): - if name == tab_name: - gr.Button(title, interactive=False) - copy_image_destinations[name] = elem - continue + for title, name in zip(['img2img', 'sketch', 'inpaint', 'inpaint sketch'], ['img2img', 'sketch', 'inpaint', 'inpaint_sketch']): + if name == tab_name: + gr.Button(title, interactive=False) + copy_image_destinations[name] = elem + continue - button = gr.Button(title) - copy_image_buttons.append((button, name, elem)) + button = gr.Button(title) + copy_image_buttons.append((button, name, elem)) - scripts.scripts_img2img.prepare_ui() + scripts.scripts_img2img.prepare_ui() - for category in ordered_ui_categories(): - if category == "prompt": - toprow.create_inline_toprow_prompts() + for category in ordered_ui_categories(): + if category == "prompt": + toprow.create_inline_toprow_prompts() - if category == "image": - with gr.Tabs(elem_id="mode_img2img"): - img2img_selected_tab = gr.State(0) + if category == "image": + with gr.Tabs(elem_id="mode_img2img"): + img2img_selected_tab = gr.State(0) - with gr.TabItem('img2img', id='img2img', elem_id="img2img_img2img_tab") as tab_img2img: - init_img = gr.Image(label="Image for img2img", elem_id="img2img_image", show_label=False, source="upload", interactive=True, type="pil", tool="editor", image_mode="RGBA", height=opts.img2img_editor_height) - add_copy_image_controls('img2img', init_img) + with gr.TabItem('img2img', id='img2img', elem_id="img2img_img2img_tab") as tab_img2img: + init_img = gr.Image(label="Image for img2img", elem_id="img2img_image", show_label=False, source="upload", interactive=True, type="pil", tool="editor", image_mode="RGBA", height=opts.img2img_editor_height) + add_copy_image_controls('img2img', init_img) - with gr.TabItem('Sketch', id='img2img_sketch', elem_id="img2img_img2img_sketch_tab") as tab_sketch: - sketch = gr.Image(label="Image for img2img", elem_id="img2img_sketch", show_label=False, source="upload", interactive=True, type="pil", tool="color-sketch", image_mode="RGB", height=opts.img2img_editor_height, brush_color=opts.img2img_sketch_default_brush_color) - add_copy_image_controls('sketch', sketch) + with gr.TabItem('Sketch', id='img2img_sketch', elem_id="img2img_img2img_sketch_tab") as tab_sketch: + sketch = gr.Image(label="Image for img2img", elem_id="img2img_sketch", show_label=False, source="upload", interactive=True, type="pil", tool="color-sketch", image_mode="RGB", height=opts.img2img_editor_height, brush_color=opts.img2img_sketch_default_brush_color) + add_copy_image_controls('sketch', sketch) - with gr.TabItem('Inpaint', id='inpaint', elem_id="img2img_inpaint_tab") as tab_inpaint: - init_img_with_mask = gr.Image(label="Image for inpainting with mask", show_label=False, elem_id="img2maskimg", source="upload", interactive=True, type="pil", tool="sketch", image_mode="RGBA", height=opts.img2img_editor_height, brush_color=opts.img2img_inpaint_mask_brush_color) - add_copy_image_controls('inpaint', init_img_with_mask) + with gr.TabItem('Inpaint', id='inpaint', elem_id="img2img_inpaint_tab") as tab_inpaint: + init_img_with_mask = gr.Image(label="Image for inpainting with mask", show_label=False, elem_id="img2maskimg", source="upload", interactive=True, type="pil", tool="sketch", image_mode="RGBA", height=opts.img2img_editor_height, brush_color=opts.img2img_inpaint_mask_brush_color) + add_copy_image_controls('inpaint', init_img_with_mask) - with gr.TabItem('Inpaint sketch', id='inpaint_sketch', elem_id="img2img_inpaint_sketch_tab") as tab_inpaint_color: - inpaint_color_sketch = gr.Image(label="Color sketch inpainting", show_label=False, elem_id="inpaint_sketch", source="upload", interactive=True, type="pil", tool="color-sketch", image_mode="RGB", height=opts.img2img_editor_height, brush_color=opts.img2img_inpaint_sketch_default_brush_color) - inpaint_color_sketch_orig = gr.State(None) - add_copy_image_controls('inpaint_sketch', inpaint_color_sketch) + with gr.TabItem('Inpaint sketch', id='inpaint_sketch', elem_id="img2img_inpaint_sketch_tab") as tab_inpaint_color: + inpaint_color_sketch = gr.Image(label="Color sketch inpainting", show_label=False, elem_id="inpaint_sketch", source="upload", interactive=True, type="pil", tool="color-sketch", image_mode="RGB", height=opts.img2img_editor_height, brush_color=opts.img2img_inpaint_sketch_default_brush_color) + inpaint_color_sketch_orig = gr.State(None) + add_copy_image_controls('inpaint_sketch', inpaint_color_sketch) - def update_orig(image, state): - if image is not None: - same_size = state is not None and state.size == image.size - has_exact_match = np.any(np.all(np.array(image) == np.array(state), axis=-1)) - edited = same_size and has_exact_match - return image if not edited or state is None else state + def update_orig(image, state): + if image is not None: + same_size = state is not None and state.size == image.size + has_exact_match = np.any(np.all(np.array(image) == np.array(state), axis=-1)) + edited = same_size and has_exact_match + return image if not edited or state is None else state - inpaint_color_sketch.change(update_orig, [inpaint_color_sketch, inpaint_color_sketch_orig], inpaint_color_sketch_orig) + inpaint_color_sketch.change(update_orig, [inpaint_color_sketch, inpaint_color_sketch_orig], inpaint_color_sketch_orig) - with gr.TabItem('Inpaint upload', id='inpaint_upload', elem_id="img2img_inpaint_upload_tab") as tab_inpaint_upload: - init_img_inpaint = gr.Image(label="Image for img2img", show_label=False, source="upload", interactive=True, type="pil", elem_id="img_inpaint_base") - init_mask_inpaint = gr.Image(label="Mask", source="upload", interactive=True, type="pil", image_mode="RGBA", elem_id="img_inpaint_mask") + with gr.TabItem('Inpaint upload', id='inpaint_upload', elem_id="img2img_inpaint_upload_tab") as tab_inpaint_upload: + init_img_inpaint = gr.Image(label="Image for img2img", show_label=False, source="upload", interactive=True, type="pil", elem_id="img_inpaint_base") + init_mask_inpaint = gr.Image(label="Mask", source="upload", interactive=True, type="pil", image_mode="RGBA", elem_id="img_inpaint_mask") - with gr.TabItem('Batch', id='batch', elem_id="img2img_batch_tab") as tab_batch: - hidden = '
Disabled when launched with --hide-ui-dir-config.' if shared.cmd_opts.hide_ui_dir_config else '' - gr.HTML( - "

Process images in a directory on the same machine where the server is running." + - "
Use an empty output directory to save pictures normally instead of writing to the output directory." + - f"
Add inpaint batch mask directory to enable inpaint batch processing." - f"{hidden}

" + with gr.TabItem('Batch', id='batch', elem_id="img2img_batch_tab") as tab_batch: + hidden = '
Disabled when launched with --hide-ui-dir-config.' if shared.cmd_opts.hide_ui_dir_config else '' + gr.HTML( + "

Process images in a directory on the same machine where the server is running." + + "
Use an empty output directory to save pictures normally instead of writing to the output directory." + + f"
Add inpaint batch mask directory to enable inpaint batch processing." + f"{hidden}

" + ) + img2img_batch_input_dir = gr.Textbox(label="Input directory", **shared.hide_dirs, elem_id="img2img_batch_input_dir") + img2img_batch_output_dir = gr.Textbox(label="Output directory", **shared.hide_dirs, elem_id="img2img_batch_output_dir") + img2img_batch_inpaint_mask_dir = gr.Textbox(label="Inpaint batch mask directory (required for inpaint batch processing only)", **shared.hide_dirs, elem_id="img2img_batch_inpaint_mask_dir") + with gr.Accordion("PNG info", open=False): + img2img_batch_use_png_info = gr.Checkbox(label="Append png info to prompts", **shared.hide_dirs, elem_id="img2img_batch_use_png_info") + img2img_batch_png_info_dir = gr.Textbox(label="PNG info directory", **shared.hide_dirs, placeholder="Leave empty to use input directory", elem_id="img2img_batch_png_info_dir") + img2img_batch_png_info_props = gr.CheckboxGroup(["Prompt", "Negative prompt", "Seed", "CFG scale", "Sampler", "Steps", "Model hash"], label="Parameters to take from png info", info="Prompts from png info will be appended to prompts set in ui.") + + img2img_tabs = [tab_img2img, tab_sketch, tab_inpaint, tab_inpaint_color, tab_inpaint_upload, tab_batch] + + for i, tab in enumerate(img2img_tabs): + tab.select(fn=lambda tabnum=i: tabnum, inputs=[], outputs=[img2img_selected_tab]) + + def copy_image(img): + if isinstance(img, dict) and 'image' in img: + return img['image'] + + return img + + for button, name, elem in copy_image_buttons: + button.click( + fn=copy_image, + inputs=[elem], + outputs=[copy_image_destinations[name]], + ) + button.click( + fn=lambda: None, + _js=f"switch_to_{name.replace(' ', '_')}", + inputs=[], + outputs=[], + ) + + with FormRow(): + resize_mode = gr.Radio(label="Resize mode", elem_id="resize_mode", choices=["Just resize", "Crop and resize", "Resize and fill", "Just resize (latent upscale)"], type="index", value="Just resize") + + if category == "sampler": + steps, sampler_name = create_sampler_and_steps_selection(sd_samplers.visible_sampler_names(), "img2img") + + elif category == "dimensions": + with FormRow(): + with gr.Column(elem_id="img2img_column_size", scale=4): + selected_scale_tab = gr.State(value=0) + + with gr.Tabs(): + with gr.Tab(label="Resize to", elem_id="img2img_tab_resize_to") as tab_scale_to: + with FormRow(): + with gr.Column(elem_id="img2img_column_size", scale=4): + width = gr.Slider(minimum=64, maximum=2048, step=8, label="Width", value=512, elem_id="img2img_width") + height = gr.Slider(minimum=64, maximum=2048, step=8, label="Height", value=512, elem_id="img2img_height") + with gr.Column(elem_id="img2img_dimensions_row", scale=1, elem_classes="dimensions-tools"): + res_switch_btn = ToolButton(value=switch_values_symbol, elem_id="img2img_res_switch_btn", tooltip="Switch width/height") + detect_image_size_btn = ToolButton(value=detect_image_size_symbol, elem_id="img2img_detect_image_size_btn", tooltip="Auto detect size from img2img") + + with gr.Tab(label="Resize by", elem_id="img2img_tab_resize_by") as tab_scale_by: + scale_by = gr.Slider(minimum=0.05, maximum=4.0, step=0.05, label="Scale", value=1.0, elem_id="img2img_scale") + + with FormRow(): + scale_by_html = FormHTML(resize_from_to_html(0, 0, 0.0), elem_id="img2img_scale_resolution_preview") + gr.Slider(label="Unused", elem_id="img2img_unused_scale_by_slider") + button_update_resize_to = gr.Button(visible=False, elem_id="img2img_update_resize_to") + + on_change_args = dict( + fn=resize_from_to_html, + _js="currentImg2imgSourceResolution", + inputs=[dummy_component, dummy_component, scale_by], + outputs=scale_by_html, + show_progress=False, ) - img2img_batch_input_dir = gr.Textbox(label="Input directory", **shared.hide_dirs, elem_id="img2img_batch_input_dir") - img2img_batch_output_dir = gr.Textbox(label="Output directory", **shared.hide_dirs, elem_id="img2img_batch_output_dir") - img2img_batch_inpaint_mask_dir = gr.Textbox(label="Inpaint batch mask directory (required for inpaint batch processing only)", **shared.hide_dirs, elem_id="img2img_batch_inpaint_mask_dir") - with gr.Accordion("PNG info", open=False): - img2img_batch_use_png_info = gr.Checkbox(label="Append png info to prompts", **shared.hide_dirs, elem_id="img2img_batch_use_png_info") - img2img_batch_png_info_dir = gr.Textbox(label="PNG info directory", **shared.hide_dirs, placeholder="Leave empty to use input directory", elem_id="img2img_batch_png_info_dir") - img2img_batch_png_info_props = gr.CheckboxGroup(["Prompt", "Negative prompt", "Seed", "CFG scale", "Sampler", "Steps", "Model hash"], label="Parameters to take from png info", info="Prompts from png info will be appended to prompts set in ui.") - img2img_tabs = [tab_img2img, tab_sketch, tab_inpaint, tab_inpaint_color, tab_inpaint_upload, tab_batch] + scale_by.release(**on_change_args) + button_update_resize_to.click(**on_change_args) - for i, tab in enumerate(img2img_tabs): - tab.select(fn=lambda tabnum=i: tabnum, inputs=[], outputs=[img2img_selected_tab]) + # the code below is meant to update the resolution label after the image in the image selection UI has changed. + # as it is now the event keeps firing continuously for inpaint edits, which ruins the page with constant requests. + # I assume this must be a gradio bug and for now we'll just do it for non-inpaint inputs. + for component in [init_img, sketch]: + component.change(fn=lambda: None, _js="updateImg2imgResizeToTextAfterChangingImage", inputs=[], outputs=[], show_progress=False) - def copy_image(img): - if isinstance(img, dict) and 'image' in img: - return img['image'] + with FormRow(): + resize_mode = gr.Radio(label="Resize mode", elem_id="resize_mode", choices=["Just resize", "Crop and resize", "Resize and fill", "Just resize (latent upscale)"], type="index", value="Just resize") - return img + scripts.scripts_img2img.prepare_ui() - for button, name, elem in copy_image_buttons: - button.click( - fn=copy_image, - inputs=[elem], - outputs=[copy_image_destinations[name]], - ) - button.click( - fn=lambda: None, - _js=f"switch_to_{name.replace(' ', '_')}", - inputs=[], - outputs=[], - ) + for category in ordered_ui_categories(): + if category == "sampler": + steps, sampler_name = create_sampler_and_steps_selection(sd_samplers.visible_sampler_names(), "img2img") - with FormRow(): - resize_mode = gr.Radio(label="Resize mode", elem_id="resize_mode", choices=["Just resize", "Crop and resize", "Resize and fill", "Just resize (latent upscale)"], type="index", value="Just resize") + elif category == "dimensions": + with FormRow(): + with gr.Column(elem_id="img2img_column_size", scale=4): + selected_scale_tab = gr.State(value=0) - if category == "sampler": - steps, sampler_name = create_sampler_and_steps_selection(sd_samplers.visible_sampler_names(), "img2img") + with gr.Tabs(): + with gr.Tab(label="Resize to", elem_id="img2img_tab_resize_to") as tab_scale_to: + with FormRow(): + with gr.Column(elem_id="img2img_column_size", scale=4): + width = gr.Slider(minimum=64, maximum=2048, step=8, label="Width", value=512, elem_id="img2img_width") + height = gr.Slider(minimum=64, maximum=2048, step=8, label="Height", value=512, elem_id="img2img_height") + with gr.Column(elem_id="img2img_dimensions_row", scale=1, elem_classes="dimensions-tools"): + res_switch_btn = ToolButton(value=switch_values_symbol, elem_id="img2img_res_switch_btn", tooltip="Switch width/height") + detect_image_size_btn = ToolButton(value=detect_image_size_symbol, elem_id="img2img_detect_image_size_btn", tooltip="Auto detect size from img2img") - elif category == "dimensions": - with FormRow(): - with gr.Column(elem_id="img2img_column_size", scale=4): - selected_scale_tab = gr.State(value=0) + with gr.Tab(label="Resize by", elem_id="img2img_tab_resize_by") as tab_scale_by: + scale_by = gr.Slider(minimum=0.05, maximum=4.0, step=0.05, label="Scale", value=1.0, elem_id="img2img_scale") - with gr.Tabs(): - with gr.Tab(label="Resize to", elem_id="img2img_tab_resize_to") as tab_scale_to: - with FormRow(): - with gr.Column(elem_id="img2img_column_size", scale=4): - width = gr.Slider(minimum=64, maximum=2048, step=8, label="Width", value=512, elem_id="img2img_width") - height = gr.Slider(minimum=64, maximum=2048, step=8, label="Height", value=512, elem_id="img2img_height") - with gr.Column(elem_id="img2img_dimensions_row", scale=1, elem_classes="dimensions-tools"): - res_switch_btn = ToolButton(value=switch_values_symbol, elem_id="img2img_res_switch_btn", tooltip="Switch width/height") - detect_image_size_btn = ToolButton(value=detect_image_size_symbol, elem_id="img2img_detect_image_size_btn", tooltip="Auto detect size from img2img") + with FormRow(): + scale_by_html = FormHTML(resize_from_to_html(0, 0, 0.0), elem_id="img2img_scale_resolution_preview") + gr.Slider(label="Unused", elem_id="img2img_unused_scale_by_slider") + button_update_resize_to = gr.Button(visible=False, elem_id="img2img_update_resize_to") - with gr.Tab(label="Resize by", elem_id="img2img_tab_resize_by") as tab_scale_by: - scale_by = gr.Slider(minimum=0.05, maximum=4.0, step=0.05, label="Scale", value=1.0, elem_id="img2img_scale") + on_change_args = dict( + fn=resize_from_to_html, + _js="currentImg2imgSourceResolution", + inputs=[dummy_component, dummy_component, scale_by], + outputs=scale_by_html, + show_progress=False, + ) - with FormRow(): - scale_by_html = FormHTML(resize_from_to_html(0, 0, 0.0), elem_id="img2img_scale_resolution_preview") - gr.Slider(label="Unused", elem_id="img2img_unused_scale_by_slider") - button_update_resize_to = gr.Button(visible=False, elem_id="img2img_update_resize_to") + scale_by.release(**on_change_args) + button_update_resize_to.click(**on_change_args) - on_change_args = dict( - fn=resize_from_to_html, - _js="currentImg2imgSourceResolution", - inputs=[dummy_component, dummy_component, scale_by], - outputs=scale_by_html, - show_progress=False, - ) + # the code below is meant to update the resolution label after the image in the image selection UI has changed. + # as it is now the event keeps firing continuously for inpaint edits, which ruins the page with constant requests. + # I assume this must be a gradio bug and for now we'll just do it for non-inpaint inputs. + for component in [init_img, sketch]: + component.change(fn=lambda: None, _js="updateImg2imgResizeToTextAfterChangingImage", inputs=[], outputs=[], show_progress=False) - scale_by.release(**on_change_args) - button_update_resize_to.click(**on_change_args) + tab_scale_to.select(fn=lambda: 0, inputs=[], outputs=[selected_scale_tab]) + tab_scale_by.select(fn=lambda: 1, inputs=[], outputs=[selected_scale_tab]) - # the code below is meant to update the resolution label after the image in the image selection UI has changed. - # as it is now the event keeps firing continuously for inpaint edits, which ruins the page with constant requests. - # I assume this must be a gradio bug and for now we'll just do it for non-inpaint inputs. - for component in [init_img, sketch]: - component.change(fn=lambda: None, _js="updateImg2imgResizeToTextAfterChangingImage", inputs=[], outputs=[], show_progress=False) - - with FormRow(): - resize_mode = gr.Radio(label="Resize mode", elem_id="resize_mode", choices=["Just resize", "Crop and resize", "Resize and fill", "Just resize (latent upscale)"], type="index", value="Just resize") - - scripts.scripts_img2img.prepare_ui() - - for category in ordered_ui_categories(): - if category == "sampler": - steps, sampler_name = create_sampler_and_steps_selection(sd_samplers.visible_sampler_names(), "img2img") - - elif category == "dimensions": - with FormRow(): - with gr.Column(elem_id="img2img_column_size", scale=4): - selected_scale_tab = gr.State(value=0) - - with gr.Tabs(): - with gr.Tab(label="Resize to", elem_id="img2img_tab_resize_to") as tab_scale_to: - with FormRow(): - with gr.Column(elem_id="img2img_column_size", scale=4): - width = gr.Slider(minimum=64, maximum=2048, step=8, label="Width", value=512, elem_id="img2img_width") - height = gr.Slider(minimum=64, maximum=2048, step=8, label="Height", value=512, elem_id="img2img_height") - with gr.Column(elem_id="img2img_dimensions_row", scale=1, elem_classes="dimensions-tools"): - res_switch_btn = ToolButton(value=switch_values_symbol, elem_id="img2img_res_switch_btn", tooltip="Switch width/height") - detect_image_size_btn = ToolButton(value=detect_image_size_symbol, elem_id="img2img_detect_image_size_btn", tooltip="Auto detect size from img2img") - - with gr.Tab(label="Resize by", elem_id="img2img_tab_resize_by") as tab_scale_by: - scale_by = gr.Slider(minimum=0.05, maximum=4.0, step=0.05, label="Scale", value=1.0, elem_id="img2img_scale") - - with FormRow(): - scale_by_html = FormHTML(resize_from_to_html(0, 0, 0.0), elem_id="img2img_scale_resolution_preview") - gr.Slider(label="Unused", elem_id="img2img_unused_scale_by_slider") - button_update_resize_to = gr.Button(visible=False, elem_id="img2img_update_resize_to") - - on_change_args = dict( - fn=resize_from_to_html, - _js="currentImg2imgSourceResolution", - inputs=[dummy_component, dummy_component, scale_by], - outputs=scale_by_html, - show_progress=False, - ) - - scale_by.release(**on_change_args) - button_update_resize_to.click(**on_change_args) - - # the code below is meant to update the resolution label after the image in the image selection UI has changed. - # as it is now the event keeps firing continuously for inpaint edits, which ruins the page with constant requests. - # I assume this must be a gradio bug and for now we'll just do it for non-inpaint inputs. - for component in [init_img, sketch]: - component.change(fn=lambda: None, _js="updateImg2imgResizeToTextAfterChangingImage", inputs=[], outputs=[], show_progress=False) - - tab_scale_to.select(fn=lambda: 0, inputs=[], outputs=[selected_scale_tab]) - tab_scale_by.select(fn=lambda: 1, inputs=[], outputs=[selected_scale_tab]) - - if opts.dimensions_and_batch_together: - with gr.Column(elem_id="img2img_column_batch"): - batch_count = gr.Slider(minimum=1, step=1, label='Batch count', value=1, elem_id="img2img_batch_count") - batch_size = gr.Slider(minimum=1, maximum=8, step=1, label='Batch size', value=1, elem_id="img2img_batch_size") - - elif category == "denoising": - denoising_strength = gr.Slider(minimum=0.0, maximum=1.0, step=0.01, label='Denoising strength', value=0.75, elem_id="img2img_denoising_strength") - - elif category == "cfg": - with gr.Row(): - cfg_scale = gr.Slider(minimum=1.0, maximum=30.0, step=0.5, label='CFG Scale', value=7.0, elem_id="img2img_cfg_scale") - image_cfg_scale = gr.Slider(minimum=0, maximum=3.0, step=0.05, label='Image CFG Scale', value=1.5, elem_id="img2img_image_cfg_scale", visible=False) - - elif category == "checkboxes": - with FormRow(elem_classes="checkboxes-row", variant="compact"): - pass - - elif category == "accordions": - with gr.Row(elem_id="img2img_accordions", elem_classes="accordions"): - scripts.scripts_img2img.setup_ui_for_section(category) - - elif category == "batch": - if not opts.dimensions_and_batch_together: - with FormRow(elem_id="img2img_column_batch"): + if opts.dimensions_and_batch_together: + with gr.Column(elem_id="img2img_column_batch"): batch_count = gr.Slider(minimum=1, step=1, label='Batch count', value=1, elem_id="img2img_batch_count") batch_size = gr.Slider(minimum=1, maximum=8, step=1, label='Batch size', value=1, elem_id="img2img_batch_size") - elif category == "override_settings": - with FormRow(elem_id="img2img_override_settings_row") as row: - override_settings = create_override_settings_dropdown('img2img', row) + elif category == "denoising": + denoising_strength = gr.Slider(minimum=0.0, maximum=1.0, step=0.01, label='Denoising strength', value=0.75, elem_id="img2img_denoising_strength") - elif category == "scripts": - with FormGroup(elem_id="img2img_script_container"): - custom_inputs = scripts.scripts_img2img.setup_ui() + elif category == "cfg": + with gr.Row(): + cfg_scale = gr.Slider(minimum=1.0, maximum=30.0, step=0.5, label='CFG Scale', value=7.0, elem_id="img2img_cfg_scale") + image_cfg_scale = gr.Slider(minimum=0, maximum=3.0, step=0.05, label='Image CFG Scale', value=1.5, elem_id="img2img_image_cfg_scale", visible=False) - elif category == "inpaint": - with FormGroup(elem_id="inpaint_controls", visible=False) as inpaint_controls: - with FormRow(): - mask_blur = gr.Slider(label='Mask blur', minimum=0, maximum=64, step=1, value=4, elem_id="img2img_mask_blur") - mask_alpha = gr.Slider(label="Mask transparency", visible=False, elem_id="img2img_mask_alpha") + elif category == "checkboxes": + with FormRow(elem_classes="checkboxes-row", variant="compact"): + pass - with FormRow(): - inpainting_mask_invert = gr.Radio(label='Mask mode', choices=['Inpaint masked', 'Inpaint not masked'], value='Inpaint masked', type="index", elem_id="img2img_mask_mode") - - with FormRow(): - inpainting_fill = gr.Radio(label='Masked content', choices=['fill', 'original', 'latent noise', 'latent nothing'], value='original', type="index", elem_id="img2img_inpainting_fill") - - with FormRow(): - with gr.Column(): - inpaint_full_res = gr.Radio(label="Inpaint area", choices=["Whole picture", "Only masked"], type="index", value="Whole picture", elem_id="img2img_inpaint_full_res") - - with gr.Column(scale=4): - inpaint_full_res_padding = gr.Slider(label='Only masked padding, pixels', minimum=0, maximum=256, step=4, value=32, elem_id="img2img_inpaint_full_res_padding") - - if category not in {"accordions"}: + elif category == "accordions": + with gr.Row(elem_id="img2img_accordions", elem_classes="accordions"): scripts.scripts_img2img.setup_ui_for_section(category) + + elif category == "batch": + if not opts.dimensions_and_batch_together: + with FormRow(elem_id="img2img_column_batch"): + batch_count = gr.Slider(minimum=1, step=1, label='Batch count', value=1, elem_id="img2img_batch_count") + batch_size = gr.Slider(minimum=1, maximum=8, step=1, label='Batch size', value=1, elem_id="img2img_batch_size") + + elif category == "override_settings": + with FormRow(elem_id="img2img_override_settings_row") as row: + override_settings = create_override_settings_dropdown('img2img', row) + + elif category == "scripts": + with FormGroup(elem_id="img2img_script_container"): + custom_inputs = scripts.scripts_img2img.setup_ui() + + elif category == "inpaint": + with FormGroup(elem_id="inpaint_controls", visible=False) as inpaint_controls: + with FormRow(): + mask_blur = gr.Slider(label='Mask blur', minimum=0, maximum=64, step=1, value=4, elem_id="img2img_mask_blur") + mask_alpha = gr.Slider(label="Mask transparency", visible=False, elem_id="img2img_mask_alpha") + + with FormRow(): + inpainting_mask_invert = gr.Radio(label='Mask mode', choices=['Inpaint masked', 'Inpaint not masked'], value='Inpaint masked', type="index", elem_id="img2img_mask_mode") + + with FormRow(): + inpainting_fill = gr.Radio(label='Masked content', choices=['fill', 'original', 'latent noise', 'latent nothing'], value='original', type="index", elem_id="img2img_inpainting_fill") + + with FormRow(): + with gr.Column(): + inpaint_full_res = gr.Radio(label="Inpaint area", choices=["Whole picture", "Only masked"], type="index", value="Whole picture", elem_id="img2img_inpaint_full_res") + + with gr.Column(scale=4): + inpaint_full_res_padding = gr.Slider(label='Only masked padding, pixels', minimum=0, maximum=256, step=4, value=32, elem_id="img2img_inpaint_full_res_padding") + + if category not in {"accordions"}: + scripts.scripts_img2img.setup_ui_for_section(category) + def select_img2img_tab(tab): return gr.update(visible=tab in [2, 3, 4]), gr.update(visible=tab == 3), From 9aa4d098f07655d99cd16e8e9c984d043dbf9006 Mon Sep 17 00:00:00 2001 From: Emily Zeng Date: Thu, 9 Nov 2023 13:25:24 -0500 Subject: [PATCH 069/342] removed changes that weren't merged properly --- modules/ui.py | 51 +-------------------------------------------------- 1 file changed, 1 insertion(+), 50 deletions(-) diff --git a/modules/ui.py b/modules/ui.py index 0faccbd34..3eec7839f 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -592,55 +592,6 @@ def create_ui(): if category == "sampler": steps, sampler_name = create_sampler_and_steps_selection(sd_samplers.visible_sampler_names(), "img2img") - elif category == "dimensions": - with FormRow(): - with gr.Column(elem_id="img2img_column_size", scale=4): - selected_scale_tab = gr.State(value=0) - - with gr.Tabs(): - with gr.Tab(label="Resize to", elem_id="img2img_tab_resize_to") as tab_scale_to: - with FormRow(): - with gr.Column(elem_id="img2img_column_size", scale=4): - width = gr.Slider(minimum=64, maximum=2048, step=8, label="Width", value=512, elem_id="img2img_width") - height = gr.Slider(minimum=64, maximum=2048, step=8, label="Height", value=512, elem_id="img2img_height") - with gr.Column(elem_id="img2img_dimensions_row", scale=1, elem_classes="dimensions-tools"): - res_switch_btn = ToolButton(value=switch_values_symbol, elem_id="img2img_res_switch_btn", tooltip="Switch width/height") - detect_image_size_btn = ToolButton(value=detect_image_size_symbol, elem_id="img2img_detect_image_size_btn", tooltip="Auto detect size from img2img") - - with gr.Tab(label="Resize by", elem_id="img2img_tab_resize_by") as tab_scale_by: - scale_by = gr.Slider(minimum=0.05, maximum=4.0, step=0.05, label="Scale", value=1.0, elem_id="img2img_scale") - - with FormRow(): - scale_by_html = FormHTML(resize_from_to_html(0, 0, 0.0), elem_id="img2img_scale_resolution_preview") - gr.Slider(label="Unused", elem_id="img2img_unused_scale_by_slider") - button_update_resize_to = gr.Button(visible=False, elem_id="img2img_update_resize_to") - - on_change_args = dict( - fn=resize_from_to_html, - _js="currentImg2imgSourceResolution", - inputs=[dummy_component, dummy_component, scale_by], - outputs=scale_by_html, - show_progress=False, - ) - - scale_by.release(**on_change_args) - button_update_resize_to.click(**on_change_args) - - # the code below is meant to update the resolution label after the image in the image selection UI has changed. - # as it is now the event keeps firing continuously for inpaint edits, which ruins the page with constant requests. - # I assume this must be a gradio bug and for now we'll just do it for non-inpaint inputs. - for component in [init_img, sketch]: - component.change(fn=lambda: None, _js="updateImg2imgResizeToTextAfterChangingImage", inputs=[], outputs=[], show_progress=False) - - with FormRow(): - resize_mode = gr.Radio(label="Resize mode", elem_id="resize_mode", choices=["Just resize", "Crop and resize", "Resize and fill", "Just resize (latent upscale)"], type="index", value="Just resize") - - scripts.scripts_img2img.prepare_ui() - - for category in ordered_ui_categories(): - if category == "sampler": - steps, sampler_name = create_sampler_and_steps_selection(sd_samplers.visible_sampler_names(), "img2img") - elif category == "dimensions": with FormRow(): with gr.Column(elem_id="img2img_column_size", scale=4): @@ -740,7 +691,7 @@ def create_ui(): if category not in {"accordions"}: scripts.scripts_img2img.setup_ui_for_section(category) - + def select_img2img_tab(tab): return gr.update(visible=tab in [2, 3, 4]), gr.update(visible=tab == 3), From ff2952f10551aab2000002079d5f862af979e964 Mon Sep 17 00:00:00 2001 From: Emily Zeng Date: Thu, 9 Nov 2023 13:35:52 -0500 Subject: [PATCH 070/342] multiline with statement for readibility --- modules/ui.py | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/modules/ui.py b/modules/ui.py index 3eec7839f..bf06776e9 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -270,7 +270,9 @@ def create_ui(): extra_tabs.__enter__() with gr.Tab("Generation", id="txt2img_generation") as txt2img_generation_tab, ResizeHandleRow(equal_height=False): - with gr.Accordion("Open for Settings", open=False), gr.Column(variant='compact', elem_id="txt2img_settings") if shared.opts.img2img_settings_accordion else gr.Column(variant='compact', elem_id="txt2img_settings"): + with gr.Accordion("Open for Settings", open=False), gr.Column(variant='compact', elem_id="txt2img_settings") \ + if shared.opts.img2img_settings_accordion else gr.Column(variant='compact', elem_id="txt2img_settings"): + scripts.scripts_txt2img.prepare_ui() for category in ordered_ui_categories(): @@ -489,7 +491,9 @@ def create_ui(): extra_tabs.__enter__() with gr.Tab("Generation", id="img2img_generation") as img2img_generation_tab, ResizeHandleRow(equal_height=False): - with gr.Accordion("Open for Settings", open=False), gr.Column(variant='compact', elem_id="img2img_settings") if shared.opts.img2img_settings_accordion else gr.Column(variant='compact', elem_id="img2img_settings"): + with gr.Accordion("Open for Settings", open=False), gr.Column(variant='compact', elem_id="img2img_settings") \ + if shared.opts.img2img_settings_accordion else gr.Column(variant='compact', elem_id="img2img_settings"): + copy_image_buttons = [] copy_image_destinations = {} From 6d77a6e1c6b27ae82b2186cfc36cc4ad2a5e9ecf Mon Sep 17 00:00:00 2001 From: "fuchen.ljl" Date: Fri, 10 Nov 2023 14:40:39 +0800 Subject: [PATCH 072/342] Update README.md Modify the stablediffusion dependency address --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 4e0834400..1c97ecbb0 100644 --- a/README.md +++ b/README.md @@ -146,7 +146,7 @@ For the purposes of getting Google and other search engines to crawl the wiki, h ## Credits Licenses for borrowed code can be found in `Settings -> Licenses` screen, and also in `html/licenses.html` file. -- Stable Diffusion - https://github.com/CompVis/stable-diffusion, https://github.com/CompVis/taming-transformers +- Stable Diffusion - https://github.com/Stability-AI/stablediffusion, https://github.com/CompVis/taming-transformers - k-diffusion - https://github.com/crowsonkb/k-diffusion.git - GFPGAN - https://github.com/TencentARC/GFPGAN.git - CodeFormer - https://github.com/sczhou/CodeFormer From 66767e3876dde8d0ef27ce00254cd6b75332f036 Mon Sep 17 00:00:00 2001 From: "Alessandro de Oliveira Faria (A.K.A. CABELO)" Date: Fri, 10 Nov 2023 03:45:44 -0300 Subject: [PATCH 073/342] - opensuse compatibility --- README.md | 2 ++ 1 file changed, 2 insertions(+) diff --git a/README.md b/README.md index 4e0834400..89e54a613 100644 --- a/README.md +++ b/README.md @@ -121,6 +121,8 @@ Alternatively, use online services (like Google Colab): sudo apt install wget git python3 python3-venv libgl1 libglib2.0-0 # Red Hat-based: sudo dnf install wget git python3 +# openSUSE-based: +sudo zypper install wget git python3 # Arch-based: sudo pacman -S wget git python3 ``` From 7ff54005fee46ce188544db75c27de61ae279001 Mon Sep 17 00:00:00 2001 From: missionfloyd Date: Thu, 9 Nov 2023 23:47:53 -0700 Subject: [PATCH 074/342] Enable prompt hotkeys in style editor --- modules/ui_prompt_styles.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/modules/ui_prompt_styles.py b/modules/ui_prompt_styles.py index 85eb3a641..d6f8d4c76 100644 --- a/modules/ui_prompt_styles.py +++ b/modules/ui_prompt_styles.py @@ -64,10 +64,10 @@ class UiPromptStyles: self.materialize = ui_components.ToolButton(value=styles_materialize_symbol, elem_id=f"{tabname}_style_apply", tooltip="Apply all selected styles from the style selction dropdown in main UI to the prompt.") with gr.Row(): - self.prompt = gr.Textbox(label="Prompt", show_label=True, elem_id=f"{tabname}_edit_style_prompt", lines=3) + self.prompt = gr.Textbox(label="Prompt", show_label=True, elem_id=f"{tabname}_edit_style_prompt", lines=3, elem_classes=["prompt"]) with gr.Row(): - self.neg_prompt = gr.Textbox(label="Negative prompt", show_label=True, elem_id=f"{tabname}_edit_style_neg_prompt", lines=3) + self.neg_prompt = gr.Textbox(label="Negative prompt", show_label=True, elem_id=f"{tabname}_edit_style_neg_prompt", lines=3, elem_classes=["prompt"]) with gr.Row(): self.save = gr.Button('Save', variant='primary', elem_id=f'{tabname}_edit_style_save', visible=False) From 6a86b3ad9bc7bb9a58dc4228ecf93a3a511ed122 Mon Sep 17 00:00:00 2001 From: "Alessandro de Oliveira Faria (A.K.A. CABELO)" Date: Fri, 10 Nov 2023 14:15:34 -0300 Subject: [PATCH 075/342] Compatibility with Debian 11, Fedora 34+ and openSUSE 15.4+ --- README.md | 4 ++-- webui.sh | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/README.md b/README.md index 89e54a613..d4aa376b1 100644 --- a/README.md +++ b/README.md @@ -120,9 +120,9 @@ Alternatively, use online services (like Google Colab): # Debian-based: sudo apt install wget git python3 python3-venv libgl1 libglib2.0-0 # Red Hat-based: -sudo dnf install wget git python3 +sudo dnf install wget git python3 gperftools-libs libglvnd-glx # openSUSE-based: -sudo zypper install wget git python3 +sudo zypper install wget git python3 libtcmalloc4 libglvnd # Arch-based: sudo pacman -S wget git python3 ``` diff --git a/webui.sh b/webui.sh index 3d0f87eed..5c23c1d8f 100755 --- a/webui.sh +++ b/webui.sh @@ -87,7 +87,7 @@ delimiter="################################################################" printf "\n%s\n" "${delimiter}" printf "\e[1m\e[32mInstall script for stable-diffusion + Web UI\n" -printf "\e[1m\e[34mTested on Debian 11 (Bullseye)\e[0m" +printf "\e[1m\e[34mTested on Debian 11 (Bullseye), Fedora 34+ and openSUSE Leap 15.4 or newer.\e[0m" printf "\n%s\n" "${delimiter}" # Do not run as root @@ -222,7 +222,7 @@ fi # Try using TCMalloc on Linux prepare_tcmalloc() { if [[ "${OSTYPE}" == "linux"* ]] && [[ -z "${NO_TCMALLOC}" ]] && [[ -z "${LD_PRELOAD}" ]]; then - TCMALLOC="$(PATH=/usr/sbin:$PATH ldconfig -p | grep -Po "libtcmalloc(_minimal|)\.so\.\d" | head -n 1)" + TCMALLOC="$(PATH=/sbin:$PATH ldconfig -p | grep -Po "libtcmalloc(_minimal|)\.so\.\d" | head -n 1)" if [[ ! -z "${TCMALLOC}" ]]; then echo "Using TCMalloc: ${TCMALLOC}" export LD_PRELOAD="${TCMALLOC}" From 5432d9301359945b595d5e6649c7a64b4bb0bfca Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Sat, 11 Nov 2023 03:38:55 +0900 Subject: [PATCH 076/342] fix added accordion settings options --- modules/ui.py | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-) diff --git a/modules/ui.py b/modules/ui.py index bf06776e9..f28de3543 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -4,6 +4,7 @@ import os import sys from functools import reduce import warnings +from contextlib import suppress import gradio as gr import gradio.utils @@ -270,9 +271,7 @@ def create_ui(): extra_tabs.__enter__() with gr.Tab("Generation", id="txt2img_generation") as txt2img_generation_tab, ResizeHandleRow(equal_height=False): - with gr.Accordion("Open for Settings", open=False), gr.Column(variant='compact', elem_id="txt2img_settings") \ - if shared.opts.img2img_settings_accordion else gr.Column(variant='compact', elem_id="txt2img_settings"): - + with gr.Accordion("Open for Settings", open=False) if shared.opts.txt2img_settings_accordion else suppress(), gr.Column(variant='compact', elem_id="txt2img_settings"): scripts.scripts_txt2img.prepare_ui() for category in ordered_ui_categories(): @@ -491,8 +490,7 @@ def create_ui(): extra_tabs.__enter__() with gr.Tab("Generation", id="img2img_generation") as img2img_generation_tab, ResizeHandleRow(equal_height=False): - with gr.Accordion("Open for Settings", open=False), gr.Column(variant='compact', elem_id="img2img_settings") \ - if shared.opts.img2img_settings_accordion else gr.Column(variant='compact', elem_id="img2img_settings"): + with gr.Accordion("Open for Settings", open=False) if shared.opts.img2img_settings_accordion else suppress(), gr.Column(variant='compact', elem_id="img2img_settings"): copy_image_buttons = [] copy_image_destinations = {} From 3a4a6c43a4ca31056d5c09bb54e3eef24e6cf864 Mon Sep 17 00:00:00 2001 From: Emily Zeng Date: Fri, 10 Nov 2023 16:06:01 -0500 Subject: [PATCH 077/342] ExitStack as alternative to suppress --- modules/ui.py | 13 ++++++++++--- 1 file changed, 10 insertions(+), 3 deletions(-) diff --git a/modules/ui.py b/modules/ui.py index f28de3543..ba0d8542b 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -4,7 +4,7 @@ import os import sys from functools import reduce import warnings -from contextlib import suppress +from contextlib import ExitStack import gradio as gr import gradio.utils @@ -271,7 +271,11 @@ def create_ui(): extra_tabs.__enter__() with gr.Tab("Generation", id="txt2img_generation") as txt2img_generation_tab, ResizeHandleRow(equal_height=False): - with gr.Accordion("Open for Settings", open=False) if shared.opts.txt2img_settings_accordion else suppress(), gr.Column(variant='compact', elem_id="txt2img_settings"): + with ExitStack() as stack: + if shared.opts.txt2img_settings_accordion: + stack.enter_context(gr.Accordion("Open for Settings", open=False)) + stack.enter_context(gr.Column(variant='compact', elem_id="txt2img_settings")) + scripts.scripts_txt2img.prepare_ui() for category in ordered_ui_categories(): @@ -490,7 +494,10 @@ def create_ui(): extra_tabs.__enter__() with gr.Tab("Generation", id="img2img_generation") as img2img_generation_tab, ResizeHandleRow(equal_height=False): - with gr.Accordion("Open for Settings", open=False) if shared.opts.img2img_settings_accordion else suppress(), gr.Column(variant='compact', elem_id="img2img_settings"): + with ExitStack() as stack: + if shared.opts.img2img_settings_accordion: + stack.enter_context(gr.Accordion("Open for Settings", open=False)) + stack.enter_context(gr.Column(variant='compact', elem_id="img2img_settings")) copy_image_buttons = [] copy_image_destinations = {} From 0fc7dc1c04a046d95588651ffc4e71a7d40378d3 Mon Sep 17 00:00:00 2001 From: wfjsw Date: Sat, 11 Nov 2023 04:01:13 -0600 Subject: [PATCH 078/342] implementing script metadata and DAG sorting mechanism --- modules/extensions.py | 80 +++++++++++++++++++++--- modules/scripts.py | 139 +++++++++++++++++++++++++++++++++++++----- 2 files changed, 196 insertions(+), 23 deletions(-) diff --git a/modules/extensions.py b/modules/extensions.py index bf9a1878f..e317a4041 100644 --- a/modules/extensions.py +++ b/modules/extensions.py @@ -1,3 +1,5 @@ +import configparser +import functools import os import threading @@ -23,8 +25,9 @@ class Extension: lock = threading.Lock() cached_fields = ['remote', 'commit_date', 'branch', 'commit_hash', 'version'] - def __init__(self, name, path, enabled=True, is_builtin=False): + def __init__(self, name, path, enabled=True, is_builtin=False, canonical_name=None): self.name = name + self.canonical_name = canonical_name or name.lower() self.path = path self.enabled = enabled self.status = '' @@ -37,6 +40,17 @@ class Extension: self.remote = None self.have_info_from_repo = False + @functools.cached_property + def metadata(self): + if os.path.isfile(os.path.join(self.path, "sd_webui_metadata.ini")): + try: + config = configparser.ConfigParser() + config.read(os.path.join(self.path, "sd_webui_metadata.ini")) + return config + except Exception: + errors.report(f"Error reading sd_webui_metadata.ini for extension {self.canonical_name}.", exc_info=True) + return None + def to_dict(self): return {x: getattr(self, x) for x in self.cached_fields} @@ -136,9 +150,6 @@ class Extension: def list_extensions(): extensions.clear() - if not os.path.isdir(extensions_dir): - return - if shared.cmd_opts.disable_all_extensions: print("*** \"--disable-all-extensions\" arg was used, will not load any extensions ***") elif shared.opts.disable_all_extensions == "all": @@ -148,18 +159,69 @@ def list_extensions(): elif shared.opts.disable_all_extensions == "extra": print("*** \"Disable all extensions\" option was set, will only load built-in extensions ***") - extension_paths = [] + extension_dependency_map = {} + + # scan through extensions directory and load metadata for dirname in [extensions_dir, extensions_builtin_dir]: if not os.path.isdir(dirname): - return + continue for extension_dirname in sorted(os.listdir(dirname)): path = os.path.join(dirname, extension_dirname) if not os.path.isdir(path): continue - extension_paths.append((extension_dirname, path, dirname == extensions_builtin_dir)) + canonical_name = extension_dirname + requires = None - for dirname, path, is_builtin in extension_paths: - extension = Extension(name=dirname, path=path, enabled=dirname not in shared.opts.disabled_extensions, is_builtin=is_builtin) + if os.path.isfile(os.path.join(path, "sd_webui_metadata.ini")): + try: + config = configparser.ConfigParser() + config.read(os.path.join(path, "sd_webui_metadata.ini")) + canonical_name = config.get("Extension", "Name", fallback=canonical_name) + requires = config.get("Extension", "Requires", fallback=None) + continue + except Exception: + errors.report(f"Error reading sd_webui_metadata.ini for extension {extension_dirname}. " + f"Will load regardless.", exc_info=True) + + canonical_name = canonical_name.lower().strip() + + # check for duplicated canonical names + if canonical_name in extension_dependency_map: + errors.report(f"Duplicate canonical name \"{canonical_name}\" found in extensions " + f"\"{extension_dirname}\" and \"{extension_dependency_map[canonical_name]['dirname']}\". " + f"The current loading extension will be discarded.", exc_info=False) + continue + + # we want to wash the data to lowercase and remove whitespaces just in case + requires = [x.strip() for x in requires.lower().split(',')] if requires else [] + + extension_dependency_map[canonical_name] = { + "dirname": extension_dirname, + "path": path, + "requires": requires, + } + + # check for requirements + for (_, extension_data) in extension_dependency_map.items(): + dirname, path, requires = extension_data['dirname'], extension_data['path'], extension_data['requires'] + requirement_met = True + for req in requires: + if req not in extension_dependency_map: + errors.report(f"Extension \"{dirname}\" requires \"{req}\" which is not installed. " + f"The current loading extension will be discarded.", exc_info=False) + requirement_met = False + break + dep_dirname = extension_dependency_map[req]['dirname'] + if dep_dirname in shared.opts.disabled_extensions: + errors.report(f"Extension \"{dirname}\" requires \"{dep_dirname}\" which is disabled. " + f"The current loading extension will be discarded.", exc_info=False) + requirement_met = False + break + + is_builtin = dirname == extensions_builtin_dir + extension = Extension(name=dirname, path=path, + enabled=dirname not in shared.opts.disabled_extensions and requirement_met, + is_builtin=is_builtin) extensions.append(extension) diff --git a/modules/scripts.py b/modules/scripts.py index 5c6e0226e..e92a34a0e 100644 --- a/modules/scripts.py +++ b/modules/scripts.py @@ -2,6 +2,7 @@ import os import re import sys import inspect +from graphlib import TopologicalSorter, CycleError from collections import namedtuple from dataclasses import dataclass @@ -314,15 +315,131 @@ ScriptClassData = namedtuple("ScriptClassData", ["script_class", "path", "basedi def list_scripts(scriptdirname, extension, *, include_extensions=True): scripts_list = [] + script_dependency_map = {} - basedir = os.path.join(paths.script_path, scriptdirname) - if os.path.exists(basedir): - for filename in sorted(os.listdir(basedir)): - scripts_list.append(ScriptFile(paths.script_path, filename, os.path.join(basedir, filename))) + # build script dependency map + + root_script_basedir = os.path.join(paths.script_path, scriptdirname) + if os.path.exists(root_script_basedir): + for filename in sorted(os.listdir(root_script_basedir)): + script_dependency_map[filename] = { + "extension": None, + "extension_dirname": None, + "script_file": ScriptFile(paths.script_path, filename, os.path.join(root_script_basedir, filename)), + "requires": [], + "load_before": [], + "load_after": [], + } if include_extensions: for ext in extensions.active(): - scripts_list += ext.list_files(scriptdirname, extension) + extension_scripts_list = ext.list_files(scriptdirname, extension) + for extension_script in extension_scripts_list: + # this is built on the assumption that script name is unique. + # I think bad thing is gonna happen if name collide in the current implementation anyway, but we + # will need to refactor here if this assumption is broken later on. + if extension_script.filename in script_dependency_map: + errors.report(f"Duplicate script name \"{extension_script.filename}\" found in extensions " + f"\"{ext.name}\" and \"{script_dependency_map[extension_script.filename]['extension_dirname'] or 'builtin'}\". " + f"The current loading file will be discarded.", exc_info=False) + continue + + relative_path = scriptdirname + "/" + extension_script.filename + + requires = None + load_before = None + load_after = None + + if ext.metadata is not None: + requires = ext.metadata.get(relative_path, "Requires", fallback=None) + load_before = ext.metadata.get(relative_path, "Before", fallback=None) + load_after = ext.metadata.get(relative_path, "After", fallback=None) + + requires = [x.strip() for x in requires.split(',')] if requires else [] + load_after = [x.strip() for x in load_after.split(',')] if load_after else [] + load_before = [x.strip() for x in load_before.split(',')] if load_before else [] + + script_dependency_map[extension_script.filename] = { + "extension": ext.canonical_name, + "extension_dirname": ext.name, + "script_file": extension_script, + "requires": requires, + "load_before": load_before, + "load_after": load_after, + } + + # resolve dependencies + + loaded_extensions = set() + for _, script_data in script_dependency_map.items(): + if script_data['extension'] is not None: + loaded_extensions.add(script_data['extension']) + + for script_filename, script_data in script_dependency_map.items(): + # load before requires inverse dependency + # in this case, append the script name into the load_after list of the specified script + for load_before_script in script_data['load_before']: + if load_before_script.startswith('ext:'): + # if this requires an extension to be loaded before + required_extension = load_before_script[4:] + for _, script_data2 in script_dependency_map.items(): + if script_data2['extension'] == required_extension: + script_data2['load_after'].append(script_filename) + break + else: + # if this requires an individual script to be loaded before + if load_before_script in script_dependency_map: + script_dependency_map[load_before_script]['load_after'].append(script_filename) + + # resolve extension name in load_after lists + for load_after_script in script_data['load_after']: + if load_after_script.startswith('ext:'): + # if this requires an extension to be loaded after + required_extension = load_after_script[4:] + for script_file_name2, script_data2 in script_dependency_map.items(): + if script_data2['extension'] == required_extension: + script_data['load_after'].append(script_file_name2) + + # remove all extension names in load_after lists + script_data['load_after'] = [x for x in script_data['load_after'] if not x.startswith('ext:')] + + # build the DAG + sorter = TopologicalSorter() + for script_filename, script_data in script_dependency_map.items(): + requirement_met = True + for required_script in script_data['requires']: + if required_script.startswith('ext:'): + # if this requires an extension to be installed + required_extension = required_script[4:] + if required_extension not in loaded_extensions: + errors.report(f"Script \"{script_filename}\" requires extension \"{required_extension}\" to " + f"be loaded, but it is not. Skipping.", + exc_info=False) + requirement_met = False + break + else: + # if this requires an individual script to be loaded + if required_script not in script_dependency_map: + errors.report(f"Script \"{script_filename}\" requires script \"{required_script}\" to " + f"be loaded, but it is not. Skipping.", + exc_info=False) + requirement_met = False + break + if not requirement_met: + continue + + sorter.add(script_filename, *script_data['load_after']) + + # sort the scripts + try: + ordered_script = sorter.static_order() + except CycleError: + errors.report("Cycle detected in script dependencies. Scripts will load in ascending order.", exc_info=True) + ordered_script = script_dependency_map.keys() + + for script_filename in ordered_script: + script_data = script_dependency_map[script_filename] + scripts_list.append(script_data['script_file']) scripts_list = [x for x in scripts_list if os.path.splitext(x.path)[1].lower() == extension and os.path.isfile(x.path)] @@ -365,15 +482,9 @@ def load_scripts(): elif issubclass(script_class, scripts_postprocessing.ScriptPostprocessing): postprocessing_scripts_data.append(ScriptClassData(script_class, scriptfile.path, scriptfile.basedir, module)) - def orderby(basedir): - # 1st webui, 2nd extensions-builtin, 3rd extensions - priority = {os.path.join(paths.script_path, "extensions-builtin"):1, paths.script_path:0} - for key in priority: - if basedir.startswith(key): - return priority[key] - return 9999 - - for scriptfile in sorted(scripts_list, key=lambda x: [orderby(x.basedir), x]): + # here the scripts_list is already ordered + # processing_script is not considered though + for scriptfile in scripts_list: try: if scriptfile.basedir != paths.script_path: sys.path = [scriptfile.basedir] + sys.path From 0d1924c48be3d02650e87b12a4f53165a8b4a599 Mon Sep 17 00:00:00 2001 From: wfjsw Date: Sat, 11 Nov 2023 04:03:55 -0600 Subject: [PATCH 079/342] populate loaded_extensions from extension list instead --- modules/scripts.py | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/modules/scripts.py b/modules/scripts.py index e92a34a0e..7cdf288df 100644 --- a/modules/scripts.py +++ b/modules/scripts.py @@ -371,9 +371,8 @@ def list_scripts(scriptdirname, extension, *, include_extensions=True): # resolve dependencies loaded_extensions = set() - for _, script_data in script_dependency_map.items(): - if script_data['extension'] is not None: - loaded_extensions.add(script_data['extension']) + for ext in extensions.active(): + loaded_extensions.add(ext.canonical_name) for script_filename, script_data in script_dependency_map.items(): # load before requires inverse dependency From bc1a450124ab643fc0c3ea7630d875afb4b84b84 Mon Sep 17 00:00:00 2001 From: wfjsw Date: Sat, 11 Nov 2023 04:08:45 -0600 Subject: [PATCH 080/342] reverse the extension load order so builtin extensions load earlier natively --- modules/extensions.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/extensions.py b/modules/extensions.py index e317a4041..7583a3b03 100644 --- a/modules/extensions.py +++ b/modules/extensions.py @@ -162,7 +162,7 @@ def list_extensions(): extension_dependency_map = {} # scan through extensions directory and load metadata - for dirname in [extensions_dir, extensions_builtin_dir]: + for dirname in [extensions_builtin_dir, extensions_dir]: if not os.path.isdir(dirname): continue From 294f8a514f982248cda1cafda30d35566f3a0321 Mon Sep 17 00:00:00 2001 From: aria1th <35677394+aria1th@users.noreply.github.com> Date: Sat, 11 Nov 2023 23:28:12 +0900 Subject: [PATCH 081/342] add hyperTile https://github.com/tfernd/HyperTile --- modules/processing.py | 27 ++++++++++++++++++++++++--- modules/shared_options.py | 2 ++ 2 files changed, 26 insertions(+), 3 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index b0e240a46..e23095343 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -799,6 +799,16 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: infotexts = [] output_images = [] + unet_object = p.sd_model.model + vae_model = p.sd_model.first_stage_model + try: + from hyper_tile import split_attention, flush + except (ImportError, ModuleNotFoundError): # pip install git+https://github.com/tfernd/HyperTile@2ef64b2800d007d305755c33550537410310d7df + split_attention = lambda *args, **kwargs: lambda x: x # return a no-op context manager + flush = lambda: None + import random + saved_rng_state = random.getstate() + random.seed(p.seed) # hyper_tile uses random, so we need to seed it with torch.no_grad(), p.sd_model.ema_scope(): with devices.autocast(): @@ -866,15 +876,25 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: shared.state.job = f"Batch {n+1} out of {p.n_iter}" with devices.without_autocast() if devices.unet_needs_upcast else devices.autocast(): - samples_ddim = p.sample(conditioning=p.c, unconditional_conditioning=p.uc, seeds=p.seeds, subseeds=p.subseeds, subseed_strength=p.subseed_strength, prompts=p.prompts) + # get largest tile size available, which is 2^x which is factor of gcd of p.width and p.height + gcd = math.gcd(p.width, p.height) + largest_tile_size_available = 1 + while gcd % (largest_tile_size_available * 2) == 0: + largest_tile_size_available *= 2 + aspect_ratio = p.width / p.height + with split_attention(vae_model, aspect_ratio=aspect_ratio, tile_size=min(largest_tile_size_available, 128), disable=not shared.opts.hypertile_split_vae_attn): + with split_attention(unet_object, aspect_ratio=aspect_ratio, tile_size=min(largest_tile_size_available, 256), swap_size=2, disable=not shared.opts.hypertile_split_unet_attn): + flush() + samples_ddim = p.sample(conditioning=p.c, unconditional_conditioning=p.uc, seeds=p.seeds, subseeds=p.subseeds, subseed_strength=p.subseed_strength, prompts=p.prompts) if getattr(samples_ddim, 'already_decoded', False): x_samples_ddim = samples_ddim else: if opts.sd_vae_decode_method != 'Full': p.extra_generation_params['VAE Decoder'] = opts.sd_vae_decode_method - - x_samples_ddim = decode_latent_batch(p.sd_model, samples_ddim, target_device=devices.cpu, check_for_nans=True) + with split_attention(vae_model, aspect_ratio=aspect_ratio, tile_size=min(largest_tile_size_available, 128), disable=not shared.opts.hypertile_split_vae_attn): + flush() + x_samples_ddim = decode_latent_batch(p.sd_model, samples_ddim, target_device=devices.cpu, check_for_nans=True) x_samples_ddim = torch.stack(x_samples_ddim).float() x_samples_ddim = torch.clamp((x_samples_ddim + 1.0) / 2.0, min=0.0, max=1.0) @@ -980,6 +1000,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: if opts.grid_save: images.save_image(grid, p.outpath_grids, "grid", p.all_seeds[0], p.all_prompts[0], opts.grid_format, info=infotext(use_main_prompt=True), short_filename=not opts.grid_extended_filename, p=p, grid=True) + random.setstate(saved_rng_state) if not p.disable_extra_networks and p.extra_network_data: extra_networks.deactivate(p, p.extra_network_data) diff --git a/modules/shared_options.py b/modules/shared_options.py index d40db5306..d96502656 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -200,6 +200,8 @@ options_templates.update(options_section(('optimizations', "Optimizations"), { "pad_cond_uncond": OptionInfo(False, "Pad prompt/negative prompt to be same length", infotext='Pad conds').info("improves performance when prompt and negative prompt have different lengths; changes seeds"), "persistent_cond_cache": OptionInfo(True, "Persistent cond cache").info("do not recalculate conds from prompts if prompts have not changed since previous calculation"), "batch_cond_uncond": OptionInfo(True, "Batch cond/uncond").info("do both conditional and unconditional denoising in one batch; uses a bit more VRAM during sampling, but improves speed; previously this was controlled by --always-batch-cond-uncond comandline argument"), + "hypertile_split_unet_attn" : OptionInfo(False, "Split attention in Unet with HyperTile").link("Github", "https://github.com/tfernd/HyperTile").info("improves performance; changes behavior, but deterministic"), + "hypertile_split_vae_attn": OptionInfo(False, "Split attention in VAE with HyperTile").link("Github", "https://github.com/tfernd/HyperTile").info("improves performance; changes behavior, but deterministic"), })) options_templates.update(options_section(('compatibility', "Compatibility"), { From 7af576e745c79a9539e40bc158e695192ae79f25 Mon Sep 17 00:00:00 2001 From: wfjsw Date: Sat, 11 Nov 2023 10:46:47 -0600 Subject: [PATCH 082/342] remove the assumption of same name --- modules/scripts.py | 81 +++++++++++++++++----------------------------- 1 file changed, 30 insertions(+), 51 deletions(-) diff --git a/modules/scripts.py b/modules/scripts.py index 7cdf288df..7ad222451 100644 --- a/modules/scripts.py +++ b/modules/scripts.py @@ -335,15 +335,9 @@ def list_scripts(scriptdirname, extension, *, include_extensions=True): for ext in extensions.active(): extension_scripts_list = ext.list_files(scriptdirname, extension) for extension_script in extension_scripts_list: - # this is built on the assumption that script name is unique. - # I think bad thing is gonna happen if name collide in the current implementation anyway, but we - # will need to refactor here if this assumption is broken later on. - if extension_script.filename in script_dependency_map: - errors.report(f"Duplicate script name \"{extension_script.filename}\" found in extensions " - f"\"{ext.name}\" and \"{script_dependency_map[extension_script.filename]['extension_dirname'] or 'builtin'}\". " - f"The current loading file will be discarded.", exc_info=False) - continue - + script_canonical_name = ext.canonical_name + "/" + extension_script.filename + if ext.is_builtin: + script_canonical_name = "builtin/" + script_canonical_name relative_path = scriptdirname + "/" + extension_script.filename requires = None @@ -359,7 +353,7 @@ def list_scripts(scriptdirname, extension, *, include_extensions=True): load_after = [x.strip() for x in load_after.split(',')] if load_after else [] load_before = [x.strip() for x in load_before.split(',')] if load_before else [] - script_dependency_map[extension_script.filename] = { + script_dependency_map[script_canonical_name] = { "extension": ext.canonical_name, "extension_dirname": ext.name, "script_file": extension_script, @@ -374,60 +368,45 @@ def list_scripts(scriptdirname, extension, *, include_extensions=True): for ext in extensions.active(): loaded_extensions.add(ext.canonical_name) - for script_filename, script_data in script_dependency_map.items(): + for script_canonical_name, script_data in script_dependency_map.items(): # load before requires inverse dependency # in this case, append the script name into the load_after list of the specified script for load_before_script in script_data['load_before']: - if load_before_script.startswith('ext:'): - # if this requires an extension to be loaded before - required_extension = load_before_script[4:] + # if this requires an individual script to be loaded before + if load_before_script in script_dependency_map: + script_dependency_map[load_before_script]['load_after'].append(script_canonical_name) + elif load_before_script in loaded_extensions: for _, script_data2 in script_dependency_map.items(): - if script_data2['extension'] == required_extension: - script_data2['load_after'].append(script_filename) + if script_data2['extension'] == load_before_script: + script_data2['load_after'].append(script_canonical_name) break - else: - # if this requires an individual script to be loaded before - if load_before_script in script_dependency_map: - script_dependency_map[load_before_script]['load_after'].append(script_filename) # resolve extension name in load_after lists - for load_after_script in script_data['load_after']: - if load_after_script.startswith('ext:'): - # if this requires an extension to be loaded after - required_extension = load_after_script[4:] - for script_file_name2, script_data2 in script_dependency_map.items(): - if script_data2['extension'] == required_extension: - script_data['load_after'].append(script_file_name2) - - # remove all extension names in load_after lists - script_data['load_after'] = [x for x in script_data['load_after'] if not x.startswith('ext:')] + for load_after_script in list(script_data['load_after']): + if load_after_script not in script_dependency_map and load_after_script in loaded_extensions: + script_data['load_after'].remove(load_after_script) + for script_canonical_name2, script_data2 in script_dependency_map.items(): + if script_data2['extension'] == load_after_script: + script_data['load_after'].remove(script_canonical_name2) + break # build the DAG sorter = TopologicalSorter() - for script_filename, script_data in script_dependency_map.items(): + for script_canonical_name, script_data in script_dependency_map.items(): requirement_met = True for required_script in script_data['requires']: - if required_script.startswith('ext:'): - # if this requires an extension to be installed - required_extension = required_script[4:] - if required_extension not in loaded_extensions: - errors.report(f"Script \"{script_filename}\" requires extension \"{required_extension}\" to " - f"be loaded, but it is not. Skipping.", - exc_info=False) - requirement_met = False - break - else: - # if this requires an individual script to be loaded - if required_script not in script_dependency_map: - errors.report(f"Script \"{script_filename}\" requires script \"{required_script}\" to " - f"be loaded, but it is not. Skipping.", - exc_info=False) - requirement_met = False - break + # if this requires an individual script to be loaded + if required_script not in script_dependency_map and required_script not in loaded_extensions: + errors.report(f"Script \"{script_canonical_name}\" " + f"requires \"{required_script}\" to " + f"be loaded, but it is not. Skipping.", + exc_info=False) + requirement_met = False + break if not requirement_met: continue - sorter.add(script_filename, *script_data['load_after']) + sorter.add(script_canonical_name, *script_data['load_after']) # sort the scripts try: @@ -436,8 +415,8 @@ def list_scripts(scriptdirname, extension, *, include_extensions=True): errors.report("Cycle detected in script dependencies. Scripts will load in ascending order.", exc_info=True) ordered_script = script_dependency_map.keys() - for script_filename in ordered_script: - script_data = script_dependency_map[script_filename] + for script_canonical_name in ordered_script: + script_data = script_dependency_map[script_canonical_name] scripts_list.append(script_data['script_file']) scripts_list = [x for x in scripts_list if os.path.splitext(x.path)[1].lower() == extension and os.path.isfile(x.path)] From 520e52f846892cc2b207b738b4180fa863c7b38f Mon Sep 17 00:00:00 2001 From: wfjsw Date: Sat, 11 Nov 2023 10:58:26 -0600 Subject: [PATCH 083/342] allow comma and whitespace as separator --- modules/extensions.py | 9 ++++++--- modules/scripts.py | 6 +++--- 2 files changed, 9 insertions(+), 6 deletions(-) diff --git a/modules/extensions.py b/modules/extensions.py index 7583a3b03..795af996e 100644 --- a/modules/extensions.py +++ b/modules/extensions.py @@ -2,6 +2,7 @@ import configparser import functools import os import threading +import re from modules import shared, errors, cache, scripts from modules.gitpython_hack import Repo @@ -48,7 +49,8 @@ class Extension: config.read(os.path.join(self.path, "sd_webui_metadata.ini")) return config except Exception: - errors.report(f"Error reading sd_webui_metadata.ini for extension {self.canonical_name}.", exc_info=True) + errors.report(f"Error reading sd_webui_metadata.ini for extension {self.canonical_name}.", + exc_info=True) return None def to_dict(self): @@ -70,6 +72,7 @@ class Extension: self.do_read_info_from_repo() return self.to_dict() + try: d = cache.cached_data_for_file('extensions-git', self.name, os.path.join(self.path, ".git"), read_from_repo) self.from_dict(d) @@ -194,8 +197,8 @@ def list_extensions(): f"The current loading extension will be discarded.", exc_info=False) continue - # we want to wash the data to lowercase and remove whitespaces just in case - requires = [x.strip() for x in requires.lower().split(',')] if requires else [] + # both "," and " " are accepted as separator + requires = list(filter(None, re.split(r"[,\s]+", requires.lower()))) if requires else [] extension_dependency_map[canonical_name] = { "dirname": extension_dirname, diff --git a/modules/scripts.py b/modules/scripts.py index 7ad222451..5dd0555dd 100644 --- a/modules/scripts.py +++ b/modules/scripts.py @@ -349,9 +349,9 @@ def list_scripts(scriptdirname, extension, *, include_extensions=True): load_before = ext.metadata.get(relative_path, "Before", fallback=None) load_after = ext.metadata.get(relative_path, "After", fallback=None) - requires = [x.strip() for x in requires.split(',')] if requires else [] - load_after = [x.strip() for x in load_after.split(',')] if load_after else [] - load_before = [x.strip() for x in load_before.split(',')] if load_before else [] + requires = list(filter(None, re.split(r"[,\s]+", requires.lower()))) if requires else [] + load_after = list(filter(None, re.split(r"[,\s]+", load_after.lower()))) if load_after else [] + load_before = list(filter(None, re.split(r"[,\s]+", load_before.lower()))) if load_before else [] script_dependency_map[script_canonical_name] = { "extension": ext.canonical_name, From 48d6102b3105bb0179c8eab14ec7930945aca326 Mon Sep 17 00:00:00 2001 From: wfjsw Date: Sat, 11 Nov 2023 11:17:26 -0600 Subject: [PATCH 084/342] fix --- modules/extensions.py | 1 - 1 file changed, 1 deletion(-) diff --git a/modules/extensions.py b/modules/extensions.py index 795af996e..5536db3ea 100644 --- a/modules/extensions.py +++ b/modules/extensions.py @@ -183,7 +183,6 @@ def list_extensions(): config.read(os.path.join(path, "sd_webui_metadata.ini")) canonical_name = config.get("Extension", "Name", fallback=canonical_name) requires = config.get("Extension", "Requires", fallback=None) - continue except Exception: errors.report(f"Error reading sd_webui_metadata.ini for extension {extension_dirname}. " f"Will load regardless.", exc_info=True) From 3bb32befe9523a6acefbab7fe099f91660f41ea9 Mon Sep 17 00:00:00 2001 From: wfjsw Date: Sat, 11 Nov 2023 11:58:19 -0600 Subject: [PATCH 085/342] bug fix --- modules/scripts.py | 25 ++++++++++++++++++------- 1 file changed, 18 insertions(+), 7 deletions(-) diff --git a/modules/scripts.py b/modules/scripts.py index 5dd0555dd..b1f4504a5 100644 --- a/modules/scripts.py +++ b/modules/scripts.py @@ -322,6 +322,9 @@ def list_scripts(scriptdirname, extension, *, include_extensions=True): root_script_basedir = os.path.join(paths.script_path, scriptdirname) if os.path.exists(root_script_basedir): for filename in sorted(os.listdir(root_script_basedir)): + if not os.path.isfile(os.path.join(root_script_basedir, filename)): + continue + script_dependency_map[filename] = { "extension": None, "extension_dirname": None, @@ -335,19 +338,27 @@ def list_scripts(scriptdirname, extension, *, include_extensions=True): for ext in extensions.active(): extension_scripts_list = ext.list_files(scriptdirname, extension) for extension_script in extension_scripts_list: + if not os.path.isfile(extension_script.path): + continue + script_canonical_name = ext.canonical_name + "/" + extension_script.filename if ext.is_builtin: script_canonical_name = "builtin/" + script_canonical_name relative_path = scriptdirname + "/" + extension_script.filename - requires = None - load_before = None - load_after = None + requires = '' + load_before = '' + load_after = '' if ext.metadata is not None: - requires = ext.metadata.get(relative_path, "Requires", fallback=None) - load_before = ext.metadata.get(relative_path, "Before", fallback=None) - load_after = ext.metadata.get(relative_path, "After", fallback=None) + requires = ext.metadata.get(relative_path, "Requires", fallback='') + load_before = ext.metadata.get(relative_path, "Before", fallback='') + load_after = ext.metadata.get(relative_path, "After", fallback='') + + # propagate directory level metadata + requires = requires + ',' + ext.metadata.get(scriptdirname, "Requires", fallback='') + load_before = load_before + ',' + ext.metadata.get(scriptdirname, "Before", fallback='') + load_after = load_after + ',' + ext.metadata.get(scriptdirname, "After", fallback='') requires = list(filter(None, re.split(r"[,\s]+", requires.lower()))) if requires else [] load_after = list(filter(None, re.split(r"[,\s]+", load_after.lower()))) if load_after else [] @@ -387,7 +398,7 @@ def list_scripts(scriptdirname, extension, *, include_extensions=True): script_data['load_after'].remove(load_after_script) for script_canonical_name2, script_data2 in script_dependency_map.items(): if script_data2['extension'] == load_after_script: - script_data['load_after'].remove(script_canonical_name2) + script_data['load_after'].append(script_canonical_name2) break # build the DAG From f6762d2ad95e3de39fc900b3fd528310e512831f Mon Sep 17 00:00:00 2001 From: Tom Haelbich <65122811+h43lb1t0@users.noreply.github.com> Date: Sun, 12 Nov 2023 14:14:16 +0100 Subject: [PATCH 086/342] dir buttons start with / so only the correct dir will be shown and not dirs with a substrings as name from the dir --- modules/ui_extra_networks.py | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/modules/ui_extra_networks.py b/modules/ui_extra_networks.py index 063bd7b80..43a94b74b 100644 --- a/modules/ui_extra_networks.py +++ b/modules/ui_extra_networks.py @@ -138,8 +138,9 @@ class ExtraNetworksPage: continue subdir = os.path.abspath(x)[len(parentdir):].replace("\\", "/") - while subdir.startswith("/"): - subdir = subdir[1:] + + if not subdir.startswith("/"): + subdir = "/" + subdir is_empty = len(os.listdir(x)) == 0 if not is_empty and not subdir.endswith("/"): From 8048f36072c8a281b8c8c79235df63a748ab7361 Mon Sep 17 00:00:00 2001 From: missionfloyd Date: Sun, 12 Nov 2023 17:12:50 -0700 Subject: [PATCH 087/342] Lint --- modules/ui_extra_networks.py | 1 - 1 file changed, 1 deletion(-) diff --git a/modules/ui_extra_networks.py b/modules/ui_extra_networks.py index 43a94b74b..bd6732856 100644 --- a/modules/ui_extra_networks.py +++ b/modules/ui_extra_networks.py @@ -138,7 +138,6 @@ class ExtraNetworksPage: continue subdir = os.path.abspath(x)[len(parentdir):].replace("\\", "/") - if not subdir.startswith("/"): subdir = "/" + subdir From 94e966956666ba13b368aaf781628085e3d4f7e3 Mon Sep 17 00:00:00 2001 From: kaalibro Date: Mon, 13 Nov 2023 14:51:06 +0600 Subject: [PATCH 088/342] Fixes generation restart not working for some users when 'Ctrl+Enter' is pressed --- script.js | 15 ++++++++++++--- 1 file changed, 12 insertions(+), 3 deletions(-) diff --git a/script.js b/script.js index 5f6ee3542..c0e678ea7 100644 --- a/script.js +++ b/script.js @@ -133,9 +133,18 @@ document.addEventListener('keydown', function(e) { if (isEnter && isModifierKey) { if (interruptButton.style.display === 'block') { interruptButton.click(); - setTimeout(function() { - generateButton.click(); - }, 500); + const callback = (mutationList) => { + for (const mutation of mutationList) { + if (mutation.type === 'attributes' && mutation.attributeName === 'style') { + if (interruptButton.style.display === 'none') { + generateButton.click(); + observer.disconnect(); + } + } + } + }; + const observer = new MutationObserver(callback); + observer.observe(interruptButton, {attributes: true}); } else { generateButton.click(); } From c1c816006e47f3b7dcf1512594fd31817242e7fa Mon Sep 17 00:00:00 2001 From: kaalibro Date: Mon, 13 Nov 2023 22:01:52 +0600 Subject: [PATCH 089/342] Adds 'Path' sorting for Extra network cards --- modules/shared_options.py | 2 +- modules/ui_extra_networks.py | 3 ++- 2 files changed, 3 insertions(+), 2 deletions(-) diff --git a/modules/shared_options.py b/modules/shared_options.py index d40db5306..8fc7ef1d2 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -235,7 +235,7 @@ options_templates.update(options_section(('extra_networks', "Extra Networks"), { "extra_networks_card_height": OptionInfo(0, "Card height for Extra Networks").info("in pixels"), "extra_networks_card_text_scale": OptionInfo(1.0, "Card text scale", gr.Slider, {"minimum": 0.0, "maximum": 2.0, "step": 0.01}).info("1 = original size"), "extra_networks_card_show_desc": OptionInfo(True, "Show description on card"), - "extra_networks_card_order_field": OptionInfo("Name", "Default order field for Extra Networks cards", gr.Dropdown, {"choices": ['Name', 'Date Created', 'Date Modified']}).needs_reload_ui(), + "extra_networks_card_order_field": OptionInfo("Path", "Default order field for Extra Networks cards", gr.Dropdown, {"choices": ['Path', 'Name', 'Date Created', 'Date Modified']}).needs_reload_ui(), "extra_networks_card_order": OptionInfo("Ascending", "Default order for Extra Networks cards", gr.Dropdown, {"choices": ['Ascending', 'Descending']}).needs_reload_ui(), "extra_networks_add_text_separator": OptionInfo(" ", "Extra networks separator").info("extra text to add before <...> when adding extra network to prompt"), "ui_extra_networks_tab_reorder": OptionInfo("", "Extra networks tab order").needs_reload_ui(), diff --git a/modules/ui_extra_networks.py b/modules/ui_extra_networks.py index 7907cd63f..f03e20337 100644 --- a/modules/ui_extra_networks.py +++ b/modules/ui_extra_networks.py @@ -279,6 +279,7 @@ class ExtraNetworksPage: "date_created": int(stat.st_ctime or 0), "date_modified": int(stat.st_mtime or 0), "name": pth.name.lower(), + "path": str(pth.parent).lower(), } def find_preview(self, path): @@ -382,7 +383,7 @@ def create_ui(interface: gr.Blocks, unrelated_tabs, tabname): related_tabs.append(tab) edit_search = gr.Textbox('', show_label=False, elem_id=tabname+"_extra_search", elem_classes="search", placeholder="Search...", visible=False, interactive=True) - dropdown_sort = gr.Dropdown(choices=['Name', 'Date Created', 'Date Modified', ], value=shared.opts.extra_networks_card_order_field, elem_id=tabname+"_extra_sort", elem_classes="sort", multiselect=False, visible=False, show_label=False, interactive=True, label=tabname+"_extra_sort_order") + dropdown_sort = gr.Dropdown(choices=['Path', 'Name', 'Date Created', 'Date Modified', ], value=shared.opts.extra_networks_card_order_field, elem_id=tabname+"_extra_sort", elem_classes="sort", multiselect=False, visible=False, show_label=False, interactive=True, label=tabname+"_extra_sort_order") button_sortorder = ToolButton(switch_values_symbol, elem_id=tabname+"_extra_sortorder", elem_classes=["sortorder"] + ([] if shared.opts.extra_networks_card_order == "Ascending" else ["sortReverse"]), visible=False, tooltip="Invert sort order") button_refresh = gr.Button('Refresh', elem_id=tabname+"_extra_refresh", visible=False) checkbox_show_dirs = gr.Checkbox(True, label='Show dirs', elem_id=tabname+"_extra_show_dirs", elem_classes="show-dirs", visible=False) From a292d2c47f51fc71cc186709bdf3706f0944b7d6 Mon Sep 17 00:00:00 2001 From: AngelBottomless Date: Wed, 15 Nov 2023 14:26:37 +0900 Subject: [PATCH 090/342] hotfix: call shared.state.end() after postprocessing done --- modules/postprocessing.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/postprocessing.py b/modules/postprocessing.py index cf04d38b0..fd0c0cc99 100644 --- a/modules/postprocessing.py +++ b/modules/postprocessing.py @@ -78,7 +78,7 @@ def run_postprocessing(extras_mode, image, image_folder, input_dir, output_dir, image_data.close() devices.torch_gc() - + shared.state.end() return outputs, ui_common.plaintext_to_html(infotext), '' From b29fc6d4de8812b25c520a46676cda13c3fe64ca Mon Sep 17 00:00:00 2001 From: aria1th <35677394+aria1th@users.noreply.github.com> Date: Sat, 11 Nov 2023 23:43:13 +0900 Subject: [PATCH 091/342] Implement Hypertile Co-Authored-By: Kieran Hunt --- modules/hypertile.py | 333 ++++++++++++++++++++++++++++++++++++++++++ modules/processing.py | 65 ++++----- 2 files changed, 358 insertions(+), 40 deletions(-) create mode 100644 modules/hypertile.py diff --git a/modules/hypertile.py b/modules/hypertile.py new file mode 100644 index 000000000..ab1c74c02 --- /dev/null +++ b/modules/hypertile.py @@ -0,0 +1,333 @@ +""" +Hypertile module for splitting attention layers in SD-1.5 U-Net and SD-1.5 VAE +Warn : The patch works well only if the input image has a width and height that are multiples of 128 +Author : @tfernd Github : https://github.com/tfernd/HyperTile +""" + +from __future__ import annotations +from typing import Callable +from typing_extensions import Literal + +import logging +from functools import wraps, cache +from contextlib import contextmanager + +import math +import torch.nn as nn +import random + +from einops import rearrange + +# TODO add SD-XL layers +DEPTH_LAYERS = { + 0: [ + # SD 1.5 U-Net (diffusers) + "down_blocks.0.attentions.0.transformer_blocks.0.attn1", + "down_blocks.0.attentions.1.transformer_blocks.0.attn1", + "up_blocks.3.attentions.0.transformer_blocks.0.attn1", + "up_blocks.3.attentions.1.transformer_blocks.0.attn1", + "up_blocks.3.attentions.2.transformer_blocks.0.attn1", + # SD 1.5 U-Net (ldm) + "input_blocks.1.1.transformer_blocks.0.attn1", + "input_blocks.2.1.transformer_blocks.0.attn1", + "output_blocks.9.1.transformer_blocks.0.attn1", + "output_blocks.10.1.transformer_blocks.0.attn1", + "output_blocks.11.1.transformer_blocks.0.attn1", + # SD 1.5 VAE + "decoder.mid_block.attentions.0", + ], + 1: [ + # SD 1.5 U-Net (diffusers) + "down_blocks.1.attentions.0.transformer_blocks.0.attn1", + "down_blocks.1.attentions.1.transformer_blocks.0.attn1", + "up_blocks.2.attentions.0.transformer_blocks.0.attn1", + "up_blocks.2.attentions.1.transformer_blocks.0.attn1", + "up_blocks.2.attentions.2.transformer_blocks.0.attn1", + # SD 1.5 U-Net (ldm) + "input_blocks.4.1.transformer_blocks.0.attn1", + "input_blocks.5.1.transformer_blocks.0.attn1", + "output_blocks.6.1.transformer_blocks.0.attn1", + "output_blocks.7.1.transformer_blocks.0.attn1", + "output_blocks.8.1.transformer_blocks.0.attn1", + ], + 2: [ + # SD 1.5 U-Net (diffusers) + "down_blocks.2.attentions.0.transformer_blocks.0.attn1", + "down_blocks.2.attentions.1.transformer_blocks.0.attn1", + "up_blocks.1.attentions.0.transformer_blocks.0.attn1", + "up_blocks.1.attentions.1.transformer_blocks.0.attn1", + "up_blocks.1.attentions.2.transformer_blocks.0.attn1", + # SD 1.5 U-Net (ldm) + "input_blocks.7.1.transformer_blocks.0.attn1", + "input_blocks.8.1.transformer_blocks.0.attn1", + "output_blocks.3.1.transformer_blocks.0.attn1", + "output_blocks.4.1.transformer_blocks.0.attn1", + "output_blocks.5.1.transformer_blocks.0.attn1", + ], + 3: [ + # SD 1.5 U-Net (diffusers) + "mid_block.attentions.0.transformer_blocks.0.attn1", + # SD 1.5 U-Net (ldm) + "middle_block.1.transformer_blocks.0.attn1", + ], +} +# XL layers, thanks for GitHub@gel-crabs for the help +DEPTH_LAYERS_XL = { + 0: [ + # SD 1.5 U-Net (diffusers) + "down_blocks.0.attentions.0.transformer_blocks.0.attn1", + "down_blocks.0.attentions.1.transformer_blocks.0.attn1", + "up_blocks.3.attentions.0.transformer_blocks.0.attn1", + "up_blocks.3.attentions.1.transformer_blocks.0.attn1", + "up_blocks.3.attentions.2.transformer_blocks.0.attn1", + # SD 1.5 U-Net (ldm) + "input_blocks.4.1.transformer_blocks.0.attn1", + "input_blocks.5.1.transformer_blocks.0.attn1", + "output_blocks.3.1.transformer_blocks.0.attn1", + "output_blocks.4.1.transformer_blocks.0.attn1", + "output_blocks.5.1.transformer_blocks.0.attn1", + # SD 1.5 VAE + "decoder.mid_block.attentions.0", + "decoder.mid.attn_1", + ], + 1: [ + # SD 1.5 U-Net (diffusers) + #"down_blocks.1.attentions.0.transformer_blocks.0.attn1", + #"down_blocks.1.attentions.1.transformer_blocks.0.attn1", + #"up_blocks.2.attentions.0.transformer_blocks.0.attn1", + #"up_blocks.2.attentions.1.transformer_blocks.0.attn1", + #"up_blocks.2.attentions.2.transformer_blocks.0.attn1", + # SD 1.5 U-Net (ldm) + "input_blocks.4.1.transformer_blocks.1.attn1", + "input_blocks.5.1.transformer_blocks.1.attn1", + "output_blocks.3.1.transformer_blocks.1.attn1", + "output_blocks.4.1.transformer_blocks.1.attn1", + "output_blocks.5.1.transformer_blocks.1.attn1", + "input_blocks.7.1.transformer_blocks.0.attn1", + "input_blocks.8.1.transformer_blocks.0.attn1", + "output_blocks.0.1.transformer_blocks.0.attn1", + "output_blocks.1.1.transformer_blocks.0.attn1", + "output_blocks.2.1.transformer_blocks.0.attn1", + "input_blocks.7.1.transformer_blocks.1.attn1", + "input_blocks.8.1.transformer_blocks.1.attn1", + "output_blocks.0.1.transformer_blocks.1.attn1", + "output_blocks.1.1.transformer_blocks.1.attn1", + "output_blocks.2.1.transformer_blocks.1.attn1", + "input_blocks.7.1.transformer_blocks.2.attn1", + "input_blocks.8.1.transformer_blocks.2.attn1", + "output_blocks.0.1.transformer_blocks.2.attn1", + "output_blocks.1.1.transformer_blocks.2.attn1", + "output_blocks.2.1.transformer_blocks.2.attn1", + "input_blocks.7.1.transformer_blocks.3.attn1", + "input_blocks.8.1.transformer_blocks.3.attn1", + "output_blocks.0.1.transformer_blocks.3.attn1", + "output_blocks.1.1.transformer_blocks.3.attn1", + "output_blocks.2.1.transformer_blocks.3.attn1", + "input_blocks.7.1.transformer_blocks.4.attn1", + "input_blocks.8.1.transformer_blocks.4.attn1", + "output_blocks.0.1.transformer_blocks.4.attn1", + "output_blocks.1.1.transformer_blocks.4.attn1", + "output_blocks.2.1.transformer_blocks.4.attn1", + "input_blocks.7.1.transformer_blocks.5.attn1", + "input_blocks.8.1.transformer_blocks.5.attn1", + "output_blocks.0.1.transformer_blocks.5.attn1", + "output_blocks.1.1.transformer_blocks.5.attn1", + "output_blocks.2.1.transformer_blocks.5.attn1", + "input_blocks.7.1.transformer_blocks.6.attn1", + "input_blocks.8.1.transformer_blocks.6.attn1", + "output_blocks.0.1.transformer_blocks.6.attn1", + "output_blocks.1.1.transformer_blocks.6.attn1", + "output_blocks.2.1.transformer_blocks.6.attn1", + "input_blocks.7.1.transformer_blocks.7.attn1", + "input_blocks.8.1.transformer_blocks.7.attn1", + "output_blocks.0.1.transformer_blocks.7.attn1", + "output_blocks.1.1.transformer_blocks.7.attn1", + "output_blocks.2.1.transformer_blocks.7.attn1", + "input_blocks.7.1.transformer_blocks.8.attn1", + "input_blocks.8.1.transformer_blocks.8.attn1", + "output_blocks.0.1.transformer_blocks.8.attn1", + "output_blocks.1.1.transformer_blocks.8.attn1", + "output_blocks.2.1.transformer_blocks.8.attn1", + "input_blocks.7.1.transformer_blocks.9.attn1", + "input_blocks.8.1.transformer_blocks.9.attn1", + "output_blocks.0.1.transformer_blocks.9.attn1", + "output_blocks.1.1.transformer_blocks.9.attn1", + "output_blocks.2.1.transformer_blocks.9.attn1", + ], + 2: [ + # SD 1.5 U-Net (diffusers) + "mid_block.attentions.0.transformer_blocks.0.attn1", + # SD 1.5 U-Net (ldm) + "middle_block.1.transformer_blocks.0.attn1", + "middle_block.1.transformer_blocks.1.attn1", + "middle_block.1.transformer_blocks.2.attn1", + "middle_block.1.transformer_blocks.3.attn1", + "middle_block.1.transformer_blocks.4.attn1", + "middle_block.1.transformer_blocks.5.attn1", + "middle_block.1.transformer_blocks.6.attn1", + "middle_block.1.transformer_blocks.7.attn1", + "middle_block.1.transformer_blocks.8.attn1", + "middle_block.1.transformer_blocks.9.attn1", + ], +} + + +RNG_INSTANCE = random.Random() + +def random_divisor(value: int, min_value: int, /, max_options: int = 1) -> int: + """ + Returns a random divisor of value that + x * min_value <= value + if max_options is 1, the behavior is deterministic + """ + min_value = min(min_value, value) + + # All big divisors of value (inclusive) + divisors = [i for i in range(min_value, value + 1) if value % i == 0] # divisors in small -> big order + + ns = [value // i for i in divisors[:max_options]] # has at least 1 element # big -> small order + + idx = RNG_INSTANCE.randint(0, len(ns) - 1) + + return ns[idx] + +def set_hypertile_seed(seed: int) -> None: + RNG_INSTANCE.seed(seed) + +def largest_tile_size_available(width:int, height:int) -> int: + """ + Calculates the largest tile size available for a given width and height + Tile size is always a power of 2 + """ + gcd = math.gcd(width, height) + largest_tile_size_available = 1 + while gcd % (largest_tile_size_available * 2) == 0: + largest_tile_size_available *= 2 + return largest_tile_size_available + +def iterative_closest_divisors(hw:int, aspect_ratio:float) -> tuple[int, int]: + """ + Finds h and w such that h*w = hw and h/w = aspect_ratio + We check all possible divisors of hw and return the closest to the aspect ratio + """ + divisors = [i for i in range(2, hw + 1) if hw % i == 0] # all divisors of hw + pairs = [(i, hw // i) for i in divisors] # all pairs of divisors of hw + ratios = [w/h for h, w in pairs] # all ratios of pairs of divisors of hw + closest_ratio = min(ratios, key=lambda x: abs(x - aspect_ratio)) # closest ratio to aspect_ratio + closest_pair = pairs[ratios.index(closest_ratio)] # closest pair of divisors to aspect_ratio + return closest_pair + +@cache +def find_hw_candidates(hw:int, aspect_ratio:float) -> tuple[int, int]: + """ + Finds h and w such that h*w = hw and h/w = aspect_ratio + """ + h, w = round(math.sqrt(hw * aspect_ratio)), round(math.sqrt(hw / aspect_ratio)) + # find h and w such that h*w = hw and h/w = aspect_ratio + if h * w != hw: + w_candidate = hw / h + # check if w is an integer + if not w_candidate.is_integer(): + h_candidate = hw / w + # check if h is an integer + if not h_candidate.is_integer(): + return iterative_closest_divisors(hw, aspect_ratio) + else: + h = int(h_candidate) + else: + w = int(w_candidate) + return h, w + +@contextmanager +def split_attention( + layer: nn.Module, + /, + aspect_ratio: float, # width/height + tile_size: int = 128, # 128 for VAE + swap_size: int = 1, # 1 for VAE + *, + disable: bool = False, + max_depth: Literal[0, 1, 2, 3] = 0, # ! Try 0 or 1 + scale_depth: bool = True, # scale the tile-size depending on the depth + is_sdxl: bool = False, # is the model SD-XL +): + # Hijacks AttnBlock from ldm and Attention from diffusers + + if disable: + logging.info(f"Attention for {layer.__class__.__qualname__} not splitted") + yield + return + + latent_tile_size = max(128, tile_size) // 8 + + def self_attn_forward(forward: Callable, depth: int, layer_name: str, module: nn.Module) -> Callable: + @wraps(forward) + def wrapper(*args, **kwargs): + x = args[0] + + # VAE + if x.ndim == 4: + b, c, h, w = x.shape + + nh = random_divisor(h, latent_tile_size, swap_size) + nw = random_divisor(w, latent_tile_size, swap_size) + + if nh * nw > 1: + x = rearrange(x, "b c (nh h) (nw w) -> (b nh nw) c h w", nh=nh, nw=nw) # split into nh * nw tiles + + out = forward(x, *args[1:], **kwargs) + + if nh * nw > 1: + out = rearrange(out, "(b nh nw) c h w -> b c (nh h) (nw w)", nh=nh, nw=nw) + + # U-Net + else: + hw: int = x.size(1) + h, w = find_hw_candidates(hw, aspect_ratio) + assert h * w == hw, f"Invalid aspect ratio {aspect_ratio} for input of shape {x.shape}, hw={hw}, h={h}, w={w}" + + factor = 2**depth if scale_depth else 1 + nh = random_divisor(h, latent_tile_size * factor, swap_size) + nw = random_divisor(w, latent_tile_size * factor, swap_size) + + module._split_sizes_hypertile.append((nh, nw)) # type: ignore + + if nh * nw > 1: + x = rearrange(x, "b (nh h nw w) c -> (b nh nw) (h w) c", h=h // nh, w=w // nw, nh=nh, nw=nw) + + out = forward(x, *args[1:], **kwargs) + + if nh * nw > 1: + out = rearrange(out, "(b nh nw) hw c -> b nh nw hw c", nh=nh, nw=nw) + out = rearrange(out, "b nh nw (h w) c -> b (nh h nw w) c", h=h // nh, w=w // nw) + + return out + + return wrapper + + # Handle hijacking the forward method and recovering afterwards + try: + if is_sdxl: + layers = DEPTH_LAYERS_XL + else: + layers = DEPTH_LAYERS + for depth in range(max_depth + 1): + for layer_name, module in layer.named_modules(): + if any(layer_name.endswith(try_name) for try_name in layers[depth]): + # print input shape for debugging + logging.debug(f"HyperTile hijacking attention layer at depth {depth}: {layer_name}") + # hijack + module._original_forward_hypertile = module.forward + module.forward = self_attn_forward(module.forward, depth, layer_name, module) + module._split_sizes_hypertile = [] + yield + finally: + for layer_name, module in layer.named_modules(): + # remove hijack + if hasattr(module, "_original_forward_hypertile"): + if module._split_sizes_hypertile: + logging.debug(f"layer {layer_name} splitted with ({module._split_sizes_hypertile})") + # recover + module.forward = module._original_forward_hypertile + del module._original_forward_hypertile + del module._split_sizes_hypertile diff --git a/modules/processing.py b/modules/processing.py index e23095343..e19a09a3c 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -24,6 +24,7 @@ from modules.shared import opts, cmd_opts, state import modules.shared as shared import modules.paths as paths import modules.face_restoration +from modules.hypertile import split_attention, set_hypertile_seed, largest_tile_size_available import modules.images as images import modules.styles import modules.sd_models as sd_models @@ -799,17 +800,6 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: infotexts = [] output_images = [] - unet_object = p.sd_model.model - vae_model = p.sd_model.first_stage_model - try: - from hyper_tile import split_attention, flush - except (ImportError, ModuleNotFoundError): # pip install git+https://github.com/tfernd/HyperTile@2ef64b2800d007d305755c33550537410310d7df - split_attention = lambda *args, **kwargs: lambda x: x # return a no-op context manager - flush = lambda: None - import random - saved_rng_state = random.getstate() - random.seed(p.seed) # hyper_tile uses random, so we need to seed it - with torch.no_grad(), p.sd_model.ema_scope(): with devices.autocast(): p.init(p.all_prompts, p.all_seeds, p.all_subseeds) @@ -871,29 +861,20 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: p.comment(comment) p.extra_generation_params.update(model_hijack.extra_generation_params) - + set_hypertile_seed(p.seed) + # add batch size + hypertile status to information to reproduce the run if p.n_iter > 1: shared.state.job = f"Batch {n+1} out of {p.n_iter}" with devices.without_autocast() if devices.unet_needs_upcast else devices.autocast(): - # get largest tile size available, which is 2^x which is factor of gcd of p.width and p.height - gcd = math.gcd(p.width, p.height) - largest_tile_size_available = 1 - while gcd % (largest_tile_size_available * 2) == 0: - largest_tile_size_available *= 2 - aspect_ratio = p.width / p.height - with split_attention(vae_model, aspect_ratio=aspect_ratio, tile_size=min(largest_tile_size_available, 128), disable=not shared.opts.hypertile_split_vae_attn): - with split_attention(unet_object, aspect_ratio=aspect_ratio, tile_size=min(largest_tile_size_available, 256), swap_size=2, disable=not shared.opts.hypertile_split_unet_attn): - flush() - samples_ddim = p.sample(conditioning=p.c, unconditional_conditioning=p.uc, seeds=p.seeds, subseeds=p.subseeds, subseed_strength=p.subseed_strength, prompts=p.prompts) + samples_ddim = p.sample(conditioning=p.c, unconditional_conditioning=p.uc, seeds=p.seeds, subseeds=p.subseeds, subseed_strength=p.subseed_strength, prompts=p.prompts) if getattr(samples_ddim, 'already_decoded', False): x_samples_ddim = samples_ddim else: if opts.sd_vae_decode_method != 'Full': p.extra_generation_params['VAE Decoder'] = opts.sd_vae_decode_method - with split_attention(vae_model, aspect_ratio=aspect_ratio, tile_size=min(largest_tile_size_available, 128), disable=not shared.opts.hypertile_split_vae_attn): - flush() + with split_attention(p.sd_model.first_stage_model, aspect_ratio = p.width / p.height, tile_size=min(largest_tile_size_available(p.width, p.height), 128), disable=not shared.opts.hypertile_split_vae_attn, is_sdxl=shared.sd_model.is_sdxl): x_samples_ddim = decode_latent_batch(p.sd_model, samples_ddim, target_device=devices.cpu, check_for_nans=True) x_samples_ddim = torch.stack(x_samples_ddim).float() @@ -1000,7 +981,6 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: if opts.grid_save: images.save_image(grid, p.outpath_grids, "grid", p.all_seeds[0], p.all_prompts[0], opts.grid_format, info=infotext(use_main_prompt=True), short_filename=not opts.grid_extended_filename, p=p, grid=True) - random.setstate(saved_rng_state) if not p.disable_extra_networks and p.extra_network_data: extra_networks.deactivate(p, p.extra_network_data) @@ -1161,24 +1141,25 @@ class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing): def sample(self, conditioning, unconditional_conditioning, seeds, subseeds, subseed_strength, prompts): self.sampler = sd_samplers.create_sampler(self.sampler_name, self.sd_model) - + aspect_ratio = self.width / self.height x = self.rng.next() - samples = self.sampler.sample(self, x, conditioning, unconditional_conditioning, image_conditioning=self.txt2img_image_conditioning(x)) + tile_size = largest_tile_size_available(self.width, self.height) + with split_attention(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=min(tile_size, 128), swap_size=1, disable=not shared.opts.hypertile_split_vae_attn, is_sdxl=shared.sd_model.is_sdxl): + with split_attention(self.sd_model.model, aspect_ratio=aspect_ratio, tile_size=min(tile_size, 256), swap_size=2, disable=not shared.opts.hypertile_split_unet_attn, is_sdxl=shared.sd_model.is_sdxl): + devices.torch_gc() + samples = self.sampler.sample(self, x, conditioning, unconditional_conditioning, image_conditioning=self.txt2img_image_conditioning(x)) del x - if not self.enable_hr: return samples if self.latent_scale_mode is None: - decoded_samples = torch.stack(decode_latent_batch(self.sd_model, samples, target_device=devices.cpu, check_for_nans=True)).to(dtype=torch.float32) + with split_attention(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=min(tile_size, 256), swap_size=1, disable=not shared.opts.hypertile_split_vae_attn, is_sdxl=shared.sd_model.is_sdxl): + decoded_samples = torch.stack(decode_latent_batch(self.sd_model, samples, target_device=devices.cpu, check_for_nans=True)).to(dtype=torch.float32) else: decoded_samples = None with sd_models.SkipWritingToConfig(): sd_models.reload_model_weights(info=self.hr_checkpoint_info) - - devices.torch_gc() - return self.sample_hr_pass(samples, decoded_samples, seeds, subseeds, subseed_strength, prompts) def sample_hr_pass(self, samples, decoded_samples, seeds, subseeds, subseed_strength, prompts): @@ -1186,7 +1167,6 @@ class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing): return samples self.is_hr_pass = True - target_width = self.hr_upscale_to_x target_height = self.hr_upscale_to_y @@ -1264,18 +1244,19 @@ class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing): if self.scripts is not None: self.scripts.before_hr(self) - - samples = self.sampler.sample_img2img(self, samples, noise, self.hr_c, self.hr_uc, steps=self.hr_second_pass_steps or self.steps, image_conditioning=image_conditioning) + tile_size = largest_tile_size_available(target_width, target_height) + with split_attention(self.sd_model.first_stage_model, aspect_ratio=target_width / target_height, tile_size=min(tile_size, 256), swap_size=1, disable=not opts.hypertile_split_vae_attn, is_sdxl=shared.sd_model.is_sdxl): + with split_attention(self.sd_model.model, aspect_ratio=target_width / target_height, tile_size=min(tile_size, 256), swap_size=3, max_depth=1,scale_depth=True, disable=not opts.hypertile_split_unet_attn, is_sdxl=shared.sd_model.is_sdxl): + samples = self.sampler.sample_img2img(self, samples, noise, self.hr_c, self.hr_uc, steps=self.hr_second_pass_steps or self.steps, image_conditioning=image_conditioning) sd_models.apply_token_merging(self.sd_model, self.get_token_merging_ratio()) self.sampler = None devices.torch_gc() - - decoded_samples = decode_latent_batch(self.sd_model, samples, target_device=devices.cpu, check_for_nans=True) + with split_attention(self.sd_model.first_stage_model, aspect_ratio=target_width / target_height, tile_size=min(tile_size, 256), swap_size=1, disable=not opts.hypertile_split_vae_attn, is_sdxl=shared.sd_model.is_sdxl): + decoded_samples = decode_latent_batch(self.sd_model, samples, target_device=devices.cpu, check_for_nans=True) self.is_hr_pass = False - return decoded_samples def close(self): @@ -1550,8 +1531,12 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): if self.initial_noise_multiplier != 1.0: self.extra_generation_params["Noise multiplier"] = self.initial_noise_multiplier x *= self.initial_noise_multiplier - - samples = self.sampler.sample_img2img(self, self.init_latent, x, conditioning, unconditional_conditioning, image_conditioning=self.image_conditioning) + aspect_ratio = self.width / self.height + tile_size = largest_tile_size_available(self.width, self.height) + with split_attention(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=min(tile_size, 128), swap_size=1, disable=not shared.opts.hypertile_split_vae_attn, is_sdxl=shared.sd_model.is_sdxl): + with split_attention(self.sd_model.model, aspect_ratio=aspect_ratio, tile_size=min(tile_size, 256), swap_size=2, disable=not shared.opts.hypertile_split_unet_attn, is_sdxl=shared.sd_model.is_sdxl): + devices.torch_gc() + samples = self.sampler.sample_img2img(self, self.init_latent, x, conditioning, unconditional_conditioning, image_conditioning=self.image_conditioning) if self.mask is not None: samples = samples * self.nmask + self.init_latent * self.mask From af45872fdb8a66ffd6a405d99120e0bacbb4a170 Mon Sep 17 00:00:00 2001 From: aria1th <35677394+aria1th@users.noreply.github.com> Date: Wed, 15 Nov 2023 15:15:14 +0900 Subject: [PATCH 092/342] copy LDM VAE key from XL --- modules/hypertile.py | 1 + 1 file changed, 1 insertion(+) diff --git a/modules/hypertile.py b/modules/hypertile.py index ab1c74c02..32d8604cc 100644 --- a/modules/hypertile.py +++ b/modules/hypertile.py @@ -35,6 +35,7 @@ DEPTH_LAYERS = { "output_blocks.11.1.transformer_blocks.0.attn1", # SD 1.5 VAE "decoder.mid_block.attentions.0", + "decoder.mid.attn_1", ], 1: [ # SD 1.5 U-Net (diffusers) From d6d0b22e6657fc84039e82ee735a57101bfe7c17 Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Wed, 15 Nov 2023 03:08:50 -0800 Subject: [PATCH 093/342] fix: ignore calc_scale() for COFT which has very small alpha --- extensions-builtin/Lora/network_oft.py | 16 +++++----------- 1 file changed, 5 insertions(+), 11 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index 93402bb28..c45a8d23a 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -99,12 +99,9 @@ class NetworkModuleOFT(network.NetworkModule): is_other_linear = type(self.sd_module) in [torch.nn.MultiheadAttention] if not is_other_linear: - #if is_other_linear and orig_weight.shape[0] != orig_weight.shape[1]: - # orig_weight=orig_weight.permute(1, 0) - oft_blocks = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) - # without this line the results are significantly worse / less accurate + # ensure skew-symmetric matrix oft_blocks = oft_blocks - oft_blocks.transpose(1, 2) R = oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) @@ -118,9 +115,6 @@ class NetworkModuleOFT(network.NetworkModule): ) merged_weight = rearrange(merged_weight, 'k m ... -> (k m) ...') - #if is_other_linear and orig_weight.shape[0] != orig_weight.shape[1]: - # orig_weight=orig_weight.permute(1, 0) - updown = merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight output_shape = orig_weight.shape else: @@ -132,10 +126,10 @@ class NetworkModuleOFT(network.NetworkModule): return self.finalize_updown(updown, orig_weight, output_shape) def calc_updown(self, orig_weight): - multiplier = self.multiplier() * self.calc_scale() - #if self.is_kohya: - # return self.calc_updown_kohya(orig_weight, multiplier) - #else: + # if alpha is a very small number as in coft, calc_scale will return a almost zero number so we ignore it + #multiplier = self.multiplier() * self.calc_scale() + multiplier = self.multiplier() + return self.calc_updown_kb(orig_weight, multiplier) # override to remove the multiplier/scale factor; it's already multiplied in get_weight From eb667e715ad3eea981f6263c143ab0422e5340c9 Mon Sep 17 00:00:00 2001 From: v0xie <28695009+v0xie@users.noreply.github.com> Date: Wed, 15 Nov 2023 18:28:48 -0800 Subject: [PATCH 094/342] feat: LyCORIS/kohya OFT network support --- extensions-builtin/Lora/network_oft.py | 108 ++++++------------------- 1 file changed, 26 insertions(+), 82 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index c45a8d23a..05c378118 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -11,8 +11,8 @@ class ModuleTypeOFT(network.ModuleType): return None -# adapted from kohya-ss' implementation https://github.com/kohya-ss/sd-scripts/blob/main/networks/oft.py -# and KohakuBlueleaf's implementation https://github.com/KohakuBlueleaf/LyCORIS/blob/dev/lycoris/modules/diag_oft.py +# Supports both kohya-ss' implementation of COFT https://github.com/kohya-ss/sd-scripts/blob/main/networks/oft.py +# and KohakuBlueleaf's implementation of OFT/COFT https://github.com/KohakuBlueleaf/LyCORIS/blob/dev/lycoris/modules/diag_oft.py class NetworkModuleOFT(network.NetworkModule): def __init__(self, net: network.Network, weights: network.NetworkWeights): @@ -25,117 +25,61 @@ class NetworkModuleOFT(network.NetworkModule): if "oft_blocks" in weights.w.keys(): self.is_kohya = True self.oft_blocks = weights.w["oft_blocks"] # (num_blocks, block_size, block_size) - self.alpha = weights.w["alpha"] + self.alpha = weights.w["alpha"] # alpha is constraint self.dim = self.oft_blocks.shape[0] # lora dim - #self.oft_blocks = rearrange(self.oft_blocks, 'k m ... -> (k m) ...') + # LyCORIS elif "oft_diag" in weights.w.keys(): self.is_kohya = False - self.oft_blocks = weights.w["oft_diag"] # (num_blocks, block_size, block_size) - - # alpha is rank if alpha is 0 or None - if self.alpha is None: - pass - self.dim = self.oft_blocks.shape[1] # FIXME: almost certainly incorrect, assumes tensor is shape [*, m, n] - else: - raise ValueError("oft_blocks or oft_diag must be in weights dict") + self.oft_blocks = weights.w["oft_diag"] + # self.alpha is unused + self.dim = self.oft_blocks.shape[1] # (num_blocks, block_size, block_size) is_linear = type(self.sd_module) in [torch.nn.Linear, torch.nn.modules.linear.NonDynamicallyQuantizableLinear] is_conv = type(self.sd_module) in [torch.nn.Conv2d] - is_other_linear = type(self.sd_module) in [torch.nn.MultiheadAttention] + is_other_linear = type(self.sd_module) in [torch.nn.MultiheadAttention] # unsupported if is_linear: self.out_dim = self.sd_module.out_features - elif is_other_linear: - self.out_dim = self.sd_module.embed_dim elif is_conv: self.out_dim = self.sd_module.out_channels - else: - raise ValueError("sd_module must be Linear or Conv") + elif is_other_linear: + self.out_dim = self.sd_module.embed_dim if self.is_kohya: self.constraint = self.alpha * self.out_dim - self.num_blocks, self.block_size = factorization(self.out_dim, self.dim) + self.num_blocks = self.dim + self.block_size = self.out_dim // self.dim else: self.constraint = None self.block_size, self.num_blocks = factorization(self.out_dim, self.dim) - def merge_weight(self, R_weight, org_weight): - R_weight = R_weight.to(org_weight.device, dtype=org_weight.dtype) - if org_weight.dim() == 4: - weight = torch.einsum("oihw, op -> pihw", org_weight, R_weight) - else: - weight = torch.einsum("oi, op -> pi", org_weight, R_weight) - return weight + def calc_updown_kb(self, orig_weight, multiplier): + oft_blocks = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) + oft_blocks = oft_blocks - oft_blocks.transpose(1, 2) # ensure skew-symmetric orthogonal matrix - def get_weight(self, oft_blocks, multiplier=None): - if self.constraint is not None: - constraint = self.constraint.to(oft_blocks.device, dtype=oft_blocks.dtype) + R = oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) + R = R * multiplier + torch.eye(self.block_size, device=orig_weight.device) - block_Q = oft_blocks - oft_blocks.transpose(1, 2) - norm_Q = torch.norm(block_Q.flatten()) - if self.constraint is not None: - new_norm_Q = torch.clamp(norm_Q, max=constraint) - else: - new_norm_Q = norm_Q - block_Q = block_Q * ((new_norm_Q + 1e-8) / (norm_Q + 1e-8)) - m_I = torch.eye(self.num_blocks, device=oft_blocks.device).unsqueeze(0).repeat(self.block_size, 1, 1) - #m_I = torch.eye(self.block_size, device=oft_blocks.device).unsqueeze(0).repeat(self.num_blocks, 1, 1) - block_R = torch.matmul(m_I + block_Q, (m_I - block_Q).inverse()) - - block_R_weighted = multiplier * block_R + (1 - multiplier) * m_I - R = torch.block_diag(*block_R_weighted) - return R - - def calc_updown_kohya(self, orig_weight, multiplier): - R = self.get_weight(self.oft_blocks, multiplier) - merged_weight = self.merge_weight(R, orig_weight) + # This errors out for MultiheadAttention, might need to be handled up-stream + merged_weight = rearrange(orig_weight, '(k n) ... -> k n ...', k=self.num_blocks, n=self.block_size) + merged_weight = torch.einsum( + 'k n m, k n ... -> k m ...', + R, + merged_weight + ) + merged_weight = rearrange(merged_weight, 'k m ... -> (k m) ...') updown = merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight output_shape = orig_weight.shape - orig_weight = orig_weight - return self.finalize_updown(updown, orig_weight, output_shape) - - def calc_updown_kb(self, orig_weight, multiplier): - is_other_linear = type(self.sd_module) in [torch.nn.MultiheadAttention] - - if not is_other_linear: - oft_blocks = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) - - # ensure skew-symmetric matrix - oft_blocks = oft_blocks - oft_blocks.transpose(1, 2) - - R = oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) - R = R * multiplier + torch.eye(self.block_size, device=orig_weight.device) - - merged_weight = rearrange(orig_weight, '(k n) ... -> k n ...', k=self.num_blocks, n=self.block_size) - merged_weight = torch.einsum( - 'k n m, k n ... -> k m ...', - R, - merged_weight - ) - merged_weight = rearrange(merged_weight, 'k m ... -> (k m) ...') - - updown = merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight - output_shape = orig_weight.shape - else: - # FIXME: skip MultiheadAttention for now - #up = self.lin_module.weight.to(orig_weight.device, dtype=orig_weight.dtype) - updown = torch.zeros([orig_weight.shape[1], orig_weight.shape[1]], device=orig_weight.device, dtype=orig_weight.dtype) - output_shape = (orig_weight.shape[1], orig_weight.shape[1]) - return self.finalize_updown(updown, orig_weight, output_shape) def calc_updown(self, orig_weight): - # if alpha is a very small number as in coft, calc_scale will return a almost zero number so we ignore it - #multiplier = self.multiplier() * self.calc_scale() + # if alpha is a very small number as in coft, calc_scale() will return a almost zero number so we ignore it multiplier = self.multiplier() - return self.calc_updown_kb(orig_weight, multiplier) # override to remove the multiplier/scale factor; it's already multiplied in get_weight def finalize_updown(self, updown, orig_weight, output_shape, ex_bias=None): - #return super().finalize_updown(updown, orig_weight, output_shape, ex_bias) - if self.bias is not None: updown = updown.reshape(self.bias.shape) updown += self.bias.to(orig_weight.device, dtype=orig_weight.dtype) From bcfaf3979a9f93e37c418b58c75b02d9570b4354 Mon Sep 17 00:00:00 2001 From: AngelBottomless Date: Thu, 16 Nov 2023 18:43:16 +0900 Subject: [PATCH 095/342] convert/add hypertile options --- modules/hypertile.py | 36 ++++++++++++++++++++++++++++++++++++ modules/processing.py | 21 +++++++++++---------- modules/shared_options.py | 6 ++++++ 3 files changed, 53 insertions(+), 10 deletions(-) diff --git a/modules/hypertile.py b/modules/hypertile.py index 32d8604cc..fee24a8ca 100644 --- a/modules/hypertile.py +++ b/modules/hypertile.py @@ -332,3 +332,39 @@ def split_attention( module.forward = module._original_forward_hypertile del module._original_forward_hypertile del module._split_sizes_hypertile + +def hypertile_context_vae(model:nn.Module, aspect_ratio:float, tile_size:int, opts): + """ + Returns context manager for VAE + """ + enabled = not opts.hypertile_split_vae_attn + swap_size = opts.hypertile_swap_size_vae + max_depth = opts.hypertile_max_depth_vae + tile_size_max = opts.hypertile_max_tile_vae + return split_attention( + model, + aspect_ratio=aspect_ratio, + tile_size=min(tile_size, tile_size_max), + swap_size=swap_size, + disable=not enabled, + max_depth=max_depth, + is_sdxl=False, + ) + +def hypertile_context_unet(model:nn.Module, aspect_ratio:float, tile_size:int, opts, is_sdxl:bool): + """ + Returns context manager for U-Net + """ + enabled = not opts.hypertile_split_unet_attn + swap_size = opts.hypertile_swap_size_unet + max_depth = opts.hypertile_max_depth_unet + tile_size_max = opts.hypertile_max_tile_unet + return split_attention( + model, + aspect_ratio=aspect_ratio, + tile_size=min(tile_size, tile_size_max), + swap_size=swap_size, + disable=not enabled, + max_depth=max_depth, + is_sdxl=is_sdxl, + ) \ No newline at end of file diff --git a/modules/processing.py b/modules/processing.py index e19a09a3c..c622ff337 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -24,7 +24,7 @@ from modules.shared import opts, cmd_opts, state import modules.shared as shared import modules.paths as paths import modules.face_restoration -from modules.hypertile import split_attention, set_hypertile_seed, largest_tile_size_available +from modules.hypertile import set_hypertile_seed, largest_tile_size_available, hypertile_context_unet, hypertile_context_vae import modules.images as images import modules.styles import modules.sd_models as sd_models @@ -874,7 +874,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: else: if opts.sd_vae_decode_method != 'Full': p.extra_generation_params['VAE Decoder'] = opts.sd_vae_decode_method - with split_attention(p.sd_model.first_stage_model, aspect_ratio = p.width / p.height, tile_size=min(largest_tile_size_available(p.width, p.height), 128), disable=not shared.opts.hypertile_split_vae_attn, is_sdxl=shared.sd_model.is_sdxl): + with hypertile_context_unet(p.sd_model.first_stage_model, aspect_ratio=p.width / p.height, tile_size=largest_tile_size_available(p.width, p.height), is_sdxl=shared.sd_model.is_sdxl, opts=shared.opts): x_samples_ddim = decode_latent_batch(p.sd_model, samples_ddim, target_device=devices.cpu, check_for_nans=True) x_samples_ddim = torch.stack(x_samples_ddim).float() @@ -1144,8 +1144,8 @@ class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing): aspect_ratio = self.width / self.height x = self.rng.next() tile_size = largest_tile_size_available(self.width, self.height) - with split_attention(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=min(tile_size, 128), swap_size=1, disable=not shared.opts.hypertile_split_vae_attn, is_sdxl=shared.sd_model.is_sdxl): - with split_attention(self.sd_model.model, aspect_ratio=aspect_ratio, tile_size=min(tile_size, 256), swap_size=2, disable=not shared.opts.hypertile_split_unet_attn, is_sdxl=shared.sd_model.is_sdxl): + with hypertile_context_vae(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, opts=shared.opts): + with hypertile_context_unet(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, is_sdxl=shared.sd_model.is_sdxl, opts=shared.opts): devices.torch_gc() samples = self.sampler.sample(self, x, conditioning, unconditional_conditioning, image_conditioning=self.txt2img_image_conditioning(x)) del x @@ -1153,7 +1153,7 @@ class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing): return samples if self.latent_scale_mode is None: - with split_attention(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=min(tile_size, 256), swap_size=1, disable=not shared.opts.hypertile_split_vae_attn, is_sdxl=shared.sd_model.is_sdxl): + with hypertile_context_vae(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, opts=shared.opts): decoded_samples = torch.stack(decode_latent_batch(self.sd_model, samples, target_device=devices.cpu, check_for_nans=True)).to(dtype=torch.float32) else: decoded_samples = None @@ -1245,15 +1245,16 @@ class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing): if self.scripts is not None: self.scripts.before_hr(self) tile_size = largest_tile_size_available(target_width, target_height) - with split_attention(self.sd_model.first_stage_model, aspect_ratio=target_width / target_height, tile_size=min(tile_size, 256), swap_size=1, disable=not opts.hypertile_split_vae_attn, is_sdxl=shared.sd_model.is_sdxl): - with split_attention(self.sd_model.model, aspect_ratio=target_width / target_height, tile_size=min(tile_size, 256), swap_size=3, max_depth=1,scale_depth=True, disable=not opts.hypertile_split_unet_attn, is_sdxl=shared.sd_model.is_sdxl): + aspect_ratio = self.width / self.height + with hypertile_context_vae(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, opts=shared.opts): + with hypertile_context_unet(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, is_sdxl=shared.sd_model.is_sdxl, opts=shared.opts): samples = self.sampler.sample_img2img(self, samples, noise, self.hr_c, self.hr_uc, steps=self.hr_second_pass_steps or self.steps, image_conditioning=image_conditioning) sd_models.apply_token_merging(self.sd_model, self.get_token_merging_ratio()) self.sampler = None devices.torch_gc() - with split_attention(self.sd_model.first_stage_model, aspect_ratio=target_width / target_height, tile_size=min(tile_size, 256), swap_size=1, disable=not opts.hypertile_split_vae_attn, is_sdxl=shared.sd_model.is_sdxl): + with hypertile_context_vae(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, opts=shared.opts): decoded_samples = decode_latent_batch(self.sd_model, samples, target_device=devices.cpu, check_for_nans=True) self.is_hr_pass = False @@ -1533,8 +1534,8 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): x *= self.initial_noise_multiplier aspect_ratio = self.width / self.height tile_size = largest_tile_size_available(self.width, self.height) - with split_attention(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=min(tile_size, 128), swap_size=1, disable=not shared.opts.hypertile_split_vae_attn, is_sdxl=shared.sd_model.is_sdxl): - with split_attention(self.sd_model.model, aspect_ratio=aspect_ratio, tile_size=min(tile_size, 256), swap_size=2, disable=not shared.opts.hypertile_split_unet_attn, is_sdxl=shared.sd_model.is_sdxl): + with hypertile_context_vae(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, opts=shared.opts): + with hypertile_context_unet(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, is_sdxl=shared.sd_model.is_sdxl, opts=shared.opts): devices.torch_gc() samples = self.sampler.sample_img2img(self, self.init_latent, x, conditioning, unconditional_conditioning, image_conditioning=self.image_conditioning) diff --git a/modules/shared_options.py b/modules/shared_options.py index d96502656..28a489069 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -202,6 +202,12 @@ options_templates.update(options_section(('optimizations', "Optimizations"), { "batch_cond_uncond": OptionInfo(True, "Batch cond/uncond").info("do both conditional and unconditional denoising in one batch; uses a bit more VRAM during sampling, but improves speed; previously this was controlled by --always-batch-cond-uncond comandline argument"), "hypertile_split_unet_attn" : OptionInfo(False, "Split attention in Unet with HyperTile").link("Github", "https://github.com/tfernd/HyperTile").info("improves performance; changes behavior, but deterministic"), "hypertile_split_vae_attn": OptionInfo(False, "Split attention in VAE with HyperTile").link("Github", "https://github.com/tfernd/HyperTile").info("improves performance; changes behavior, but deterministic"), + "hypertile_max_depth_vae" : OptionInfo(3, "Max depth for VAE HyperTile hijack", gr.Slider, {"minimum": 0, "maximum": 3, "step": 1}).link("Github", "https://github.com/tfernd/HyperTile"), + "hypertile_max_depth_unet" : OptionInfo(3, "Max depth for Unet HyperTile hijack", gr.Slider, {"minimum": 0, "maximum": 3, "step": 1}).link("Github", "https://github.com/tfernd/HyperTile"), + "hypertile_max_tile_vae" : OptionInfo(128, "Max tile size for VAE HyperTile hijack", gr.Slider, {"minimum": 0, "maximum": 512, "step": 16}).link("Github", "https://github.com/tfernd/HyperTile"), + "hypertile_max_tile_unet" : OptionInfo(256, "Max tile size for Unet HyperTile hijack", gr.Slider, {"minimum": 0, "maximum": 512, "step": 16}).link("Github", "https://github.com/tfernd/HyperTile"), + "hypertile_swap_size_unet": OptionInfo(3, "Swap size for Unet HyperTile hijack", gr.Slider, {"minimum": 0, "maximum": 6, "step": 1}).link("Github", "https://github.com/tfernd/HyperTile"), + "hypertile_swap_size_vae": OptionInfo(3, "Swap size for VAE HyperTile hijack", gr.Slider, {"minimum": 0, "maximum": 6, "step": 1}).link("Github", "https://github.com/tfernd/HyperTile"), })) options_templates.update(options_section(('compatibility', "Compatibility"), { From 472c22cc8a46b825545d5c86bd2745269430d7b0 Mon Sep 17 00:00:00 2001 From: AngelBottomless Date: Thu, 16 Nov 2023 19:03:45 +0900 Subject: [PATCH 096/342] fix ruff - add newline --- modules/hypertile.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/hypertile.py b/modules/hypertile.py index fee24a8ca..86acecdc0 100644 --- a/modules/hypertile.py +++ b/modules/hypertile.py @@ -367,4 +367,4 @@ def hypertile_context_unet(model:nn.Module, aspect_ratio:float, tile_size:int, o disable=not enabled, max_depth=max_depth, is_sdxl=is_sdxl, - ) \ No newline at end of file + ) From 236eb82c3a91960ba5db7b82efbe0f6a9fd7cf24 Mon Sep 17 00:00:00 2001 From: Lucas Daniel Velazquez M <19197331+Luxter77@users.noreply.github.com> Date: Thu, 16 Nov 2023 13:20:33 -0300 Subject: [PATCH 097/342] Adds tqdm handler to logging_config.py for progress bar integration --- modules/logging_config.py | 14 ++++++++++++++ 1 file changed, 14 insertions(+) diff --git a/modules/logging_config.py b/modules/logging_config.py index 7db23d4b6..ce831b5c6 100644 --- a/modules/logging_config.py +++ b/modules/logging_config.py @@ -1,6 +1,19 @@ import os import logging +from tqdm.auto import tqdm + +class TqdmLoggingHandler(logging.Handler): + def __init__(self, level=logging.INFO): + super().__init__(level) + + def emit(self, record): + try: + msg = self.format(record) + tqdm.write(msg) + self.flush() + except Exception: + self.handleError(record) def setup_logging(loglevel): if loglevel is None: @@ -12,5 +25,6 @@ def setup_logging(loglevel): level=log_level, format='%(asctime)s %(levelname)s [%(name)s] %(message)s', datefmt='%Y-%m-%d %H:%M:%S', + handlers=[TqdmLoggingHandler()] ) From cdb60a690dcd35e865eb0caef6c6d8ff64e1b0d5 Mon Sep 17 00:00:00 2001 From: Lucas Daniel Velazquez M <19197331+Luxter77@users.noreply.github.com> Date: Thu, 16 Nov 2023 16:43:59 -0300 Subject: [PATCH 098/342] Take into account tqdm not being installed before first boot for logging --- modules/logging_config.py | 37 ++++++++++++++++++++++++------------- 1 file changed, 24 insertions(+), 13 deletions(-) diff --git a/modules/logging_config.py b/modules/logging_config.py index ce831b5c6..99ed2855b 100644 --- a/modules/logging_config.py +++ b/modules/logging_config.py @@ -1,30 +1,41 @@ import os import logging -from tqdm.auto import tqdm +try: + from tqdm.auto import tqdm -class TqdmLoggingHandler(logging.Handler): - def __init__(self, level=logging.INFO): - super().__init__(level) + class TqdmLoggingHandler(logging.Handler): + def __init__(self, level=logging.INFO): + super().__init__(level) - def emit(self, record): - try: - msg = self.format(record) - tqdm.write(msg) - self.flush() - except Exception: - self.handleError(record) + def emit(self, record): + try: + msg = self.format(record) + tqdm.write(msg) + self.flush() + except Exception: + self.handleError(record) + + TQDM_IMPORTED = True +except ImportError: + # tqdm does not exist before first launch + # I will import once the UI finishes seting up the enviroment and reloads. + TQDM_IMPORTED = False def setup_logging(loglevel): if loglevel is None: loglevel = os.environ.get("SD_WEBUI_LOG_LEVEL") + loghandlers = [] + + if TQDM_IMPORTED: + loghandlers.append(TqdmLoggingHandler()) + if loglevel: log_level = getattr(logging, loglevel.upper(), None) or logging.INFO logging.basicConfig( level=log_level, format='%(asctime)s %(levelname)s [%(name)s] %(message)s', datefmt='%Y-%m-%d %H:%M:%S', - handlers=[TqdmLoggingHandler()] + handlers=[] ) - From 7021cdb1de12be3071ecb67aa8d2e34e7a0ec5ab Mon Sep 17 00:00:00 2001 From: Your Name Date: Thu, 16 Nov 2023 17:53:57 -0300 Subject: [PATCH 099/342] actually adds handler to logging_config.py --- modules/logging_config.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/logging_config.py b/modules/logging_config.py index 99ed2855b..792698756 100644 --- a/modules/logging_config.py +++ b/modules/logging_config.py @@ -37,5 +37,5 @@ def setup_logging(loglevel): level=log_level, format='%(asctime)s %(levelname)s [%(name)s] %(message)s', datefmt='%Y-%m-%d %H:%M:%S', - handlers=[] + handlers=loghandlers ) From c40be2252ab1c8c289562db208c5ac6618bd8545 Mon Sep 17 00:00:00 2001 From: aria1th <35677394+aria1th@users.noreply.github.com> Date: Fri, 17 Nov 2023 09:22:27 +0900 Subject: [PATCH 100/342] Fix critical issue - unet apply --- modules/processing.py | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index c622ff337..2fda7f332 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -874,7 +874,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: else: if opts.sd_vae_decode_method != 'Full': p.extra_generation_params['VAE Decoder'] = opts.sd_vae_decode_method - with hypertile_context_unet(p.sd_model.first_stage_model, aspect_ratio=p.width / p.height, tile_size=largest_tile_size_available(p.width, p.height), is_sdxl=shared.sd_model.is_sdxl, opts=shared.opts): + with hypertile_context_unet(p.sd_model.model, aspect_ratio=p.width / p.height, tile_size=largest_tile_size_available(p.width, p.height), is_sdxl=shared.sd_model.is_sdxl, opts=shared.opts): x_samples_ddim = decode_latent_batch(p.sd_model, samples_ddim, target_device=devices.cpu, check_for_nans=True) x_samples_ddim = torch.stack(x_samples_ddim).float() @@ -1145,7 +1145,7 @@ class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing): x = self.rng.next() tile_size = largest_tile_size_available(self.width, self.height) with hypertile_context_vae(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, opts=shared.opts): - with hypertile_context_unet(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, is_sdxl=shared.sd_model.is_sdxl, opts=shared.opts): + with hypertile_context_unet(self.sd_model.model, aspect_ratio=aspect_ratio, tile_size=tile_size, is_sdxl=shared.sd_model.is_sdxl, opts=shared.opts): devices.torch_gc() samples = self.sampler.sample(self, x, conditioning, unconditional_conditioning, image_conditioning=self.txt2img_image_conditioning(x)) del x @@ -1247,7 +1247,7 @@ class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing): tile_size = largest_tile_size_available(target_width, target_height) aspect_ratio = self.width / self.height with hypertile_context_vae(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, opts=shared.opts): - with hypertile_context_unet(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, is_sdxl=shared.sd_model.is_sdxl, opts=shared.opts): + with hypertile_context_unet(self.sd_model.model, aspect_ratio=aspect_ratio, tile_size=tile_size, is_sdxl=shared.sd_model.is_sdxl, opts=shared.opts): samples = self.sampler.sample_img2img(self, samples, noise, self.hr_c, self.hr_uc, steps=self.hr_second_pass_steps or self.steps, image_conditioning=image_conditioning) sd_models.apply_token_merging(self.sd_model, self.get_token_merging_ratio()) @@ -1535,7 +1535,7 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): aspect_ratio = self.width / self.height tile_size = largest_tile_size_available(self.width, self.height) with hypertile_context_vae(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, opts=shared.opts): - with hypertile_context_unet(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, is_sdxl=shared.sd_model.is_sdxl, opts=shared.opts): + with hypertile_context_unet(self.sd_model.model, aspect_ratio=aspect_ratio, tile_size=tile_size, is_sdxl=shared.sd_model.is_sdxl, opts=shared.opts): devices.torch_gc() samples = self.sampler.sample_img2img(self, self.init_latent, x, conditioning, unconditional_conditioning, image_conditioning=self.image_conditioning) From c0725ba2d098a6a78610e7d96ee75f63a32d4e52 Mon Sep 17 00:00:00 2001 From: aria1th <35677394+aria1th@users.noreply.github.com> Date: Fri, 17 Nov 2023 09:34:50 +0900 Subject: [PATCH 101/342] Fix inverted option issue I'm pretty sure I was sleepy while implementing this --- modules/hypertile.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/modules/hypertile.py b/modules/hypertile.py index 86acecdc0..3a1468c6b 100644 --- a/modules/hypertile.py +++ b/modules/hypertile.py @@ -337,7 +337,7 @@ def hypertile_context_vae(model:nn.Module, aspect_ratio:float, tile_size:int, op """ Returns context manager for VAE """ - enabled = not opts.hypertile_split_vae_attn + enabled = opts.hypertile_split_vae_attn swap_size = opts.hypertile_swap_size_vae max_depth = opts.hypertile_max_depth_vae tile_size_max = opts.hypertile_max_tile_vae @@ -355,7 +355,7 @@ def hypertile_context_unet(model:nn.Module, aspect_ratio:float, tile_size:int, o """ Returns context manager for U-Net """ - enabled = not opts.hypertile_split_unet_attn + enabled = opts.hypertile_split_unet_attn swap_size = opts.hypertile_swap_size_unet max_depth = opts.hypertile_max_depth_unet tile_size_max = opts.hypertile_max_tile_unet From ffd0f8ddc309688636ac1ac10d82b72ab6b466bf Mon Sep 17 00:00:00 2001 From: aria1th <35677394+aria1th@users.noreply.github.com> Date: Fri, 17 Nov 2023 09:54:33 +0900 Subject: [PATCH 102/342] set empty value for SD XL 3rd layer --- modules/hypertile.py | 1 + 1 file changed, 1 insertion(+) diff --git a/modules/hypertile.py b/modules/hypertile.py index 3a1468c6b..be898fce4 100644 --- a/modules/hypertile.py +++ b/modules/hypertile.py @@ -170,6 +170,7 @@ DEPTH_LAYERS_XL = { "middle_block.1.transformer_blocks.8.attn1", "middle_block.1.transformer_blocks.9.attn1", ], + 3 : [] # TODO - separate layers for SD-XL } From 97431f29feb17ffc96ca95e9b3efec87be9d8b3a Mon Sep 17 00:00:00 2001 From: aria1th <35677394+aria1th@users.noreply.github.com> Date: Fri, 17 Nov 2023 10:05:28 +0900 Subject: [PATCH 103/342] fix double gc and decoding with unet context --- modules/processing.py | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index 2fda7f332..36c2be5e5 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -874,7 +874,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: else: if opts.sd_vae_decode_method != 'Full': p.extra_generation_params['VAE Decoder'] = opts.sd_vae_decode_method - with hypertile_context_unet(p.sd_model.model, aspect_ratio=p.width / p.height, tile_size=largest_tile_size_available(p.width, p.height), is_sdxl=shared.sd_model.is_sdxl, opts=shared.opts): + with hypertile_context_vae(p.sd_model.first_stage_model, aspect_ratio=p.width / p.height, tile_size=largest_tile_size_available(p.width, p.height), opts=shared.opts): x_samples_ddim = decode_latent_batch(p.sd_model, samples_ddim, target_device=devices.cpu, check_for_nans=True) x_samples_ddim = torch.stack(x_samples_ddim).float() @@ -1146,11 +1146,11 @@ class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing): tile_size = largest_tile_size_available(self.width, self.height) with hypertile_context_vae(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, opts=shared.opts): with hypertile_context_unet(self.sd_model.model, aspect_ratio=aspect_ratio, tile_size=tile_size, is_sdxl=shared.sd_model.is_sdxl, opts=shared.opts): - devices.torch_gc() samples = self.sampler.sample(self, x, conditioning, unconditional_conditioning, image_conditioning=self.txt2img_image_conditioning(x)) del x if not self.enable_hr: return samples + devices.torch_gc() if self.latent_scale_mode is None: with hypertile_context_vae(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, opts=shared.opts): @@ -1536,7 +1536,6 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): tile_size = largest_tile_size_available(self.width, self.height) with hypertile_context_vae(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, opts=shared.opts): with hypertile_context_unet(self.sd_model.model, aspect_ratio=aspect_ratio, tile_size=tile_size, is_sdxl=shared.sd_model.is_sdxl, opts=shared.opts): - devices.torch_gc() samples = self.sampler.sample_img2img(self, self.init_latent, x, conditioning, unconditional_conditioning, image_conditioning=self.image_conditioning) if self.mask is not None: From 4f2a4a361511ca3b8cdd7373f6c7d723583e8fdb Mon Sep 17 00:00:00 2001 From: storyicon Date: Fri, 17 Nov 2023 09:48:18 +0000 Subject: [PATCH 104/342] feat: fix randn found element of type float at pos 2 Signed-off-by: storyicon --- modules/rng.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/rng.py b/modules/rng.py index 9e8ba2ee9..8934d39bf 100644 --- a/modules/rng.py +++ b/modules/rng.py @@ -110,7 +110,7 @@ class ImageRNG: self.is_first = True def first(self): - noise_shape = self.shape if self.seed_resize_from_h <= 0 or self.seed_resize_from_w <= 0 else (self.shape[0], self.seed_resize_from_h // 8, self.seed_resize_from_w // 8) + noise_shape = self.shape if self.seed_resize_from_h <= 0 or self.seed_resize_from_w <= 0 else (self.shape[0], int(self.seed_resize_from_h) // 8, int(self.seed_resize_from_w // 8)) xs = [] From bde439ef67776be126d6a8c569a23d54dbc3e707 Mon Sep 17 00:00:00 2001 From: wfjsw Date: Sun, 19 Nov 2023 00:58:47 -0600 Subject: [PATCH 105/342] use metadata.ini for meta filename --- modules/extensions.py | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/modules/extensions.py b/modules/extensions.py index 5536db3ea..f3988d02e 100644 --- a/modules/extensions.py +++ b/modules/extensions.py @@ -43,13 +43,13 @@ class Extension: @functools.cached_property def metadata(self): - if os.path.isfile(os.path.join(self.path, "sd_webui_metadata.ini")): + if os.path.isfile(os.path.join(self.path, "metadata.ini")): try: config = configparser.ConfigParser() - config.read(os.path.join(self.path, "sd_webui_metadata.ini")) + config.read(os.path.join(self.path, "metadata.ini")) return config except Exception: - errors.report(f"Error reading sd_webui_metadata.ini for extension {self.canonical_name}.", + errors.report(f"Error reading metadata.ini for extension {self.canonical_name}.", exc_info=True) return None @@ -177,14 +177,14 @@ def list_extensions(): canonical_name = extension_dirname requires = None - if os.path.isfile(os.path.join(path, "sd_webui_metadata.ini")): + if os.path.isfile(os.path.join(path, "metadata.ini")): try: config = configparser.ConfigParser() - config.read(os.path.join(path, "sd_webui_metadata.ini")) + config.read(os.path.join(path, "metadata.ini")) canonical_name = config.get("Extension", "Name", fallback=canonical_name) requires = config.get("Extension", "Requires", fallback=None) except Exception: - errors.report(f"Error reading sd_webui_metadata.ini for extension {extension_dirname}. " + errors.report(f"Error reading metadata.ini for extension {extension_dirname}. " f"Will load regardless.", exc_info=True) canonical_name = canonical_name.lower().strip() From 598da5cd4928618b166886d3485ce30ce3a43490 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Sun, 19 Nov 2023 15:50:06 +0800 Subject: [PATCH 106/342] Use options instead of cmd_args --- modules/cmd_args.py | 2 -- modules/devices.py | 25 +++++++++------- modules/initialize_util.py | 1 + modules/sd_models.py | 61 ++++++++++++++++++++------------------ modules/shared_options.py | 1 + scripts/xyz_grid.py | 1 + 6 files changed, 49 insertions(+), 42 deletions(-) diff --git a/modules/cmd_args.py b/modules/cmd_args.py index 088d5deaa..a9fb9bfa3 100644 --- a/modules/cmd_args.py +++ b/modules/cmd_args.py @@ -118,5 +118,3 @@ parser.add_argument('--timeout-keep-alive', type=int, default=30, help='set time parser.add_argument("--disable-all-extensions", action='store_true', help="prevent all extensions from running regardless of any other settings", default=False) parser.add_argument("--disable-extra-extensions", action='store_true', help="prevent all extensions except built-in from running regardless of any other settings", default=False) parser.add_argument("--skip-load-model-at-start", action='store_true', help="if load a model at web start, only take effect when --nowebui", ) -parser.add_argument("--opt-unet-fp8-storage", action='store_true', help="use fp8 for SD UNet to save vram", default=False) -parser.add_argument("--opt-unet-fp8-storage-xl", action='store_true', help="use fp8 for SD UNet to save vram", default=False) diff --git a/modules/devices.py b/modules/devices.py index d7c905c28..03e7bdb7c 100644 --- a/modules/devices.py +++ b/modules/devices.py @@ -20,15 +20,15 @@ def cuda_no_autocast(device_id=None) -> bool: if device_id is None: device_id = get_cuda_device_id() return ( - torch.cuda.get_device_capability(device_id) == (7, 5) + torch.cuda.get_device_capability(device_id) == (7, 5) and torch.cuda.get_device_name(device_id).startswith("NVIDIA GeForce GTX 16") ) def get_cuda_device_id(): return ( - int(shared.cmd_opts.device_id) - if shared.cmd_opts.device_id is not None and shared.cmd_opts.device_id.isdigit() + int(shared.cmd_opts.device_id) + if shared.cmd_opts.device_id is not None and shared.cmd_opts.device_id.isdigit() else 0 ) or torch.cuda.current_device() @@ -116,16 +116,19 @@ patch_module_list = [ torch.nn.LayerNorm, ] + +def manual_cast_forward(self, *args, **kwargs): + org_dtype = next(self.parameters()).dtype + self.to(dtype) + args = [arg.to(dtype) if isinstance(arg, torch.Tensor) else arg for arg in args] + kwargs = {k: v.to(dtype) if isinstance(v, torch.Tensor) else v for k, v in kwargs.items()} + result = self.org_forward(*args, **kwargs) + self.to(org_dtype) + return result + + @contextlib.contextmanager def manual_autocast(): - def manual_cast_forward(self, *args, **kwargs): - org_dtype = next(self.parameters()).dtype - self.to(dtype) - args = [arg.to(dtype) if isinstance(arg, torch.Tensor) else arg for arg in args] - kwargs = {k: v.to(dtype) if isinstance(v, torch.Tensor) else v for k, v in kwargs.items()} - result = self.org_forward(*args, **kwargs) - self.to(org_dtype) - return result for module_type in patch_module_list: org_forward = module_type.forward module_type.forward = manual_cast_forward diff --git a/modules/initialize_util.py b/modules/initialize_util.py index 2e9b6d895..1b11ead61 100644 --- a/modules/initialize_util.py +++ b/modules/initialize_util.py @@ -177,6 +177,7 @@ def configure_opts_onchange(): shared.opts.onchange("temp_dir", ui_tempdir.on_tmpdir_changed) shared.opts.onchange("gradio_theme", shared.reload_gradio_theme) shared.opts.onchange("cross_attention_optimization", wrap_queued_call(lambda: sd_hijack.model_hijack.redo_hijack(shared.sd_model)), call=False) + shared.opts.onchange("fp8_storage", wrap_queued_call(lambda: sd_models.reload_model_weights()), call=False) startup_timer.record("opts onchange") diff --git a/modules/sd_models.py b/modules/sd_models.py index a6c8b2fa7..eb4914347 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -339,10 +339,28 @@ class SkipWritingToConfig: SkipWritingToConfig.skip = self.previous +def check_fp8(model): + if model is None: + return None + if devices.get_optimal_device_name() == "mps": + enable_fp8 = False + elif shared.opts.fp8_storage == "Enable": + enable_fp8 = True + elif getattr(model, "is_sdxl", False) and shared.opts.fp8_storage == "Enable for SDXL": + enable_fp8 = True + else: + enable_fp8 = False + return enable_fp8 + + def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer): sd_model_hash = checkpoint_info.calculate_shorthash() timer.record("calculate hash") + if not check_fp8(model) and devices.fp8: + # prevent model to load state dict in fp8 + model.half() + if not SkipWritingToConfig.skip: shared.opts.data["sd_model_checkpoint"] = checkpoint_info.title @@ -395,34 +413,16 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer devices.dtype_unet = torch.float16 timer.record("apply half()") - if devices.get_optimal_device_name() == "mps": - enable_fp8 = False - elif shared.cmd_opts.opt_unet_fp8_storage: - enable_fp8 = True - elif model.is_sdxl and shared.cmd_opts.opt_unet_fp8_storage_xl: - enable_fp8 = True - else: - enable_fp8 = False - - if enable_fp8: + if check_fp8(model): devices.fp8 = True - if model.is_sdxl: - cond_stage = model.conditioner - else: - cond_stage = model.cond_stage_model - - for module in cond_stage.modules(): - if isinstance(module, torch.nn.Linear): + first_stage = model.first_stage_model + model.first_stage_model = None + for module in model.modules(): + if isinstance(module, torch.nn.Conv2d): module.to(torch.float8_e4m3fn) - - if devices.device == devices.cpu: - for module in model.model.diffusion_model.modules(): - if isinstance(module, torch.nn.Conv2d): - module.to(torch.float8_e4m3fn) - elif isinstance(module, torch.nn.Linear): - module.to(torch.float8_e4m3fn) - else: - model.model.diffusion_model = model.model.diffusion_model.to(torch.float8_e4m3fn) + elif isinstance(module, torch.nn.Linear): + module.to(torch.float8_e4m3fn) + model.first_stage_model = first_stage timer.record("apply fp8") else: devices.fp8 = False @@ -769,7 +769,7 @@ def reuse_model_from_already_loaded(sd_model, checkpoint_info, timer): return None -def reload_model_weights(sd_model=None, info=None): +def reload_model_weights(sd_model=None, info=None, forced_reload=False): checkpoint_info = info or select_checkpoint() timer = Timer() @@ -781,11 +781,14 @@ def reload_model_weights(sd_model=None, info=None): current_checkpoint_info = None else: current_checkpoint_info = sd_model.sd_checkpoint_info - if sd_model.sd_model_checkpoint == checkpoint_info.filename: + if check_fp8(sd_model) != devices.fp8: + # load from state dict again to prevent extra numerical errors + forced_reload = True + elif sd_model.sd_model_checkpoint == checkpoint_info.filename: return sd_model sd_model = reuse_model_from_already_loaded(sd_model, checkpoint_info, timer) - if sd_model is not None and sd_model.sd_checkpoint_info.filename == checkpoint_info.filename: + if not forced_reload and sd_model is not None and sd_model.sd_checkpoint_info.filename == checkpoint_info.filename: return sd_model if sd_model is not None: diff --git a/modules/shared_options.py b/modules/shared_options.py index f1003f218..d27f35e96 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -200,6 +200,7 @@ options_templates.update(options_section(('optimizations', "Optimizations"), { "pad_cond_uncond": OptionInfo(False, "Pad prompt/negative prompt to be same length", infotext='Pad conds').info("improves performance when prompt and negative prompt have different lengths; changes seeds"), "persistent_cond_cache": OptionInfo(True, "Persistent cond cache").info("do not recalculate conds from prompts if prompts have not changed since previous calculation"), "batch_cond_uncond": OptionInfo(True, "Batch cond/uncond").info("do both conditional and unconditional denoising in one batch; uses a bit more VRAM during sampling, but improves speed; previously this was controlled by --always-batch-cond-uncond comandline argument"), + "fp8_storage": OptionInfo("Disable", "FP8 weight", gr.Dropdown, {"choices": ["Disable", "Enable for SDXL", "Enable"]}).info("Use FP8 to store Linear/Conv layers' weight. Require pytorch>=2.1.0."), })) options_templates.update(options_section(('compatibility', "Compatibility"), { diff --git a/scripts/xyz_grid.py b/scripts/xyz_grid.py index 0dc255bc4..b2250c04d 100644 --- a/scripts/xyz_grid.py +++ b/scripts/xyz_grid.py @@ -270,6 +270,7 @@ axis_options = [ AxisOption("Refiner checkpoint", str, apply_field('refiner_checkpoint'), format_value=format_remove_path, confirm=confirm_checkpoints_or_none, cost=1.0, choices=lambda: ['None'] + sorted(sd_models.checkpoints_list, key=str.casefold)), AxisOption("Refiner switch at", float, apply_field('refiner_switch_at')), AxisOption("RNG source", str, apply_override("randn_source"), choices=lambda: ["GPU", "CPU", "NV"]), + AxisOption("FP8 mode", str, apply_override("fp8_storage"), cost=0.9, choices=lambda: ["Disable", "Enable for SDXL", "Enable"]), ] From 890181e1d456b613bf60f6e8378dc68b39011af9 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Sun, 19 Nov 2023 15:54:39 +0800 Subject: [PATCH 107/342] Update the xformers/torch versions --- modules/errors.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/modules/errors.py b/modules/errors.py index 8c339464d..a3498c11f 100644 --- a/modules/errors.py +++ b/modules/errors.py @@ -93,8 +93,8 @@ def check_versions(): import torch import gradio - expected_torch_version = "2.0.0" - expected_xformers_version = "0.0.20" + expected_torch_version = "2.1.0" + expected_xformers_version = "0.0.22.post7" expected_gradio_version = "3.41.2" if version.parse(torch.__version__) < version.parse(expected_torch_version): From f383af2729ec2d1969200218577ab19dd78f7d48 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Sun, 19 Nov 2023 15:56:23 +0800 Subject: [PATCH 108/342] update xformers/torch versions --- modules/launch_utils.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/launch_utils.py b/modules/launch_utils.py index 636da679e..c225bbc18 100644 --- a/modules/launch_utils.py +++ b/modules/launch_utils.py @@ -312,7 +312,7 @@ def prepare_environment(): torch_command = os.environ.get('TORCH_COMMAND', f"pip install torch==2.1.0 torchvision==0.16.0 --extra-index-url {torch_index_url}") requirements_file = os.environ.get('REQS_FILE', "requirements_versions.txt") - xformers_package = os.environ.get('XFORMERS_PACKAGE', 'xformers==0.0.20') + xformers_package = os.environ.get('XFORMERS_PACKAGE', 'xformers==0.0.22.post7') clip_package = os.environ.get('CLIP_PACKAGE', "https://github.com/openai/CLIP/archive/d50d76daa670286dd6cacf3bcd80b5e4823fc8e1.zip") openclip_package = os.environ.get('OPENCLIP_PACKAGE', "https://github.com/mlfoundations/open_clip/archive/bb6e834e9c70d9c27d0dc3ecedeebeaeb1ffad6b.zip") From 043d2edcf6a543f236f1f3cb70ac72e7b3b357b6 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Sun, 19 Nov 2023 15:56:31 +0800 Subject: [PATCH 109/342] Better naming --- modules/devices.py | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/modules/devices.py b/modules/devices.py index 03e7bdb7c..c19a7f402 100644 --- a/modules/devices.py +++ b/modules/devices.py @@ -128,7 +128,7 @@ def manual_cast_forward(self, *args, **kwargs): @contextlib.contextmanager -def manual_autocast(): +def manual_cast(): for module_type in patch_module_list: org_forward = module_type.forward module_type.forward = manual_cast_forward @@ -148,10 +148,10 @@ def autocast(disable=False): return torch.autocast("cpu", dtype=torch.bfloat16, enabled=True) if fp8 and (dtype == torch.float32 or shared.cmd_opts.precision == "full" or cuda_no_autocast()): - return manual_autocast() + return manual_cast() if has_mps() and shared.cmd_opts.precision != "full": - return manual_autocast() + return manual_cast() if dtype == torch.float32 or shared.cmd_opts.precision == "full": return contextlib.nullcontext() From dea5e43c8359b663d5599efc99278c258747db61 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Sat, 18 Nov 2023 04:23:03 +0900 Subject: [PATCH 110/342] Option to show batch img2img results in UI shared.opts.img2img_batch_show_results_limit limit the number of images return to the UI for batch img2img default limit 32 0 no images are shown -1 unlimited, all images are shown --- modules/img2img.py | 24 ++++++++++++++++++++---- modules/shared_options.py | 1 + 2 files changed, 21 insertions(+), 4 deletions(-) diff --git a/modules/img2img.py b/modules/img2img.py index 52cb577a6..c583290a0 100644 --- a/modules/img2img.py +++ b/modules/img2img.py @@ -44,6 +44,8 @@ def process_batch(p, input_dir, output_dir, inpaint_mask_dir, args, to_scale=Fal steps = p.steps override_settings = p.override_settings sd_model_checkpoint_override = get_closet_checkpoint_match(override_settings.get("sd_model_checkpoint", None)) + batch_results = None + discard_further_results = False for i, image in enumerate(images): state.job = f"{i+1} out of {len(images)}" if state.skipped: @@ -127,7 +129,21 @@ def process_batch(p, input_dir, output_dir, inpaint_mask_dir, args, to_scale=Fal if proc is None: p.override_settings.pop('save_images_replace_action', None) - process_images(p) + proc = process_images(p) + + if not discard_further_results and proc: + if batch_results: + batch_results.images.extend(proc.images) + batch_results.infotexts.extend(proc.infotexts) + else: + batch_results = proc + + if 0 <= shared.opts.img2img_batch_show_results_limit < len(batch_results.images): + discard_further_results = True + batch_results.images = batch_results.images[:int(shared.opts.img2img_batch_show_results_limit)] + batch_results.infotexts = batch_results.infotexts[:int(shared.opts.img2img_batch_show_results_limit)] + + return batch_results def img2img(id_task: str, mode: int, prompt: str, negative_prompt: str, prompt_styles, init_img, sketch, init_img_with_mask, inpaint_color_sketch, inpaint_color_sketch_orig, init_img_inpaint, init_mask_inpaint, steps: int, sampler_name: str, mask_blur: int, mask_alpha: float, inpainting_fill: int, n_iter: int, batch_size: int, cfg_scale: float, image_cfg_scale: float, denoising_strength: float, selected_scale_tab: int, height: int, width: int, scale_by: float, resize_mode: int, inpaint_full_res: bool, inpaint_full_res_padding: int, inpainting_mask_invert: int, img2img_batch_input_dir: str, img2img_batch_output_dir: str, img2img_batch_inpaint_mask_dir: str, override_settings_texts, img2img_batch_use_png_info: bool, img2img_batch_png_info_props: list, img2img_batch_png_info_dir: str, request: gr.Request, *args): @@ -212,10 +228,10 @@ def img2img(id_task: str, mode: int, prompt: str, negative_prompt: str, prompt_s with closing(p): if is_batch: assert not shared.cmd_opts.hide_ui_dir_config, "Launched with --hide-ui-dir-config, batch img2img disabled" + processed = process_batch(p, img2img_batch_input_dir, img2img_batch_output_dir, img2img_batch_inpaint_mask_dir, args, to_scale=selected_scale_tab == 1, scale_by=scale_by, use_png_info=img2img_batch_use_png_info, png_info_props=img2img_batch_png_info_props, png_info_dir=img2img_batch_png_info_dir) - process_batch(p, img2img_batch_input_dir, img2img_batch_output_dir, img2img_batch_inpaint_mask_dir, args, to_scale=selected_scale_tab == 1, scale_by=scale_by, use_png_info=img2img_batch_use_png_info, png_info_props=img2img_batch_png_info_props, png_info_dir=img2img_batch_png_info_dir) - - processed = Processed(p, [], p.seed, "") + if processed is None: + processed = Processed(p, [], p.seed, "") else: processed = modules.scripts.scripts_img2img.run(p, *args) if processed is None: diff --git a/modules/shared_options.py b/modules/shared_options.py index d40db5306..1ee8c7ad1 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -189,6 +189,7 @@ options_templates.update(options_section(('img2img', "img2img"), { "img2img_inpaint_sketch_default_brush_color": OptionInfo("#ffffff", "Inpaint sketch initial brush color", ui_components.FormColorPicker, {}).info("default brush color of img2img inpaint sketch").needs_reload_ui(), "return_mask": OptionInfo(False, "For inpainting, include the greyscale mask in results for web"), "return_mask_composite": OptionInfo(False, "For inpainting, include masked composite in results for web"), + "img2img_batch_show_results_limit": OptionInfo(32, "Show the first N batch img2img results in UI", gr.Slider, {"minimum": -1, "maximum": 1000, "step": 1}).info('0: disable, -1: show all images. Too many images can cause lag'), })) options_templates.update(options_section(('optimizations', "Optimizations"), { From 6d337bf23dae990e7b6717da4d5f2e54f212685c Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Mon, 20 Nov 2023 01:38:31 +0900 Subject: [PATCH 111/342] save sysinfo as .json GitHub now allows uploading of .json files in issues --- modules/launch_utils.py | 2 +- modules/ui.py | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/modules/launch_utils.py b/modules/launch_utils.py index 8cdbafa50..264ec9ca6 100644 --- a/modules/launch_utils.py +++ b/modules/launch_utils.py @@ -441,7 +441,7 @@ def dump_sysinfo(): import datetime text = sysinfo.get() - filename = f"sysinfo-{datetime.datetime.utcnow().strftime('%Y-%m-%d-%H-%M')}.txt" + filename = f"sysinfo-{datetime.datetime.utcnow().strftime('%Y-%m-%d-%H-%M')}.json" with open(filename, "w", encoding="utf8") as file: file.write(text) diff --git a/modules/ui.py b/modules/ui.py index ba0d8542b..b82f3c5e8 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -1308,7 +1308,7 @@ def setup_ui_api(app): from fastapi.responses import PlainTextResponse text = sysinfo.get() - filename = f"sysinfo-{datetime.datetime.utcnow().strftime('%Y-%m-%d-%H-%M')}.txt" + filename = f"sysinfo-{datetime.datetime.utcnow().strftime('%Y-%m-%d-%H-%M')}.json" return PlainTextResponse(text, headers={'Content-Disposition': f'{"attachment" if attachment else "inline"}; filename="{filename}"'}) From b2e039d07bed76350120ff448964c907a3b5e4a3 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Mon, 20 Nov 2023 14:05:32 +0800 Subject: [PATCH 112/342] Update webui-macos-env.sh --- webui-macos-env.sh | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/webui-macos-env.sh b/webui-macos-env.sh index 24bc5c426..db7e8b1a0 100644 --- a/webui-macos-env.sh +++ b/webui-macos-env.sh @@ -11,7 +11,7 @@ fi export install_dir="$HOME" export COMMANDLINE_ARGS="--skip-torch-cuda-test --upcast-sampling --no-half-vae --use-cpu interrogate" -export TORCH_COMMAND="pip install torch==2.0.1 torchvision==0.15.2" +export TORCH_COMMAND="pip install torch==2.1.0 torchvision==0.16.0" export PYTORCH_ENABLE_MPS_FALLBACK=1 #################################################################### From 9b471436b2226458a767077707ea102e331b5d78 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Mon, 20 Nov 2023 14:47:09 +0300 Subject: [PATCH 113/342] rework extensions metadata: use custom sorter that doesn't mess the order as much and ignores cyclic errors, use classes with named fields instead of dictionaries, eliminate some duplicated code --- modules/extensions.py | 132 +++++++++++++++++---------------- modules/scripts.py | 165 +++++++++++++++++++----------------------- 2 files changed, 146 insertions(+), 151 deletions(-) diff --git a/modules/extensions.py b/modules/extensions.py index f3988d02e..1899cd529 100644 --- a/modules/extensions.py +++ b/modules/extensions.py @@ -1,5 +1,6 @@ +from __future__ import annotations + import configparser -import functools import os import threading import re @@ -8,7 +9,6 @@ from modules import shared, errors, cache, scripts from modules.gitpython_hack import Repo from modules.paths_internal import extensions_dir, extensions_builtin_dir, script_path # noqa: F401 -extensions = [] os.makedirs(extensions_dir, exist_ok=True) @@ -22,13 +22,56 @@ def active(): return [x for x in extensions if x.enabled] +class ExtensionMetadata: + filename = "metadata.ini" + config: configparser.ConfigParser + canonical_name: str + requires: list + + def __init__(self, path, canonical_name): + self.config = configparser.ConfigParser() + + filepath = os.path.join(path, self.filename) + if os.path.isfile(filepath): + try: + self.config.read(filepath) + except Exception: + errors.report(f"Error reading {self.filename} for extension {canonical_name}.", exc_info=True) + + self.canonical_name = self.config.get("Extension", "Name", fallback=canonical_name) + self.canonical_name = canonical_name.lower().strip() + + self.requires = self.get_script_requirements("Requires", "Extension") + + def get_script_requirements(self, field, section, extra_section=None): + """reads a list of requirements from the config; field is the name of the field in the ini file, + like Requires or Before, and section is the name of the [section] in the ini file; additionally, + reads more requirements from [extra_section] if specified.""" + + x = self.config.get(section, field, fallback='') + + if extra_section: + x = x + ', ' + self.config.get(extra_section, field, fallback='') + + return self.parse_list(x.lower()) + + def parse_list(self, text): + """converts a line from config ("ext1 ext2, ext3 ") into a python list (["ext1", "ext2", "ext3"])""" + + if not text: + return [] + + # both "," and " " are accepted as separator + return [x for x in re.split(r"[,\s]+", text.strip()) if x] + + class Extension: lock = threading.Lock() cached_fields = ['remote', 'commit_date', 'branch', 'commit_hash', 'version'] + metadata: ExtensionMetadata - def __init__(self, name, path, enabled=True, is_builtin=False, canonical_name=None): + def __init__(self, name, path, enabled=True, is_builtin=False, metadata=None): self.name = name - self.canonical_name = canonical_name or name.lower() self.path = path self.enabled = enabled self.status = '' @@ -40,18 +83,8 @@ class Extension: self.branch = None self.remote = None self.have_info_from_repo = False - - @functools.cached_property - def metadata(self): - if os.path.isfile(os.path.join(self.path, "metadata.ini")): - try: - config = configparser.ConfigParser() - config.read(os.path.join(self.path, "metadata.ini")) - return config - except Exception: - errors.report(f"Error reading metadata.ini for extension {self.canonical_name}.", - exc_info=True) - return None + self.metadata = metadata if metadata else ExtensionMetadata(self.path, name.lower()) + self.canonical_name = metadata.canonical_name def to_dict(self): return {x: getattr(self, x) for x in self.cached_fields} @@ -162,7 +195,7 @@ def list_extensions(): elif shared.opts.disable_all_extensions == "extra": print("*** \"Disable all extensions\" option was set, will only load built-in extensions ***") - extension_dependency_map = {} + loaded_extensions = {} # scan through extensions directory and load metadata for dirname in [extensions_builtin_dir, extensions_dir]: @@ -175,55 +208,30 @@ def list_extensions(): continue canonical_name = extension_dirname - requires = None - - if os.path.isfile(os.path.join(path, "metadata.ini")): - try: - config = configparser.ConfigParser() - config.read(os.path.join(path, "metadata.ini")) - canonical_name = config.get("Extension", "Name", fallback=canonical_name) - requires = config.get("Extension", "Requires", fallback=None) - except Exception: - errors.report(f"Error reading metadata.ini for extension {extension_dirname}. " - f"Will load regardless.", exc_info=True) - - canonical_name = canonical_name.lower().strip() + metadata = ExtensionMetadata(path, canonical_name) # check for duplicated canonical names - if canonical_name in extension_dependency_map: - errors.report(f"Duplicate canonical name \"{canonical_name}\" found in extensions " - f"\"{extension_dirname}\" and \"{extension_dependency_map[canonical_name]['dirname']}\". " - f"The current loading extension will be discarded.", exc_info=False) + already_loaded_extension = loaded_extensions.get(metadata.canonical_name) + if already_loaded_extension is not None: + errors.report(f'Duplicate canonical name "{canonical_name}" found in extensions "{extension_dirname}" and "{already_loaded_extension.name}". Former will be discarded.', exc_info=False) continue - # both "," and " " are accepted as separator - requires = list(filter(None, re.split(r"[,\s]+", requires.lower()))) if requires else [] - - extension_dependency_map[canonical_name] = { - "dirname": extension_dirname, - "path": path, - "requires": requires, - } + is_builtin = dirname == extensions_builtin_dir + extension = Extension(name=extension_dirname, path=path, enabled=extension_dirname not in shared.opts.disabled_extensions, is_builtin=is_builtin, metadata=metadata) + extensions.append(extension) + loaded_extensions[canonical_name] = extension # check for requirements - for (_, extension_data) in extension_dependency_map.items(): - dirname, path, requires = extension_data['dirname'], extension_data['path'], extension_data['requires'] - requirement_met = True - for req in requires: - if req not in extension_dependency_map: - errors.report(f"Extension \"{dirname}\" requires \"{req}\" which is not installed. " - f"The current loading extension will be discarded.", exc_info=False) - requirement_met = False - break - dep_dirname = extension_dependency_map[req]['dirname'] - if dep_dirname in shared.opts.disabled_extensions: - errors.report(f"Extension \"{dirname}\" requires \"{dep_dirname}\" which is disabled. " - f"The current loading extension will be discarded.", exc_info=False) - requirement_met = False - break + for extension in extensions: + for req in extension.metadata.requires: + required_extension = loaded_extensions.get(req) + if required_extension is None: + errors.report(f'Extension "{extension.name}" requires "{req}" which is not installed.', exc_info=False) + continue - is_builtin = dirname == extensions_builtin_dir - extension = Extension(name=dirname, path=path, - enabled=dirname not in shared.opts.disabled_extensions and requirement_met, - is_builtin=is_builtin) - extensions.append(extension) + if not extension.enabled: + errors.report(f'Extension "{extension.name}" requires "{required_extension.name}" which is disabled.', exc_info=False) + continue + + +extensions: list[Extension] = [] diff --git a/modules/scripts.py b/modules/scripts.py index b1f4504a5..b0689a23d 100644 --- a/modules/scripts.py +++ b/modules/scripts.py @@ -2,7 +2,6 @@ import os import re import sys import inspect -from graphlib import TopologicalSorter, CycleError from collections import namedtuple from dataclasses import dataclass @@ -312,27 +311,57 @@ scripts_data = [] postprocessing_scripts_data = [] ScriptClassData = namedtuple("ScriptClassData", ["script_class", "path", "basedir", "module"]) +def topological_sort(dependencies): + """Accepts a dictionary mapping name to its dependencies, returns a list of names ordered according to dependencies. + Ignores errors relating to missing dependeencies or circular dependencies + """ + + visited = {} + result = [] + + def inner(name): + visited[name] = True + + for dep in dependencies.get(name, []): + if dep in dependencies and dep not in visited: + inner(dep) + + result.append(name) + + for depname in dependencies: + if depname not in visited: + inner(depname) + + return result + + +@dataclass +class ScriptWithDependencies: + script_canonical_name: str + file: ScriptFile + requires: list + load_before: list + load_after: list + def list_scripts(scriptdirname, extension, *, include_extensions=True): - scripts_list = [] - script_dependency_map = {} + scripts = {} + + loaded_extensions = {ext.canonical_name: ext for ext in extensions.active()} + loaded_extensions_scripts = {ext.canonical_name: [] for ext in extensions.active()} # build script dependency map - root_script_basedir = os.path.join(paths.script_path, scriptdirname) if os.path.exists(root_script_basedir): for filename in sorted(os.listdir(root_script_basedir)): if not os.path.isfile(os.path.join(root_script_basedir, filename)): continue - script_dependency_map[filename] = { - "extension": None, - "extension_dirname": None, - "script_file": ScriptFile(paths.script_path, filename, os.path.join(root_script_basedir, filename)), - "requires": [], - "load_before": [], - "load_after": [], - } + if os.path.splitext(filename)[1].lower() != extension: + continue + + script_file = ScriptFile(paths.script_path, filename, os.path.join(root_script_basedir, filename)) + scripts[filename] = ScriptWithDependencies(filename, script_file, [], [], []) if include_extensions: for ext in extensions.active(): @@ -341,96 +370,54 @@ def list_scripts(scriptdirname, extension, *, include_extensions=True): if not os.path.isfile(extension_script.path): continue - script_canonical_name = ext.canonical_name + "/" + extension_script.filename - if ext.is_builtin: - script_canonical_name = "builtin/" + script_canonical_name + script_canonical_name = ("builtin/" if ext.is_builtin else "") + ext.canonical_name + "/" + extension_script.filename relative_path = scriptdirname + "/" + extension_script.filename - requires = '' - load_before = '' - load_after = '' + script = ScriptWithDependencies( + script_canonical_name=script_canonical_name, + file=extension_script, + requires=ext.metadata.get_script_requirements("Requires", relative_path, scriptdirname), + load_before=ext.metadata.get_script_requirements("Before", relative_path, scriptdirname), + load_after=ext.metadata.get_script_requirements("After", relative_path, scriptdirname), + ) - if ext.metadata is not None: - requires = ext.metadata.get(relative_path, "Requires", fallback='') - load_before = ext.metadata.get(relative_path, "Before", fallback='') - load_after = ext.metadata.get(relative_path, "After", fallback='') + scripts[script_canonical_name] = script + loaded_extensions_scripts[ext.canonical_name].append(script) - # propagate directory level metadata - requires = requires + ',' + ext.metadata.get(scriptdirname, "Requires", fallback='') - load_before = load_before + ',' + ext.metadata.get(scriptdirname, "Before", fallback='') - load_after = load_after + ',' + ext.metadata.get(scriptdirname, "After", fallback='') - - requires = list(filter(None, re.split(r"[,\s]+", requires.lower()))) if requires else [] - load_after = list(filter(None, re.split(r"[,\s]+", load_after.lower()))) if load_after else [] - load_before = list(filter(None, re.split(r"[,\s]+", load_before.lower()))) if load_before else [] - - script_dependency_map[script_canonical_name] = { - "extension": ext.canonical_name, - "extension_dirname": ext.name, - "script_file": extension_script, - "requires": requires, - "load_before": load_before, - "load_after": load_after, - } - - # resolve dependencies - - loaded_extensions = set() - for ext in extensions.active(): - loaded_extensions.add(ext.canonical_name) - - for script_canonical_name, script_data in script_dependency_map.items(): + for script_canonical_name, script in scripts.items(): # load before requires inverse dependency # in this case, append the script name into the load_after list of the specified script - for load_before_script in script_data['load_before']: + for load_before in script.load_before: # if this requires an individual script to be loaded before - if load_before_script in script_dependency_map: - script_dependency_map[load_before_script]['load_after'].append(script_canonical_name) - elif load_before_script in loaded_extensions: - for _, script_data2 in script_dependency_map.items(): - if script_data2['extension'] == load_before_script: - script_data2['load_after'].append(script_canonical_name) - break + other_script = scripts.get(load_before) + if other_script: + other_script.load_after.append(script_canonical_name) - # resolve extension name in load_after lists - for load_after_script in list(script_data['load_after']): - if load_after_script not in script_dependency_map and load_after_script in loaded_extensions: - script_data['load_after'].remove(load_after_script) - for script_canonical_name2, script_data2 in script_dependency_map.items(): - if script_data2['extension'] == load_after_script: - script_data['load_after'].append(script_canonical_name2) - break + # if this requires an extension + other_extension_scripts = loaded_extensions_scripts.get(load_before) + if other_extension_scripts: + for other_script in other_extension_scripts: + other_script.load_after.append(script_canonical_name) - # build the DAG - sorter = TopologicalSorter() - for script_canonical_name, script_data in script_dependency_map.items(): - requirement_met = True - for required_script in script_data['requires']: - # if this requires an individual script to be loaded - if required_script not in script_dependency_map and required_script not in loaded_extensions: - errors.report(f"Script \"{script_canonical_name}\" " - f"requires \"{required_script}\" to " - f"be loaded, but it is not. Skipping.", - exc_info=False) - requirement_met = False - break - if not requirement_met: - continue + # if After mentions an extension, remove it and instead add all of its scripts + for load_after in list(script.load_after): + if load_after not in scripts and load_after in loaded_extensions_scripts: + script.load_after.remove(load_after) - sorter.add(script_canonical_name, *script_data['load_after']) + for other_script in loaded_extensions_scripts.get(load_after, []): + script.load_after.append(other_script.script_canonical_name) - # sort the scripts - try: - ordered_script = sorter.static_order() - except CycleError: - errors.report("Cycle detected in script dependencies. Scripts will load in ascending order.", exc_info=True) - ordered_script = script_dependency_map.keys() + dependencies = {} - for script_canonical_name in ordered_script: - script_data = script_dependency_map[script_canonical_name] - scripts_list.append(script_data['script_file']) + for script_canonical_name, script in scripts.items(): + for required_script in script.requires: + if required_script not in scripts and required_script not in loaded_extensions: + errors.report(f'Script "{script_canonical_name}" requires "{required_script}" to be loaded, but it is not.', exc_info=False) - scripts_list = [x for x in scripts_list if os.path.splitext(x.path)[1].lower() == extension and os.path.isfile(x.path)] + dependencies[script_canonical_name] = script.load_after + + ordered_scripts = topological_sort(dependencies) + scripts_list = [scripts[script_canonical_name].file for script_canonical_name in ordered_scripts] return scripts_list From 314ae1535ea172fcdb0f5b3b2eecc5d4ce9112b5 Mon Sep 17 00:00:00 2001 From: Tom Haelbich Date: Mon, 20 Nov 2023 16:19:54 +0100 Subject: [PATCH 114/342] added option for default behavior of dir buttons --- modules/shared_options.py | 1 + modules/ui_extra_networks.py | 9 +++++++-- 2 files changed, 8 insertions(+), 2 deletions(-) diff --git a/modules/shared_options.py b/modules/shared_options.py index 00b273faa..1d2dca797 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -224,6 +224,7 @@ options_templates.update(options_section(('interrogate', "Interrogate"), { options_templates.update(options_section(('extra_networks', "Extra Networks"), { "extra_networks_show_hidden_directories": OptionInfo(True, "Show hidden directories").info("directory is hidden if its name starts with \".\"."), + "extra_networks_dir_button_function": OptionInfo(False, "Add a '/' to the beginning of directory buttons").info("Buttons will display the contents of the selected directory without acting as a search filter."), "extra_networks_hidden_models": OptionInfo("When searched", "Show cards for models in hidden directories", gr.Radio, {"choices": ["Always", "When searched", "Never"]}).info('"When searched" option will only show the item when the search string has 4 characters or more'), "extra_networks_default_multiplier": OptionInfo(1.0, "Default multiplier for extra networks", gr.Slider, {"minimum": 0.0, "maximum": 2.0, "step": 0.01}), "extra_networks_card_width": OptionInfo(0, "Card width for Extra Networks").info("in pixels"), diff --git a/modules/ui_extra_networks.py b/modules/ui_extra_networks.py index bd6732856..27a37295f 100644 --- a/modules/ui_extra_networks.py +++ b/modules/ui_extra_networks.py @@ -138,8 +138,13 @@ class ExtraNetworksPage: continue subdir = os.path.abspath(x)[len(parentdir):].replace("\\", "/") - if not subdir.startswith("/"): - subdir = "/" + subdir + + if shared.opts.extra_networks_dir_button_function: + if not subdir.startswith("/"): + subdir = "/" + subdir + else: + while subdir.startswith("/"): + subdir = subdir[1:] is_empty = len(os.listdir(x)) == 0 if not is_empty and not subdir.endswith("/"): From 58c19545c83fa6925c9ce2216ee64964eb5129ce Mon Sep 17 00:00:00 2001 From: hidenorly Date: Tue, 21 Nov 2023 01:13:53 +0900 Subject: [PATCH 115/342] Add FP32 fallback support on sd_vae_approx This tries to execute interpolate with FP32 if it failed. Background is that on some environment such as Mx chip MacOS devices, we get error as follows: ``` "torch/nn/functional.py", line 3931, in interpolate return torch._C._nn.upsample_nearest2d(input, output_size, scale_factors) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ RuntimeError: "upsample_nearest2d_channels_last" not implemented for 'Half' ``` In this case, ```--no-half``` doesn't help to solve. Therefore this commits add the FP32 fallback execution to solve it. Note that the submodule may require additional modifications. The following is the example modification on the other submodule. ```repositories/stable-diffusion-stability-ai/ldm/modules/diffusionmodules/openaimodel.py class Upsample(nn.Module): ..snip.. def forward(self, x): assert x.shape[1] == self.channels if self.dims == 3: x = F.interpolate( x, (x.shape[2], x.shape[3] * 2, x.shape[4] * 2), mode="nearest" ) else: try: x = F.interpolate(x, scale_factor=2, mode="nearest") except: x = F.interpolate(x.to(th.float32), scale_factor=2, mode="nearest").to(x.dtype) if self.use_conv: x = self.conv(x) return x ..snip.. ``` You can see the FP32 fallback execution as same as sd_vae_approx.py. --- modules/sd_vae_approx.py | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/modules/sd_vae_approx.py b/modules/sd_vae_approx.py index 3965e223e..8370493f9 100644 --- a/modules/sd_vae_approx.py +++ b/modules/sd_vae_approx.py @@ -21,7 +21,13 @@ class VAEApprox(nn.Module): def forward(self, x): extra = 11 - x = nn.functional.interpolate(x, (x.shape[2] * 2, x.shape[3] * 2)) + try: + x = nn.functional.interpolate(x, (x.shape[2] * 2, x.shape[3] * 2)) + except RuntimeError as e: + if "not implemented for" in str(e) and "Half" in str(e): + x = nn.functional.interpolate(x.to(torch.float32), (x.shape[2] * 2, x.shape[3] * 2)).to(x.dtype) + else: + print(f"An unexpected RuntimeError occurred: {str(e)}") x = nn.functional.pad(x, (extra, extra, extra, extra)) for layer in [self.conv1, self.conv2, self.conv3, self.conv4, self.conv5, self.conv6, self.conv7, self.conv8, ]: From 8aa51f682c17d85f4585b9471860224568d25e95 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Tue, 21 Nov 2023 08:32:00 +0300 Subject: [PATCH 116/342] fix [Bug]: (Dev Branch) Placing "Dimensions" first in "ui_reorder_list" prevents start #14047 --- modules/ui.py | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/modules/ui.py b/modules/ui.py index b82f3c5e8..08e0ad775 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -635,12 +635,6 @@ def create_ui(): scale_by.release(**on_change_args) button_update_resize_to.click(**on_change_args) - # the code below is meant to update the resolution label after the image in the image selection UI has changed. - # as it is now the event keeps firing continuously for inpaint edits, which ruins the page with constant requests. - # I assume this must be a gradio bug and for now we'll just do it for non-inpaint inputs. - for component in [init_img, sketch]: - component.change(fn=lambda: None, _js="updateImg2imgResizeToTextAfterChangingImage", inputs=[], outputs=[], show_progress=False) - tab_scale_to.select(fn=lambda: 0, inputs=[], outputs=[selected_scale_tab]) tab_scale_by.select(fn=lambda: 1, inputs=[], outputs=[selected_scale_tab]) @@ -701,6 +695,12 @@ def create_ui(): if category not in {"accordions"}: scripts.scripts_img2img.setup_ui_for_section(category) + # the code below is meant to update the resolution label after the image in the image selection UI has changed. + # as it is now the event keeps firing continuously for inpaint edits, which ruins the page with constant requests. + # I assume this must be a gradio bug and for now we'll just do it for non-inpaint inputs. + for component in [init_img, sketch]: + component.change(fn=lambda: None, _js="updateImg2imgResizeToTextAfterChangingImage", inputs=[], outputs=[], show_progress=False) + def select_img2img_tab(tab): return gr.update(visible=tab in [2, 3, 4]), gr.update(visible=tab == 3), From 370a77f8e78e65a8a1339289d684cb43df142f70 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Tue, 21 Nov 2023 19:59:34 +0800 Subject: [PATCH 117/342] Option for using fp16 weight when apply lora --- extensions-builtin/Lora/networks.py | 16 ++++++++++++---- modules/initialize_util.py | 1 + modules/sd_models.py | 14 +++++++++++--- modules/shared_options.py | 1 + 4 files changed, 25 insertions(+), 7 deletions(-) diff --git a/extensions-builtin/Lora/networks.py b/extensions-builtin/Lora/networks.py index 0170dbfbd..d22ed8438 100644 --- a/extensions-builtin/Lora/networks.py +++ b/extensions-builtin/Lora/networks.py @@ -388,18 +388,26 @@ def network_apply_weights(self: Union[torch.nn.Conv2d, torch.nn.Linear, torch.nn if module is not None and hasattr(self, 'weight'): try: with torch.no_grad(): - updown, ex_bias = module.calc_updown(self.weight) + if getattr(self, 'fp16_weight', None) is None: + weight = self.weight + bias = self.bias + else: + weight = self.fp16_weight.clone().to(self.weight.device) + bias = getattr(self, 'fp16_bias', None) + if bias is not None: + bias = bias.clone().to(self.bias.device) + updown, ex_bias = module.calc_updown(weight) - if len(self.weight.shape) == 4 and self.weight.shape[1] == 9: + if len(weight.shape) == 4 and weight.shape[1] == 9: # inpainting model. zero pad updown to make channel[1] 4 to 9 updown = torch.nn.functional.pad(updown, (0, 0, 0, 0, 0, 5)) - self.weight.copy_((self.weight.to(dtype=updown.dtype) + updown).to(dtype=self.weight.dtype)) + self.weight.copy_((weight.to(dtype=updown.dtype) + updown).to(dtype=self.weight.dtype)) if ex_bias is not None and hasattr(self, 'bias'): if self.bias is None: self.bias = torch.nn.Parameter(ex_bias).to(self.weight.dtype) else: - self.bias.copy_((self.bias.to(dtype=ex_bias.dtype) + ex_bias).to(dtype=self.bias.dtype)) + self.bias.copy_((bias + ex_bias).to(dtype=self.bias.dtype)) except RuntimeError as e: logging.debug(f"Network {net.name} layer {network_layer_name}: {e}") extra_network_lora.errors[net.name] = extra_network_lora.errors.get(net.name, 0) + 1 diff --git a/modules/initialize_util.py b/modules/initialize_util.py index 1b11ead61..7fb1d8d50 100644 --- a/modules/initialize_util.py +++ b/modules/initialize_util.py @@ -178,6 +178,7 @@ def configure_opts_onchange(): shared.opts.onchange("gradio_theme", shared.reload_gradio_theme) shared.opts.onchange("cross_attention_optimization", wrap_queued_call(lambda: sd_hijack.model_hijack.redo_hijack(shared.sd_model)), call=False) shared.opts.onchange("fp8_storage", wrap_queued_call(lambda: sd_models.reload_model_weights()), call=False) + shared.opts.onchange("cache_fp16_weight", wrap_queued_call(lambda: sd_models.reload_model_weights()), call=False) startup_timer.record("opts onchange") diff --git a/modules/sd_models.py b/modules/sd_models.py index eb4914347..0a7777f1c 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -413,14 +413,22 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer devices.dtype_unet = torch.float16 timer.record("apply half()") + for module in model.modules(): + if hasattr(module, 'fp16_weight'): + del module.fp16_weight + if hasattr(module, 'fp16_bias'): + del module.fp16_bias + if check_fp8(model): devices.fp8 = True first_stage = model.first_stage_model model.first_stage_model = None for module in model.modules(): - if isinstance(module, torch.nn.Conv2d): - module.to(torch.float8_e4m3fn) - elif isinstance(module, torch.nn.Linear): + if isinstance(module, (torch.nn.Conv2d, torch.nn.Linear)): + if shared.opts.cache_fp16_weight: + module.fp16_weight = module.weight.clone().half() + if module.bias is not None: + module.fp16_bias = module.bias.clone().half() module.to(torch.float8_e4m3fn) model.first_stage_model = first_stage timer.record("apply fp8") diff --git a/modules/shared_options.py b/modules/shared_options.py index d27f35e96..eaa9f1352 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -201,6 +201,7 @@ options_templates.update(options_section(('optimizations', "Optimizations"), { "persistent_cond_cache": OptionInfo(True, "Persistent cond cache").info("do not recalculate conds from prompts if prompts have not changed since previous calculation"), "batch_cond_uncond": OptionInfo(True, "Batch cond/uncond").info("do both conditional and unconditional denoising in one batch; uses a bit more VRAM during sampling, but improves speed; previously this was controlled by --always-batch-cond-uncond comandline argument"), "fp8_storage": OptionInfo("Disable", "FP8 weight", gr.Dropdown, {"choices": ["Disable", "Enable for SDXL", "Enable"]}).info("Use FP8 to store Linear/Conv layers' weight. Require pytorch>=2.1.0."), + "cache_fp16_weight": OptionInfo(False, "Cache FP16 weight for LoRA").info("Cache fp16 weight when enabling FP8, will increase the quality of LoRA. Use more system ram."), })) options_templates.update(options_section(('compatibility', "Compatibility"), { From f5d719d1f1baa775d838aa75d9af1971bcc78e8f Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Wed, 22 Nov 2023 01:45:56 +0800 Subject: [PATCH 118/342] Add forced reload for fp16 cache --- modules/initialize_util.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/initialize_util.py b/modules/initialize_util.py index 7fb1d8d50..b6767138d 100644 --- a/modules/initialize_util.py +++ b/modules/initialize_util.py @@ -178,7 +178,7 @@ def configure_opts_onchange(): shared.opts.onchange("gradio_theme", shared.reload_gradio_theme) shared.opts.onchange("cross_attention_optimization", wrap_queued_call(lambda: sd_hijack.model_hijack.redo_hijack(shared.sd_model)), call=False) shared.opts.onchange("fp8_storage", wrap_queued_call(lambda: sd_models.reload_model_weights()), call=False) - shared.opts.onchange("cache_fp16_weight", wrap_queued_call(lambda: sd_models.reload_model_weights()), call=False) + shared.opts.onchange("cache_fp16_weight", wrap_queued_call(lambda: sd_models.reload_model_weights(forced_reload=True)), call=False) startup_timer.record("opts onchange") From 8fe1e195228162a4510925de05015f361efa1087 Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Wed, 22 Nov 2023 18:01:34 +0200 Subject: [PATCH 119/342] Update ruff to 0.1.6 --- .github/workflows/on_pull_request.yaml | 2 +- pyproject.toml | 1 + 2 files changed, 2 insertions(+), 1 deletion(-) diff --git a/.github/workflows/on_pull_request.yaml b/.github/workflows/on_pull_request.yaml index 78e608ee9..9e44c806a 100644 --- a/.github/workflows/on_pull_request.yaml +++ b/.github/workflows/on_pull_request.yaml @@ -20,7 +20,7 @@ jobs: # not to have GHA download an (at the time of writing) 4 GB cache # of PyTorch and other dependencies. - name: Install Ruff - run: pip install ruff==0.0.272 + run: pip install ruff==0.1.6 - name: Run Ruff run: ruff . lint-js: diff --git a/pyproject.toml b/pyproject.toml index 80541a8f3..d03036e7d 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -16,6 +16,7 @@ exclude = [ ignore = [ "E501", # Line too long + "E721", # Do not compare types, use `isinstance` "E731", # Do not assign a `lambda` expression, use a `def` "I001", # Import block is un-sorted or un-formatted From 066afda2f6f650fe108d285a239d08d59d92590d Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Wed, 22 Nov 2023 18:02:39 +0200 Subject: [PATCH 120/342] Simplify restart_sampler (suggested by ruff) --- modules/sd_samplers_extra.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/sd_samplers_extra.py b/modules/sd_samplers_extra.py index 1b981ca80..72fd0aa5e 100644 --- a/modules/sd_samplers_extra.py +++ b/modules/sd_samplers_extra.py @@ -60,7 +60,7 @@ def restart_sampler(model, x, sigmas, extra_args=None, callback=None, disable=No sigma_restart = get_sigmas_karras(restart_steps, sigmas[min_idx].item(), sigmas[max_idx].item(), device=sigmas.device)[:-1] while restart_times > 0: restart_times -= 1 - step_list.extend([(old_sigma, new_sigma) for (old_sigma, new_sigma) in zip(sigma_restart[:-1], sigma_restart[1:])]) + step_list.extend(zip(sigma_restart[:-1], sigma_restart[1:])) last_sigma = None for old_sigma, new_sigma in tqdm.tqdm(step_list, disable=disable): From ac2a981c4f30d77cdb674948fe0e2aa7264a93e1 Mon Sep 17 00:00:00 2001 From: wfjsw Date: Wed, 22 Nov 2023 22:40:24 -0600 Subject: [PATCH 121/342] use extension name for determining an extension is installed in the index --- modules/ui_extensions.py | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-) diff --git a/modules/ui_extensions.py b/modules/ui_extensions.py index c0a73b573..b67088811 100644 --- a/modules/ui_extensions.py +++ b/modules/ui_extensions.py @@ -335,6 +335,11 @@ def normalize_git_url(url): return url +def get_extension_dirname_from_url(url): + *parts, last_part = url.split('/') + return normalize_git_url(last_part) + + def install_extension_from_url(dirname, url, branch_name=None): check_access() @@ -346,10 +351,7 @@ def install_extension_from_url(dirname, url, branch_name=None): assert url, 'No URL specified' if dirname is None or dirname == "": - *parts, last_part = url.split('/') - last_part = normalize_git_url(last_part) - - dirname = last_part + dirname = get_extension_dirname_from_url(url) target_dir = os.path.join(extensions.extensions_dir, dirname) assert not os.path.exists(target_dir), f'Extension directory already exists: {target_dir}' @@ -449,7 +451,7 @@ def get_date(info: dict, key): def refresh_available_extensions_from_data(hide_tags, sort_column, filter_text=""): extlist = available_extensions["extensions"] - installed_extension_urls = {normalize_git_url(extension.remote): extension.name for extension in extensions.extensions} + installed_extensions = {extension.name for extension in extensions.extensions} tags = available_extensions.get("tags", {}) tags_to_hide = set(hide_tags) @@ -482,7 +484,7 @@ def refresh_available_extensions_from_data(hide_tags, sort_column, filter_text=" if url is None: continue - existing = installed_extension_urls.get(normalize_git_url(url), None) + existing = get_extension_dirname_from_url(url) in installed_extensions extension_tags = extension_tags + ["installed"] if existing else extension_tags if any(x for x in extension_tags if x in tags_to_hide): From 86b99b1e98fcdd6e7e5f6017071944364e01e6ad Mon Sep 17 00:00:00 2001 From: Jabasukuriputo Wang Date: Fri, 24 Nov 2023 11:28:54 -0600 Subject: [PATCH 122/342] Move exception_records related methods to errors.py --- modules/errors.py | 18 ++++++++++++++++-- modules/sysinfo.py | 17 +---------------- 2 files changed, 17 insertions(+), 18 deletions(-) diff --git a/modules/errors.py b/modules/errors.py index 192cd8ffd..ac9f1ee5e 100644 --- a/modules/errors.py +++ b/modules/errors.py @@ -6,6 +6,21 @@ import traceback exception_records = [] +def format_traceback(tb): + return [[f"{x.filename}, line {x.lineno}, {x.name}", x.line] for x in traceback.extract_tb(tb)] + + +def format_exception(e, tb): + return {"exception": str(e), "traceback": format_traceback(tb)} + + +def get_exceptions(): + try: + return list(reversed(exception_records)) + except Exception as e: + return str(e) + + def record_exception(): _, e, tb = sys.exc_info() if e is None: @@ -14,8 +29,7 @@ def record_exception(): if exception_records and exception_records[-1] == e: return - from modules import sysinfo - exception_records.append(sysinfo.format_exception(e, tb)) + exception_records.append(format_exception(e, tb)) if len(exception_records) > 5: exception_records.pop(0) diff --git a/modules/sysinfo.py b/modules/sysinfo.py index 7d906e1fe..226b204d9 100644 --- a/modules/sysinfo.py +++ b/modules/sysinfo.py @@ -85,7 +85,7 @@ def get_dict(): "Checksum": checksum_token, "Commandline": sys.argv, "Torch env info": get_torch_sysinfo(), - "Exceptions": get_exceptions(), + "Exceptions": errors.get_exceptions(), "CPU": { "model": platform.processor(), "count logical": psutil.cpu_count(logical=True), @@ -105,21 +105,6 @@ def get_dict(): return res -def format_traceback(tb): - return [[f"{x.filename}, line {x.lineno}, {x.name}", x.line] for x in traceback.extract_tb(tb)] - - -def format_exception(e, tb): - return {"exception": str(e), "traceback": format_traceback(tb)} - - -def get_exceptions(): - try: - return list(reversed(errors.exception_records)) - except Exception as e: - return str(e) - - def get_environment(): return {k: os.environ[k] for k in sorted(os.environ) if k in environment_whitelist} From 5cedc8f9b2b51f392e7c8f5e29286466e3bee8d6 Mon Sep 17 00:00:00 2001 From: Jabasukuriputo Wang Date: Fri, 24 Nov 2023 11:30:30 -0600 Subject: [PATCH 123/342] remove traceback in sysinfo --- modules/sysinfo.py | 1 - 1 file changed, 1 deletion(-) diff --git a/modules/sysinfo.py b/modules/sysinfo.py index 226b204d9..1d058950a 100644 --- a/modules/sysinfo.py +++ b/modules/sysinfo.py @@ -1,7 +1,6 @@ import json import os import sys -import traceback import platform import hashlib From 40ac134c553ac824d4a96666bba14d550300daa5 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Sat, 25 Nov 2023 12:35:09 +0800 Subject: [PATCH 124/342] Fix pre-fp8 --- modules/sd_models.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/sd_models.py b/modules/sd_models.py index 0a7777f1c..90437c87a 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -357,7 +357,7 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer sd_model_hash = checkpoint_info.calculate_shorthash() timer.record("calculate hash") - if not check_fp8(model) and devices.fp8: + if devices.fp8: # prevent model to load state dict in fp8 model.half() From 3a9bf4ac10d99feb81b0e637417a108d3fa5ac06 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sun, 26 Nov 2023 08:29:12 +0300 Subject: [PATCH 125/342] move file --- {modules => extensions-builtin/hypertile}/hypertile.py | 0 1 file changed, 0 insertions(+), 0 deletions(-) rename {modules => extensions-builtin/hypertile}/hypertile.py (100%) diff --git a/modules/hypertile.py b/extensions-builtin/hypertile/hypertile.py similarity index 100% rename from modules/hypertile.py rename to extensions-builtin/hypertile/hypertile.py From d2e0c1ca132f4f0d98b77397a9f353d4ad8e7c4b Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sun, 26 Nov 2023 10:51:45 +0300 Subject: [PATCH 126/342] rework hypertile into a built-in extension --- README.md | 1 + extensions-builtin/hypertile/hypertile.py | 215 ++++++++---------- .../hypertile/scripts/hypertile_script.py | 73 ++++++ modules/processing.py | 37 ++- modules/shared_options.py | 8 - 5 files changed, 183 insertions(+), 151 deletions(-) create mode 100644 extensions-builtin/hypertile/scripts/hypertile_script.py diff --git a/README.md b/README.md index 25ba070e7..3b3f93adc 100644 --- a/README.md +++ b/README.md @@ -174,5 +174,6 @@ Licenses for borrowed code can be found in `Settings -> Licenses` screen, and al - TAESD - Ollin Boer Bohan - https://github.com/madebyollin/taesd - LyCORIS - KohakuBlueleaf - Restart sampling - lambertae - https://github.com/Newbeeer/diffusion_restart_sampling +- Hypertile - tfernd - https://github.com/tfernd/HyperTile - Initial Gradio script - posted on 4chan by an Anonymous user. Thank you Anonymous user. - (You) diff --git a/extensions-builtin/hypertile/hypertile.py b/extensions-builtin/hypertile/hypertile.py index be898fce4..a40c13118 100644 --- a/extensions-builtin/hypertile/hypertile.py +++ b/extensions-builtin/hypertile/hypertile.py @@ -1,10 +1,13 @@ """ Hypertile module for splitting attention layers in SD-1.5 U-Net and SD-1.5 VAE -Warn : The patch works well only if the input image has a width and height that are multiples of 128 -Author : @tfernd Github : https://github.com/tfernd/HyperTile +Warn: The patch works well only if the input image has a width and height that are multiples of 128 +Original author: @tfernd Github: https://github.com/tfernd/HyperTile """ from __future__ import annotations + +import functools +from dataclasses import dataclass from typing import Callable from typing_extensions import Literal @@ -18,6 +21,19 @@ import random from einops import rearrange + +@dataclass +class HypertileParams: + depth = 0 + layer_name = "" + tile_size: int = 0 + swap_size: int = 0 + aspect_ratio: float = 1.0 + forward = None + enabled = False + + + # TODO add SD-XL layers DEPTH_LAYERS = { 0: [ @@ -176,6 +192,7 @@ DEPTH_LAYERS_XL = { RNG_INSTANCE = random.Random() + def random_divisor(value: int, min_value: int, /, max_options: int = 1) -> int: """ Returns a random divisor of value that @@ -193,10 +210,13 @@ def random_divisor(value: int, min_value: int, /, max_options: int = 1) -> int: return ns[idx] + def set_hypertile_seed(seed: int) -> None: RNG_INSTANCE.seed(seed) -def largest_tile_size_available(width:int, height:int) -> int: + +@functools.cache +def largest_tile_size_available(width: int, height: int) -> int: """ Calculates the largest tile size available for a given width and height Tile size is always a power of 2 @@ -207,6 +227,7 @@ def largest_tile_size_available(width:int, height:int) -> int: largest_tile_size_available *= 2 return largest_tile_size_available + def iterative_closest_divisors(hw:int, aspect_ratio:float) -> tuple[int, int]: """ Finds h and w such that h*w = hw and h/w = aspect_ratio @@ -219,6 +240,7 @@ def iterative_closest_divisors(hw:int, aspect_ratio:float) -> tuple[int, int]: closest_pair = pairs[ratios.index(closest_ratio)] # closest pair of divisors to aspect_ratio return closest_pair + @cache def find_hw_candidates(hw:int, aspect_ratio:float) -> tuple[int, int]: """ @@ -240,132 +262,87 @@ def find_hw_candidates(hw:int, aspect_ratio:float) -> tuple[int, int]: w = int(w_candidate) return h, w -@contextmanager -def split_attention( - layer: nn.Module, - /, - aspect_ratio: float, # width/height - tile_size: int = 128, # 128 for VAE - swap_size: int = 1, # 1 for VAE - *, - disable: bool = False, - max_depth: Literal[0, 1, 2, 3] = 0, # ! Try 0 or 1 - scale_depth: bool = True, # scale the tile-size depending on the depth - is_sdxl: bool = False, # is the model SD-XL -): - # Hijacks AttnBlock from ldm and Attention from diffusers - if disable: - logging.info(f"Attention for {layer.__class__.__qualname__} not splitted") - yield - return +def self_attn_forward(params: HypertileParams, scale_depth=True) -> Callable: - latent_tile_size = max(128, tile_size) // 8 + @wraps(params.forward) + def wrapper(*args, **kwargs): + if not params.enabled: + return params.forward(*args, **kwargs) - def self_attn_forward(forward: Callable, depth: int, layer_name: str, module: nn.Module) -> Callable: - @wraps(forward) - def wrapper(*args, **kwargs): - x = args[0] + latent_tile_size = max(128, params.tile_size) // 8 + x = args[0] - # VAE - if x.ndim == 4: - b, c, h, w = x.shape + # VAE + if x.ndim == 4: + b, c, h, w = x.shape - nh = random_divisor(h, latent_tile_size, swap_size) - nw = random_divisor(w, latent_tile_size, swap_size) + nh = random_divisor(h, latent_tile_size, params.swap_size) + nw = random_divisor(w, latent_tile_size, params.swap_size) - if nh * nw > 1: - x = rearrange(x, "b c (nh h) (nw w) -> (b nh nw) c h w", nh=nh, nw=nw) # split into nh * nw tiles + if nh * nw > 1: + x = rearrange(x, "b c (nh h) (nw w) -> (b nh nw) c h w", nh=nh, nw=nw) # split into nh * nw tiles - out = forward(x, *args[1:], **kwargs) + out = params.forward(x, *args[1:], **kwargs) - if nh * nw > 1: - out = rearrange(out, "(b nh nw) c h w -> b c (nh h) (nw w)", nh=nh, nw=nw) + if nh * nw > 1: + out = rearrange(out, "(b nh nw) c h w -> b c (nh h) (nw w)", nh=nh, nw=nw) - # U-Net - else: - hw: int = x.size(1) - h, w = find_hw_candidates(hw, aspect_ratio) - assert h * w == hw, f"Invalid aspect ratio {aspect_ratio} for input of shape {x.shape}, hw={hw}, h={h}, w={w}" - - factor = 2**depth if scale_depth else 1 - nh = random_divisor(h, latent_tile_size * factor, swap_size) - nw = random_divisor(w, latent_tile_size * factor, swap_size) - - module._split_sizes_hypertile.append((nh, nw)) # type: ignore - - if nh * nw > 1: - x = rearrange(x, "b (nh h nw w) c -> (b nh nw) (h w) c", h=h // nh, w=w // nw, nh=nh, nw=nw) - - out = forward(x, *args[1:], **kwargs) - - if nh * nw > 1: - out = rearrange(out, "(b nh nw) hw c -> b nh nw hw c", nh=nh, nw=nw) - out = rearrange(out, "b nh nw (h w) c -> b (nh h nw w) c", h=h // nh, w=w // nw) - - return out - - return wrapper - - # Handle hijacking the forward method and recovering afterwards - try: - if is_sdxl: - layers = DEPTH_LAYERS_XL + # U-Net else: - layers = DEPTH_LAYERS - for depth in range(max_depth + 1): - for layer_name, module in layer.named_modules(): + hw: int = x.size(1) + h, w = find_hw_candidates(hw, params.aspect_ratio) + assert h * w == hw, f"Invalid aspect ratio {params.aspect_ratio} for input of shape {x.shape}, hw={hw}, h={h}, w={w}" + + factor = 2 ** params.depth if scale_depth else 1 + nh = random_divisor(h, latent_tile_size * factor, params.swap_size) + nw = random_divisor(w, latent_tile_size * factor, params.swap_size) + + if nh * nw > 1: + x = rearrange(x, "b (nh h nw w) c -> (b nh nw) (h w) c", h=h // nh, w=w // nw, nh=nh, nw=nw) + + out = params.forward(x, *args[1:], **kwargs) + + if nh * nw > 1: + out = rearrange(out, "(b nh nw) hw c -> b nh nw hw c", nh=nh, nw=nw) + out = rearrange(out, "b nh nw (h w) c -> b (nh h nw w) c", h=h // nh, w=w // nw) + + return out + + return wrapper + + +def hypertile_hook_model(model: nn.Module, width, height, *, enable=False, tile_size_max=128, swap_size=1, max_depth=3, is_sdxl=False): + hypertile_layers = getattr(model, "__webui_hypertile_layers", None) + if hypertile_layers is None: + if not enable: + return + + hypertile_layers = {} + layers = DEPTH_LAYERS_XL if is_sdxl else DEPTH_LAYERS + + for depth in range(4): + for layer_name, module in model.named_modules(): if any(layer_name.endswith(try_name) for try_name in layers[depth]): - # print input shape for debugging - logging.debug(f"HyperTile hijacking attention layer at depth {depth}: {layer_name}") - # hijack - module._original_forward_hypertile = module.forward - module.forward = self_attn_forward(module.forward, depth, layer_name, module) - module._split_sizes_hypertile = [] - yield - finally: - for layer_name, module in layer.named_modules(): - # remove hijack - if hasattr(module, "_original_forward_hypertile"): - if module._split_sizes_hypertile: - logging.debug(f"layer {layer_name} splitted with ({module._split_sizes_hypertile})") - # recover - module.forward = module._original_forward_hypertile - del module._original_forward_hypertile - del module._split_sizes_hypertile + params = HypertileParams() + module.__webui_hypertile_params = params + params.forward = module.forward + params.depth = depth + params.layer_name = layer_name + module.forward = self_attn_forward(params) -def hypertile_context_vae(model:nn.Module, aspect_ratio:float, tile_size:int, opts): - """ - Returns context manager for VAE - """ - enabled = opts.hypertile_split_vae_attn - swap_size = opts.hypertile_swap_size_vae - max_depth = opts.hypertile_max_depth_vae - tile_size_max = opts.hypertile_max_tile_vae - return split_attention( - model, - aspect_ratio=aspect_ratio, - tile_size=min(tile_size, tile_size_max), - swap_size=swap_size, - disable=not enabled, - max_depth=max_depth, - is_sdxl=False, - ) + hypertile_layers[layer_name] = 1 -def hypertile_context_unet(model:nn.Module, aspect_ratio:float, tile_size:int, opts, is_sdxl:bool): - """ - Returns context manager for U-Net - """ - enabled = opts.hypertile_split_unet_attn - swap_size = opts.hypertile_swap_size_unet - max_depth = opts.hypertile_max_depth_unet - tile_size_max = opts.hypertile_max_tile_unet - return split_attention( - model, - aspect_ratio=aspect_ratio, - tile_size=min(tile_size, tile_size_max), - swap_size=swap_size, - disable=not enabled, - max_depth=max_depth, - is_sdxl=is_sdxl, - ) + model.__webui_hypertile_layers = hypertile_layers + + aspect_ratio = width / height + tile_size = min(largest_tile_size_available(width, height), tile_size_max) + + for layer_name, module in model.named_modules(): + if layer_name in hypertile_layers: + params = module.__webui_hypertile_params + + params.tile_size = tile_size + params.swap_size = swap_size + params.aspect_ratio = aspect_ratio + params.enabled = enable and params.depth <= max_depth diff --git a/extensions-builtin/hypertile/scripts/hypertile_script.py b/extensions-builtin/hypertile/scripts/hypertile_script.py new file mode 100644 index 000000000..3cc29cd1f --- /dev/null +++ b/extensions-builtin/hypertile/scripts/hypertile_script.py @@ -0,0 +1,73 @@ +import hypertile +from modules import scripts, script_callbacks, shared + + +class ScriptHypertile(scripts.Script): + name = "Hypertile" + + def title(self): + return self.name + + def show(self, is_img2img): + return scripts.AlwaysVisible + + def process(self, p, *args): + hypertile.set_hypertile_seed(p.all_seeds[0]) + + configure_hypertile(p.width, p.height, enable_unet=shared.opts.hypertile_enable_unet) + + def before_hr(self, p, *args): + configure_hypertile(p.hr_upscale_to_x, p.hr_upscale_to_y, enable_unet=shared.opts.hypertile_enable_unet_secondpass or shared.opts.hypertile_enable_unet) + + +def configure_hypertile(width, height, enable_unet=True): + hypertile.hypertile_hook_model( + shared.sd_model.first_stage_model, + width, + height, + swap_size=shared.opts.hypertile_swap_size_vae, + max_depth=shared.opts.hypertile_max_depth_vae, + tile_size_max=shared.opts.hypertile_max_tile_vae, + enable=shared.opts.hypertile_enable_vae, + ) + + hypertile.hypertile_hook_model( + shared.sd_model.model, + width, + height, + swap_size=shared.opts.hypertile_swap_size_unet, + max_depth=shared.opts.hypertile_max_depth_unet, + tile_size_max=shared.opts.hypertile_max_tile_unet, + enable=enable_unet, + is_sdxl=shared.sd_model.is_sdxl + ) + + +def on_ui_settings(): + import gradio as gr + + options = { + "hypertile_explanation": shared.OptionHTML(""" + Hypertile optimizes the self-attention layer within U-Net and VAE models, + resulting in a reduction in computation time ranging from 1 to 4 times. The larger the generated image is, the greater the + benefit. + """), + + "hypertile_enable_unet": shared.OptionInfo(False, "Enable Hypertile U-Net").info("noticeable change in details of the generated picture; if enabled, overrides the setting below"), + "hypertile_enable_unet_secondpass": shared.OptionInfo(False, "Enable Hypertile U-Net for hires fix second pass"), + "hypertile_max_depth_unet": shared.OptionInfo(3, "Hypertile U-Net max depth", gr.Slider, {"minimum": 0, "maximum": 3, "step": 1}), + "hypertile_max_tile_unet": shared.OptionInfo(256, "Hypertile U-net max tile size", gr.Slider, {"minimum": 0, "maximum": 512, "step": 16}), + "hypertile_swap_size_unet": shared.OptionInfo(3, "Hypertile U-net swap size", gr.Slider, {"minimum": 0, "maximum": 6, "step": 1}), + + "hypertile_enable_vae": shared.OptionInfo(False, "Enable Hypertile VAE").info("minimal change in the generated picture"), + "hypertile_max_depth_vae": shared.OptionInfo(3, "Hypertile VAE max depth", gr.Slider, {"minimum": 0, "maximum": 3, "step": 1}), + "hypertile_max_tile_vae": shared.OptionInfo(128, "Hypertile VAE max tile size", gr.Slider, {"minimum": 0, "maximum": 512, "step": 16}), + "hypertile_swap_size_vae": shared.OptionInfo(3, "Hypertile VAE swap size ", gr.Slider, {"minimum": 0, "maximum": 6, "step": 1}), + } + + for name, opt in options.items(): + opt.section = ('hypertile', "Hypertile") + shared.opts.add_option(name, opt) + + +script_callbacks.on_ui_settings(on_ui_settings) diff --git a/modules/processing.py b/modules/processing.py index 36c2be5e5..ac58ef869 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -24,7 +24,6 @@ from modules.shared import opts, cmd_opts, state import modules.shared as shared import modules.paths as paths import modules.face_restoration -from modules.hypertile import set_hypertile_seed, largest_tile_size_available, hypertile_context_unet, hypertile_context_vae import modules.images as images import modules.styles import modules.sd_models as sd_models @@ -861,8 +860,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: p.comment(comment) p.extra_generation_params.update(model_hijack.extra_generation_params) - set_hypertile_seed(p.seed) - # add batch size + hypertile status to information to reproduce the run + if p.n_iter > 1: shared.state.job = f"Batch {n+1} out of {p.n_iter}" @@ -874,8 +872,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: else: if opts.sd_vae_decode_method != 'Full': p.extra_generation_params['VAE Decoder'] = opts.sd_vae_decode_method - with hypertile_context_vae(p.sd_model.first_stage_model, aspect_ratio=p.width / p.height, tile_size=largest_tile_size_available(p.width, p.height), opts=shared.opts): - x_samples_ddim = decode_latent_batch(p.sd_model, samples_ddim, target_device=devices.cpu, check_for_nans=True) + x_samples_ddim = decode_latent_batch(p.sd_model, samples_ddim, target_device=devices.cpu, check_for_nans=True) x_samples_ddim = torch.stack(x_samples_ddim).float() x_samples_ddim = torch.clamp((x_samples_ddim + 1.0) / 2.0, min=0.0, max=1.0) @@ -1141,25 +1138,23 @@ class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing): def sample(self, conditioning, unconditional_conditioning, seeds, subseeds, subseed_strength, prompts): self.sampler = sd_samplers.create_sampler(self.sampler_name, self.sd_model) - aspect_ratio = self.width / self.height + x = self.rng.next() - tile_size = largest_tile_size_available(self.width, self.height) - with hypertile_context_vae(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, opts=shared.opts): - with hypertile_context_unet(self.sd_model.model, aspect_ratio=aspect_ratio, tile_size=tile_size, is_sdxl=shared.sd_model.is_sdxl, opts=shared.opts): - samples = self.sampler.sample(self, x, conditioning, unconditional_conditioning, image_conditioning=self.txt2img_image_conditioning(x)) + samples = self.sampler.sample(self, x, conditioning, unconditional_conditioning, image_conditioning=self.txt2img_image_conditioning(x)) del x + if not self.enable_hr: return samples devices.torch_gc() if self.latent_scale_mode is None: - with hypertile_context_vae(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, opts=shared.opts): - decoded_samples = torch.stack(decode_latent_batch(self.sd_model, samples, target_device=devices.cpu, check_for_nans=True)).to(dtype=torch.float32) + decoded_samples = torch.stack(decode_latent_batch(self.sd_model, samples, target_device=devices.cpu, check_for_nans=True)).to(dtype=torch.float32) else: decoded_samples = None with sd_models.SkipWritingToConfig(): sd_models.reload_model_weights(info=self.hr_checkpoint_info) + return self.sample_hr_pass(samples, decoded_samples, seeds, subseeds, subseed_strength, prompts) def sample_hr_pass(self, samples, decoded_samples, seeds, subseeds, subseed_strength, prompts): @@ -1244,18 +1239,15 @@ class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing): if self.scripts is not None: self.scripts.before_hr(self) - tile_size = largest_tile_size_available(target_width, target_height) - aspect_ratio = self.width / self.height - with hypertile_context_vae(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, opts=shared.opts): - with hypertile_context_unet(self.sd_model.model, aspect_ratio=aspect_ratio, tile_size=tile_size, is_sdxl=shared.sd_model.is_sdxl, opts=shared.opts): - samples = self.sampler.sample_img2img(self, samples, noise, self.hr_c, self.hr_uc, steps=self.hr_second_pass_steps or self.steps, image_conditioning=image_conditioning) + + samples = self.sampler.sample_img2img(self, samples, noise, self.hr_c, self.hr_uc, steps=self.hr_second_pass_steps or self.steps, image_conditioning=image_conditioning) sd_models.apply_token_merging(self.sd_model, self.get_token_merging_ratio()) self.sampler = None devices.torch_gc() - with hypertile_context_vae(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, opts=shared.opts): - decoded_samples = decode_latent_batch(self.sd_model, samples, target_device=devices.cpu, check_for_nans=True) + + decoded_samples = decode_latent_batch(self.sd_model, samples, target_device=devices.cpu, check_for_nans=True) self.is_hr_pass = False return decoded_samples @@ -1532,11 +1524,8 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): if self.initial_noise_multiplier != 1.0: self.extra_generation_params["Noise multiplier"] = self.initial_noise_multiplier x *= self.initial_noise_multiplier - aspect_ratio = self.width / self.height - tile_size = largest_tile_size_available(self.width, self.height) - with hypertile_context_vae(self.sd_model.first_stage_model, aspect_ratio=aspect_ratio, tile_size=tile_size, opts=shared.opts): - with hypertile_context_unet(self.sd_model.model, aspect_ratio=aspect_ratio, tile_size=tile_size, is_sdxl=shared.sd_model.is_sdxl, opts=shared.opts): - samples = self.sampler.sample_img2img(self, self.init_latent, x, conditioning, unconditional_conditioning, image_conditioning=self.image_conditioning) + + samples = self.sampler.sample_img2img(self, self.init_latent, x, conditioning, unconditional_conditioning, image_conditioning=self.image_conditioning) if self.mask is not None: samples = samples * self.nmask + self.init_latent * self.mask diff --git a/modules/shared_options.py b/modules/shared_options.py index 28a489069..d40db5306 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -200,14 +200,6 @@ options_templates.update(options_section(('optimizations', "Optimizations"), { "pad_cond_uncond": OptionInfo(False, "Pad prompt/negative prompt to be same length", infotext='Pad conds').info("improves performance when prompt and negative prompt have different lengths; changes seeds"), "persistent_cond_cache": OptionInfo(True, "Persistent cond cache").info("do not recalculate conds from prompts if prompts have not changed since previous calculation"), "batch_cond_uncond": OptionInfo(True, "Batch cond/uncond").info("do both conditional and unconditional denoising in one batch; uses a bit more VRAM during sampling, but improves speed; previously this was controlled by --always-batch-cond-uncond comandline argument"), - "hypertile_split_unet_attn" : OptionInfo(False, "Split attention in Unet with HyperTile").link("Github", "https://github.com/tfernd/HyperTile").info("improves performance; changes behavior, but deterministic"), - "hypertile_split_vae_attn": OptionInfo(False, "Split attention in VAE with HyperTile").link("Github", "https://github.com/tfernd/HyperTile").info("improves performance; changes behavior, but deterministic"), - "hypertile_max_depth_vae" : OptionInfo(3, "Max depth for VAE HyperTile hijack", gr.Slider, {"minimum": 0, "maximum": 3, "step": 1}).link("Github", "https://github.com/tfernd/HyperTile"), - "hypertile_max_depth_unet" : OptionInfo(3, "Max depth for Unet HyperTile hijack", gr.Slider, {"minimum": 0, "maximum": 3, "step": 1}).link("Github", "https://github.com/tfernd/HyperTile"), - "hypertile_max_tile_vae" : OptionInfo(128, "Max tile size for VAE HyperTile hijack", gr.Slider, {"minimum": 0, "maximum": 512, "step": 16}).link("Github", "https://github.com/tfernd/HyperTile"), - "hypertile_max_tile_unet" : OptionInfo(256, "Max tile size for Unet HyperTile hijack", gr.Slider, {"minimum": 0, "maximum": 512, "step": 16}).link("Github", "https://github.com/tfernd/HyperTile"), - "hypertile_swap_size_unet": OptionInfo(3, "Swap size for Unet HyperTile hijack", gr.Slider, {"minimum": 0, "maximum": 6, "step": 1}).link("Github", "https://github.com/tfernd/HyperTile"), - "hypertile_swap_size_vae": OptionInfo(3, "Swap size for VAE HyperTile hijack", gr.Slider, {"minimum": 0, "maximum": 6, "step": 1}).link("Github", "https://github.com/tfernd/HyperTile"), })) options_templates.update(options_section(('compatibility', "Compatibility"), { From c5a0c59a83c950c64bc44427d3478aaa78c296cf Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sun, 26 Nov 2023 11:36:17 +0300 Subject: [PATCH 127/342] do not save HTML explanations from options page to config --- modules/options.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/modules/options.py b/modules/options.py index ab40aff73..7703d80ec 100644 --- a/modules/options.py +++ b/modules/options.py @@ -76,7 +76,7 @@ class Options: def __init__(self, data_labels: dict[str, OptionInfo], restricted_opts): self.data_labels = data_labels - self.data = {k: v.default for k, v in self.data_labels.items()} + self.data = {k: v.default for k, v in self.data_labels.items() if not v.do_not_save} self.restricted_opts = restricted_opts def __setattr__(self, key, value): @@ -210,7 +210,7 @@ class Options: def add_option(self, key, info): self.data_labels[key] = info - if key not in self.data: + if key not in self.data and not info.do_not_save: self.data[key] = info.default def reorder(self): From d1750e5eca6fd95db3516928cad18b32e557f56f Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sun, 26 Nov 2023 11:37:12 +0300 Subject: [PATCH 128/342] fix linter errors --- extensions-builtin/hypertile/hypertile.py | 3 --- 1 file changed, 3 deletions(-) diff --git a/extensions-builtin/hypertile/hypertile.py b/extensions-builtin/hypertile/hypertile.py index a40c13118..feb02fd27 100644 --- a/extensions-builtin/hypertile/hypertile.py +++ b/extensions-builtin/hypertile/hypertile.py @@ -9,11 +9,8 @@ from __future__ import annotations import functools from dataclasses import dataclass from typing import Callable -from typing_extensions import Literal -import logging from functools import wraps, cache -from contextlib import contextmanager import math import torch.nn as nn From 29f04149b60bcf6e8e2b41a161d6cc7e8981710f Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sun, 26 Nov 2023 12:07:33 +0300 Subject: [PATCH 129/342] update torch to 2.1.0 --- modules/errors.py | 4 ++-- modules/launch_utils.py | 6 +++--- webui-macos-env.sh | 2 +- 3 files changed, 6 insertions(+), 6 deletions(-) diff --git a/modules/errors.py b/modules/errors.py index eb234a838..c534a5d66 100644 --- a/modules/errors.py +++ b/modules/errors.py @@ -107,8 +107,8 @@ def check_versions(): import torch import gradio - expected_torch_version = "2.0.0" - expected_xformers_version = "0.0.20" + expected_torch_version = "2.1.0" + expected_xformers_version = "0.0.22.post7" expected_gradio_version = "3.41.2" if version.parse(torch.__version__) < version.parse(expected_torch_version): diff --git a/modules/launch_utils.py b/modules/launch_utils.py index 264ec9ca6..1f2b6c5e8 100644 --- a/modules/launch_utils.py +++ b/modules/launch_utils.py @@ -308,11 +308,11 @@ def requirements_met(requirements_file): def prepare_environment(): - torch_index_url = os.environ.get('TORCH_INDEX_URL', "https://download.pytorch.org/whl/cu118") - torch_command = os.environ.get('TORCH_COMMAND', f"pip install torch==2.0.1 torchvision==0.15.2 --extra-index-url {torch_index_url}") + torch_index_url = os.environ.get('TORCH_INDEX_URL', "https://download.pytorch.org/whl/cu121") + torch_command = os.environ.get('TORCH_COMMAND', f"pip install torch==2.1.0 torchvision==0.16.0 --extra-index-url {torch_index_url}") requirements_file = os.environ.get('REQS_FILE', "requirements_versions.txt") - xformers_package = os.environ.get('XFORMERS_PACKAGE', 'xformers==0.0.20') + xformers_package = os.environ.get('XFORMERS_PACKAGE', 'xformers==0.0.22.post7') clip_package = os.environ.get('CLIP_PACKAGE', "https://github.com/openai/CLIP/archive/d50d76daa670286dd6cacf3bcd80b5e4823fc8e1.zip") openclip_package = os.environ.get('OPENCLIP_PACKAGE', "https://github.com/mlfoundations/open_clip/archive/bb6e834e9c70d9c27d0dc3ecedeebeaeb1ffad6b.zip") diff --git a/webui-macos-env.sh b/webui-macos-env.sh index 24bc5c426..db7e8b1a0 100644 --- a/webui-macos-env.sh +++ b/webui-macos-env.sh @@ -11,7 +11,7 @@ fi export install_dir="$HOME" export COMMANDLINE_ARGS="--skip-torch-cuda-test --upcast-sampling --no-half-vae --use-cpu interrogate" -export TORCH_COMMAND="pip install torch==2.0.1 torchvision==0.15.2" +export TORCH_COMMAND="pip install torch==2.1.0 torchvision==0.16.0" export PYTORCH_ENABLE_MPS_FALLBACK=1 #################################################################### From 2a40d3c603448d15e209814366f2d6ab25e52398 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sun, 26 Nov 2023 14:58:47 +0300 Subject: [PATCH 130/342] compact prompt layout: preserve scroll when switching between lora tabs --- javascript/extraNetworks.js | 4 ++++ modules/ui_extra_networks.py | 5 ++++- style.css | 12 ++++++++++-- 3 files changed, 18 insertions(+), 3 deletions(-) diff --git a/javascript/extraNetworks.js b/javascript/extraNetworks.js index a1bf29a8c..a787372cf 100644 --- a/javascript/extraNetworks.js +++ b/javascript/extraNetworks.js @@ -130,6 +130,10 @@ function extraNetworksMovePromptToTab(tabname, id, showPrompt, showNegativePromp } else { promptContainer.insertBefore(prompt, promptContainer.firstChild); } + + if (elem) { + elem.classList.toggle('extra-page-prompts-active', showNegativePrompt || showPrompt); + } } diff --git a/modules/ui_extra_networks.py b/modules/ui_extra_networks.py index f03e20337..f3b23cc9c 100644 --- a/modules/ui_extra_networks.py +++ b/modules/ui_extra_networks.py @@ -370,6 +370,9 @@ def create_ui(interface: gr.Blocks, unrelated_tabs, tabname): for page in ui.stored_extra_pages: with gr.Tab(page.title, elem_id=f"{tabname}_{page.id_page}", elem_classes=["extra-page"]) as tab: + with gr.Column(elem_id=f"{tabname}_{page.id_page}_prompts", elem_classes=["extra-page-prompts"]): + pass + elem_id = f"{tabname}_{page.id_page}_cards_html" page_elem = gr.HTML('Loading...', elem_id=elem_id) ui.pages.append(page_elem) @@ -400,7 +403,7 @@ def create_ui(interface: gr.Blocks, unrelated_tabs, tabname): allow_prompt = "true" if page.allow_prompt else "false" allow_negative_prompt = "true" if page.allow_negative_prompt else "false" - jscode = 'extraNetworksTabSelected("' + tabname + '", "' + f"{tabname}_{page.id_page}" + '", ' + allow_prompt + ', ' + allow_negative_prompt + ');' + jscode = 'extraNetworksTabSelected("' + tabname + '", "' + f"{tabname}_{page.id_page}_prompts" + '", ' + allow_prompt + ', ' + allow_negative_prompt + ');' tab.select(fn=lambda: [gr.update(visible=True) for _ in tab_controls], _js='function(){ ' + jscode + ' }', inputs=[], outputs=tab_controls, show_progress=False) diff --git a/style.css b/style.css index 731620226..f8b42636d 100644 --- a/style.css +++ b/style.css @@ -840,8 +840,16 @@ footer { /* extra networks UI */ -.extra-page .prompt{ - margin: 0 0 0.5em 0; +.extra-page > div.gap{ + gap: 0; +} + +.extra-page-prompts{ + margin-bottom: 0; +} + +.extra-page-prompts.extra-page-prompts-active{ + margin-bottom: 1em; } .extra-network-cards{ From a15dd151ffb4d11556028b34561058bc44930427 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Sun, 26 Nov 2023 21:55:50 +0900 Subject: [PATCH 131/342] json.dump(ensure_ascii=False) improve json readability --- modules/cache.py | 2 +- modules/options.py | 2 +- modules/ui_extensions.py | 2 +- modules/ui_extra_networks_user_metadata.py | 2 +- modules/ui_loadsave.py | 2 +- 5 files changed, 5 insertions(+), 5 deletions(-) diff --git a/modules/cache.py b/modules/cache.py index ff26a2132..2d37e7b99 100644 --- a/modules/cache.py +++ b/modules/cache.py @@ -32,7 +32,7 @@ def dump_cache(): with cache_lock: cache_filename_tmp = cache_filename + "-" with open(cache_filename_tmp, "w", encoding="utf8") as file: - json.dump(cache_data, file, indent=4) + json.dump(cache_data, file, indent=4, ensure_ascii=False) os.replace(cache_filename_tmp, cache_filename) diff --git a/modules/options.py b/modules/options.py index 7703d80ec..40cb47991 100644 --- a/modules/options.py +++ b/modules/options.py @@ -158,7 +158,7 @@ class Options: assert not cmd_opts.freeze_settings, "saving settings is disabled" with open(filename, "w", encoding="utf8") as file: - json.dump(self.data, file, indent=4) + json.dump(self.data, file, indent=4, ensure_ascii=False) def same_type(self, x, y): if x is None or y is None: diff --git a/modules/ui_extensions.py b/modules/ui_extensions.py index c0a73b573..96dc9db2c 100644 --- a/modules/ui_extensions.py +++ b/modules/ui_extensions.py @@ -65,7 +65,7 @@ def save_config_state(name): filename = os.path.join(config_states_dir, f"{timestamp}_{name}.json") print(f"Saving backup of webui/extension state to {filename}.") with open(filename, "w", encoding="utf-8") as f: - json.dump(current_config_state, f, indent=4) + json.dump(current_config_state, f, indent=4, ensure_ascii=False) config_states.list_config_states() new_value = next(iter(config_states.all_config_states.keys()), "Current") new_choices = ["Current"] + list(config_states.all_config_states.keys()) diff --git a/modules/ui_extra_networks_user_metadata.py b/modules/ui_extra_networks_user_metadata.py index bfec140cc..36a807fcd 100644 --- a/modules/ui_extra_networks_user_metadata.py +++ b/modules/ui_extra_networks_user_metadata.py @@ -134,7 +134,7 @@ class UserMetadataEditor: basename, ext = os.path.splitext(filename) with open(basename + '.json', "w", encoding="utf8") as file: - json.dump(metadata, file, indent=4) + json.dump(metadata, file, indent=4, ensure_ascii=False) def save_user_metadata(self, name, desc, notes): user_metadata = self.get_user_metadata(name) diff --git a/modules/ui_loadsave.py b/modules/ui_loadsave.py index eb20ff258..7826786cc 100644 --- a/modules/ui_loadsave.py +++ b/modules/ui_loadsave.py @@ -141,7 +141,7 @@ class UiLoadsave: def write_to_file(self, current_ui_settings): with open(self.filename, "w", encoding="utf8") as file: - json.dump(current_ui_settings, file, indent=4) + json.dump(current_ui_settings, file, indent=4, ensure_ascii=False) def dump_defaults(self): """saves default values to a file unless tjhe file is present and there was an error loading default values at start""" From f0f100e67b78f686dc73cf3c8cad422e45cc9b8a Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sun, 26 Nov 2023 17:56:16 +0300 Subject: [PATCH 132/342] add categories to settings --- javascript/settings.js | 25 +++++++++++++ modules/options.py | 75 +++++++++++++++++++++++++++++++++++---- modules/shared_options.py | 49 ++++++++++++++----------- style.css | 9 +++++ 4 files changed, 130 insertions(+), 28 deletions(-) diff --git a/javascript/settings.js b/javascript/settings.js index 4e79ec003..e6009290a 100644 --- a/javascript/settings.js +++ b/javascript/settings.js @@ -44,3 +44,28 @@ onUiLoaded(function() { buttonShowAllPages.addEventListener("click", settingsShowAllTabs); }); + + +onOptionsChanged(function() { + if (gradioApp().querySelector('#settings .settings-category')) return; + + var sectionMap = {}; + gradioApp().querySelectorAll('#settings > div > button').forEach(function(x) { + sectionMap[x.textContent.trim()] = x; + }); + + opts._categories.forEach(function(x) { + var section = x[0]; + var category = x[1]; + + var span = document.createElement('SPAN'); + span.textContent = category; + span.className = 'settings-category'; + + var sectionElem = sectionMap[section]; + if (!sectionElem) return; + + sectionElem.parentElement.insertBefore(span, sectionElem); + }); +}); + diff --git a/modules/options.py b/modules/options.py index 40cb47991..4fead690c 100644 --- a/modules/options.py +++ b/modules/options.py @@ -1,5 +1,6 @@ import json import sys +from dataclasses import dataclass import gradio as gr @@ -8,13 +9,14 @@ from modules.shared_cmd_options import cmd_opts class OptionInfo: - def __init__(self, default=None, label="", component=None, component_args=None, onchange=None, section=None, refresh=None, comment_before='', comment_after='', infotext=None, restrict_api=False): + def __init__(self, default=None, label="", component=None, component_args=None, onchange=None, section=None, refresh=None, comment_before='', comment_after='', infotext=None, restrict_api=False, category_id=None): self.default = default self.label = label self.component = component self.component_args = component_args self.onchange = onchange self.section = section + self.category_id = category_id self.refresh = refresh self.do_not_save = False @@ -63,7 +65,11 @@ class OptionHTML(OptionInfo): def options_section(section_identifier, options_dict): for v in options_dict.values(): - v.section = section_identifier + if len(section_identifier) == 2: + v.section = section_identifier + elif len(section_identifier) == 3: + v.section = section_identifier[0:2] + v.category_id = section_identifier[2] return options_dict @@ -206,6 +212,17 @@ class Options: d = {k: self.data.get(k, v.default) for k, v in self.data_labels.items()} d["_comments_before"] = {k: v.comment_before for k, v in self.data_labels.items() if v.comment_before is not None} d["_comments_after"] = {k: v.comment_after for k, v in self.data_labels.items() if v.comment_after is not None} + + item_categories = {} + for item in self.data_labels.values(): + category = categories.mapping.get(item.category_id) + category = "Uncategorized" if category is None else category.label + if category not in item_categories: + item_categories[category] = item.section[1] + + # _categories is a list of pairs: [section, category]. Each section (a setting page) will get a special heading above it with the category as text. + d["_categories"] = [[v, k] for k, v in item_categories.items()] + [["Defaults", "Other"]] + return json.dumps(d) def add_option(self, key, info): @@ -214,15 +231,40 @@ class Options: self.data[key] = info.default def reorder(self): - """reorder settings so that all items related to section always go together""" + """Reorder settings so that: + - all items related to section always go together + - all sections belonging to a category go together + - sections inside a category are ordered alphabetically + - categories are ordered by creation order + + Category is a superset of sections: for category "postprocessing" there could be multiple sections: "face restoration", "upscaling". + + This function also changes items' category_id so that all items belonging to a section have the same category_id. + """ + + category_ids = {} + section_categories = {} - section_ids = {} settings_items = self.data_labels.items() for _, item in settings_items: - if item.section not in section_ids: - section_ids[item.section] = len(section_ids) + if item.section not in section_categories: + section_categories[item.section] = item.category_id - self.data_labels = dict(sorted(settings_items, key=lambda x: section_ids[x[1].section])) + for _, item in settings_items: + item.category_id = section_categories.get(item.section) + + for category_id in categories.mapping: + if category_id not in category_ids: + category_ids[category_id] = len(category_ids) + + def sort_key(x): + item: OptionInfo = x[1] + category_order = category_ids.get(item.category_id, len(category_ids)) + section_order = item.section[1] + + return category_order, section_order + + self.data_labels = dict(sorted(settings_items, key=sort_key)) def cast_value(self, key, value): """casts an arbitrary to the same type as this setting's value with key @@ -245,3 +287,22 @@ class Options: value = expected_type(value) return value + + +@dataclass +class OptionsCategory: + id: str + label: str + +class OptionsCategories: + def __init__(self): + self.mapping = {} + + def register_category(self, category_id, label): + if category_id in self.mapping: + return category_id + + self.mapping[category_id] = OptionsCategory(category_id, label) + + +categories = OptionsCategories() diff --git a/modules/shared_options.py b/modules/shared_options.py index 9bcd7914b..04e68a712 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -3,7 +3,7 @@ import gradio as gr from modules import localization, ui_components, shared_items, shared, interrogate, shared_gradio_themes from modules.paths_internal import models_path, script_path, data_path, sd_configs_path, sd_default_config, sd_model_file, default_sd_model_file, extensions_dir, extensions_builtin_dir # noqa: F401 from modules.shared_cmd_options import cmd_opts -from modules.options import options_section, OptionInfo, OptionHTML +from modules.options import options_section, OptionInfo, OptionHTML, categories options_templates = {} hide_dirs = shared.hide_dirs @@ -21,7 +21,14 @@ restricted_opts = { "outdir_init_images" } -options_templates.update(options_section(('saving-images', "Saving images/grids"), { +categories.register_category("saving", "Saving images") +categories.register_category("sd", "Stable Diffusion") +categories.register_category("ui", "User Interface") +categories.register_category("system", "System") +categories.register_category("postprocessing", "Postprocessing") +categories.register_category("training", "Training") + +options_templates.update(options_section(('saving-images', "Saving images/grids", "saving"), { "samples_save": OptionInfo(True, "Always save all generated images"), "samples_format": OptionInfo('png', 'File format for images'), "samples_filename_pattern": OptionInfo("", "Images filename pattern", component_args=hide_dirs).link("wiki", "https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Custom-Images-Filename-Name-and-Subdirectory"), @@ -67,7 +74,7 @@ options_templates.update(options_section(('saving-images', "Saving images/grids" "notification_volume": OptionInfo(100, "Notification sound volume", gr.Slider, {"minimum": 0, "maximum": 100, "step": 1}).info("in %"), })) -options_templates.update(options_section(('saving-paths', "Paths for saving"), { +options_templates.update(options_section(('saving-paths', "Paths for saving", "saving"), { "outdir_samples": OptionInfo("", "Output directory for images; if empty, defaults to three directories below", component_args=hide_dirs), "outdir_txt2img_samples": OptionInfo("outputs/txt2img-images", 'Output directory for txt2img images', component_args=hide_dirs), "outdir_img2img_samples": OptionInfo("outputs/img2img-images", 'Output directory for img2img images', component_args=hide_dirs), @@ -79,7 +86,7 @@ options_templates.update(options_section(('saving-paths', "Paths for saving"), { "outdir_init_images": OptionInfo("outputs/init-images", "Directory for saving init images when using img2img", component_args=hide_dirs), })) -options_templates.update(options_section(('saving-to-dirs', "Saving to a directory"), { +options_templates.update(options_section(('saving-to-dirs', "Saving to a directory", "saving"), { "save_to_dirs": OptionInfo(True, "Save images to a subdirectory"), "grid_save_to_dirs": OptionInfo(True, "Save grids to a subdirectory"), "use_save_to_dirs_for_ui": OptionInfo(False, "When using \"Save\" button, save images to a subdirectory"), @@ -87,21 +94,21 @@ options_templates.update(options_section(('saving-to-dirs', "Saving to a directo "directories_max_prompt_words": OptionInfo(8, "Max prompt words for [prompt_words] pattern", gr.Slider, {"minimum": 1, "maximum": 20, "step": 1, **hide_dirs}), })) -options_templates.update(options_section(('upscaling', "Upscaling"), { +options_templates.update(options_section(('upscaling', "Upscaling", "postprocessing"), { "ESRGAN_tile": OptionInfo(192, "Tile size for ESRGAN upscalers.", gr.Slider, {"minimum": 0, "maximum": 512, "step": 16}).info("0 = no tiling"), "ESRGAN_tile_overlap": OptionInfo(8, "Tile overlap for ESRGAN upscalers.", gr.Slider, {"minimum": 0, "maximum": 48, "step": 1}).info("Low values = visible seam"), "realesrgan_enabled_models": OptionInfo(["R-ESRGAN 4x+", "R-ESRGAN 4x+ Anime6B"], "Select which Real-ESRGAN models to show in the web UI.", gr.CheckboxGroup, lambda: {"choices": shared_items.realesrgan_models_names()}), "upscaler_for_img2img": OptionInfo(None, "Upscaler for img2img", gr.Dropdown, lambda: {"choices": [x.name for x in shared.sd_upscalers]}), })) -options_templates.update(options_section(('face-restoration', "Face restoration"), { +options_templates.update(options_section(('face-restoration', "Face restoration", "postprocessing"), { "face_restoration": OptionInfo(False, "Restore faces", infotext='Face restoration').info("will use a third-party model on generation result to reconstruct faces"), "face_restoration_model": OptionInfo("CodeFormer", "Face restoration model", gr.Radio, lambda: {"choices": [x.name() for x in shared.face_restorers]}), "code_former_weight": OptionInfo(0.5, "CodeFormer weight", gr.Slider, {"minimum": 0, "maximum": 1, "step": 0.01}).info("0 = maximum effect; 1 = minimum effect"), "face_restoration_unload": OptionInfo(False, "Move face restoration model from VRAM into RAM after processing"), })) -options_templates.update(options_section(('system', "System"), { +options_templates.update(options_section(('system', "System", "system"), { "auto_launch_browser": OptionInfo("Local", "Automatically open webui in browser on startup", gr.Radio, lambda: {"choices": ["Disable", "Local", "Remote"]}), "enable_console_prompts": OptionInfo(shared.cmd_opts.enable_console_prompts, "Print prompts to console when generating with txt2img and img2img."), "show_warnings": OptionInfo(False, "Show warnings in console.").needs_reload_ui(), @@ -116,13 +123,13 @@ options_templates.update(options_section(('system', "System"), { "dump_stacks_on_signal": OptionInfo(False, "Print stack traces before exiting the program with ctrl+c."), })) -options_templates.update(options_section(('API', "API"), { +options_templates.update(options_section(('API', "API", "system"), { "api_enable_requests": OptionInfo(True, "Allow http:// and https:// URLs for input images in API", restrict_api=True), "api_forbid_local_requests": OptionInfo(True, "Forbid URLs to local resources", restrict_api=True), "api_useragent": OptionInfo("", "User agent for requests", restrict_api=True), })) -options_templates.update(options_section(('training', "Training"), { +options_templates.update(options_section(('training', "Training", "training"), { "unload_models_when_training": OptionInfo(False, "Move VAE and CLIP to RAM when training if possible. Saves VRAM."), "pin_memory": OptionInfo(False, "Turn on pin_memory for DataLoader. Makes training slightly faster but can increase memory usage."), "save_optimizer_state": OptionInfo(False, "Saves Optimizer state as separate *.optim file. Training of embedding or HN can be resumed with the matching optim file."), @@ -137,7 +144,7 @@ options_templates.update(options_section(('training', "Training"), { "training_tensorboard_flush_every": OptionInfo(120, "How often, in seconds, to flush the pending tensorboard events and summaries to disk."), })) -options_templates.update(options_section(('sd', "Stable Diffusion"), { +options_templates.update(options_section(('sd', "Stable Diffusion", "sd"), { "sd_model_checkpoint": OptionInfo(None, "Stable Diffusion checkpoint", gr.Dropdown, lambda: {"choices": shared_items.list_checkpoint_tiles(shared.opts.sd_checkpoint_dropdown_use_short)}, refresh=shared_items.refresh_checkpoints, infotext='Model hash'), "sd_checkpoints_limit": OptionInfo(1, "Maximum number of checkpoints loaded at the same time", gr.Slider, {"minimum": 1, "maximum": 10, "step": 1}), "sd_checkpoints_keep_in_cpu": OptionInfo(True, "Only keep one model on device").info("will keep models other than the currently used one in RAM rather than VRAM"), @@ -154,14 +161,14 @@ options_templates.update(options_section(('sd', "Stable Diffusion"), { "hires_fix_refiner_pass": OptionInfo("second pass", "Hires fix: which pass to enable refiner for", gr.Radio, {"choices": ["first pass", "second pass", "both passes"]}, infotext="Hires refiner"), })) -options_templates.update(options_section(('sdxl', "Stable Diffusion XL"), { +options_templates.update(options_section(('sdxl', "Stable Diffusion XL", "sd"), { "sdxl_crop_top": OptionInfo(0, "crop top coordinate"), "sdxl_crop_left": OptionInfo(0, "crop left coordinate"), "sdxl_refiner_low_aesthetic_score": OptionInfo(2.5, "SDXL low aesthetic score", gr.Number).info("used for refiner model negative prompt"), "sdxl_refiner_high_aesthetic_score": OptionInfo(6.0, "SDXL high aesthetic score", gr.Number).info("used for refiner model prompt"), })) -options_templates.update(options_section(('vae', "VAE"), { +options_templates.update(options_section(('vae', "VAE", "sd"), { "sd_vae_explanation": OptionHTML(""" VAE is a neural network that transforms a standard RGB image into latent space representation and back. Latent space representation is what stable diffusion is working on during sampling @@ -176,7 +183,7 @@ For img2img, VAE is used to process user's input image before the sampling, and "sd_vae_decode_method": OptionInfo("Full", "VAE type for decode", gr.Radio, {"choices": ["Full", "TAESD"]}, infotext='VAE Decoder').info("method to decode latent to image"), })) -options_templates.update(options_section(('img2img', "img2img"), { +options_templates.update(options_section(('img2img', "img2img", "sd"), { "inpainting_mask_weight": OptionInfo(1.0, "Inpainting conditioning mask strength", gr.Slider, {"minimum": 0.0, "maximum": 1.0, "step": 0.01}, infotext='Conditional mask weight'), "initial_noise_multiplier": OptionInfo(1.0, "Noise multiplier for img2img", gr.Slider, {"minimum": 0.0, "maximum": 1.5, "step": 0.001}, infotext='Noise multiplier'), "img2img_extra_noise": OptionInfo(0.0, "Extra noise multiplier for img2img and hires fix", gr.Slider, {"minimum": 0.0, "maximum": 1.0, "step": 0.01}, infotext='Extra noise').info("0 = disabled (default); should be lower than denoising strength"), @@ -192,7 +199,7 @@ options_templates.update(options_section(('img2img', "img2img"), { "img2img_batch_show_results_limit": OptionInfo(32, "Show the first N batch img2img results in UI", gr.Slider, {"minimum": -1, "maximum": 1000, "step": 1}).info('0: disable, -1: show all images. Too many images can cause lag'), })) -options_templates.update(options_section(('optimizations', "Optimizations"), { +options_templates.update(options_section(('optimizations', "Optimizations", "sd"), { "cross_attention_optimization": OptionInfo("Automatic", "Cross attention optimization", gr.Dropdown, lambda: {"choices": shared_items.cross_attention_optimizations()}), "s_min_uncond": OptionInfo(0.0, "Negative Guidance minimum sigma", gr.Slider, {"minimum": 0.0, "maximum": 15.0, "step": 0.01}).link("PR", "https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/9177").info("skip negative prompt for some steps when the image is almost ready; 0=disable, higher=faster"), "token_merging_ratio": OptionInfo(0.0, "Token merging ratio", gr.Slider, {"minimum": 0.0, "maximum": 0.9, "step": 0.1}, infotext='Token merging ratio').link("PR", "https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/9256").info("0=disable, higher=faster"), @@ -203,7 +210,7 @@ options_templates.update(options_section(('optimizations', "Optimizations"), { "batch_cond_uncond": OptionInfo(True, "Batch cond/uncond").info("do both conditional and unconditional denoising in one batch; uses a bit more VRAM during sampling, but improves speed; previously this was controlled by --always-batch-cond-uncond comandline argument"), })) -options_templates.update(options_section(('compatibility', "Compatibility"), { +options_templates.update(options_section(('compatibility', "Compatibility", "sd"), { "use_old_emphasis_implementation": OptionInfo(False, "Use old emphasis implementation. Can be useful to reproduce old seeds."), "use_old_karras_scheduler_sigmas": OptionInfo(False, "Use old karras scheduler sigmas (0.1 to 10)."), "no_dpmpp_sde_batch_determinism": OptionInfo(False, "Do not make DPM++ SDE deterministic across different batch sizes."), @@ -228,7 +235,7 @@ options_templates.update(options_section(('interrogate', "Interrogate"), { "deepbooru_filter_tags": OptionInfo("", "deepbooru: filter out those tags").info("separate by comma"), })) -options_templates.update(options_section(('extra_networks', "Extra Networks"), { +options_templates.update(options_section(('extra_networks', "Extra Networks", "sd"), { "extra_networks_show_hidden_directories": OptionInfo(True, "Show hidden directories").info("directory is hidden if its name starts with \".\"."), "extra_networks_hidden_models": OptionInfo("When searched", "Show cards for models in hidden directories", gr.Radio, {"choices": ["Always", "When searched", "Never"]}).info('"When searched" option will only show the item when the search string has 4 characters or more'), "extra_networks_default_multiplier": OptionInfo(1.0, "Default multiplier for extra networks", gr.Slider, {"minimum": 0.0, "maximum": 2.0, "step": 0.01}), @@ -245,7 +252,7 @@ options_templates.update(options_section(('extra_networks', "Extra Networks"), { "sd_hypernetwork": OptionInfo("None", "Add hypernetwork to prompt", gr.Dropdown, lambda: {"choices": ["None", *shared.hypernetworks]}, refresh=shared_items.reload_hypernetworks), })) -options_templates.update(options_section(('ui', "User interface"), { +options_templates.update(options_section(('ui', "User interface", "ui"), { "localization": OptionInfo("None", "Localization", gr.Dropdown, lambda: {"choices": ["None"] + list(localization.localizations.keys())}, refresh=lambda: localization.list_localizations(cmd_opts.localizations_dir)).needs_reload_ui(), "gradio_theme": OptionInfo("Default", "Gradio theme", ui_components.DropdownEditable, lambda: {"choices": ["Default"] + shared_gradio_themes.gradio_hf_hub_themes}).info("you can also manually enter any of themes from the gallery.").needs_reload_ui(), "gradio_themes_cache": OptionInfo(True, "Cache gradio themes locally").info("disable to update the selected Gradio theme"), @@ -280,7 +287,7 @@ options_templates.update(options_section(('ui', "User interface"), { })) -options_templates.update(options_section(('infotext', "Infotext"), { +options_templates.update(options_section(('infotext', "Infotext", "ui"), { "add_model_hash_to_info": OptionInfo(True, "Add model hash to generation information"), "add_model_name_to_info": OptionInfo(True, "Add model name to generation information"), "add_user_name_to_info": OptionInfo(False, "Add user name to generation information when authenticated"), @@ -295,7 +302,7 @@ options_templates.update(options_section(('infotext', "Infotext"), { })) -options_templates.update(options_section(('ui', "Live previews"), { +options_templates.update(options_section(('ui', "Live previews", "ui"), { "show_progressbar": OptionInfo(True, "Show progressbar"), "live_previews_enable": OptionInfo(True, "Show live previews of the created image"), "live_previews_image_format": OptionInfo("png", "Live preview file format", gr.Radio, {"choices": ["jpeg", "png", "webp"]}), @@ -308,7 +315,7 @@ options_templates.update(options_section(('ui', "Live previews"), { "live_preview_fast_interrupt": OptionInfo(False, "Return image with chosen live preview method on interrupt").info("makes interrupts faster"), })) -options_templates.update(options_section(('sampler-params', "Sampler parameters"), { +options_templates.update(options_section(('sampler-params', "Sampler parameters", "sd"), { "hide_samplers": OptionInfo([], "Hide samplers in user interface", gr.CheckboxGroup, lambda: {"choices": [x.name for x in shared_items.list_samplers()]}).needs_reload_ui(), "eta_ddim": OptionInfo(0.0, "Eta for DDIM", gr.Slider, {"minimum": 0.0, "maximum": 1.0, "step": 0.01}, infotext='Eta DDIM').info("noise multiplier; higher = more unpredictable results"), "eta_ancestral": OptionInfo(1.0, "Eta for k-diffusion samplers", gr.Slider, {"minimum": 0.0, "maximum": 1.0, "step": 0.01}, infotext='Eta').info("noise multiplier; currently only applies to ancestral samplers (i.e. Euler a) and SDE samplers"), @@ -330,7 +337,7 @@ options_templates.update(options_section(('sampler-params', "Sampler parameters" 'uni_pc_lower_order_final': OptionInfo(True, "UniPC lower order final", infotext='UniPC lower order final'), })) -options_templates.update(options_section(('postprocessing', "Postprocessing"), { +options_templates.update(options_section(('postprocessing', "Postprocessing", "postprocessing"), { 'postprocessing_enable_in_main_ui': OptionInfo([], "Enable postprocessing operations in txt2img and img2img tabs", ui_components.DropdownMulti, lambda: {"choices": [x.name for x in shared_items.postprocessing_scripts()]}), 'postprocessing_operation_order': OptionInfo([], "Postprocessing operation order", ui_components.DropdownMulti, lambda: {"choices": [x.name for x in shared_items.postprocessing_scripts()]}), 'upscaling_max_images_in_cache': OptionInfo(5, "Maximum number of images in upscaling cache", gr.Slider, {"minimum": 0, "maximum": 10, "step": 1}), diff --git a/style.css b/style.css index f8b42636d..6e3ca8411 100644 --- a/style.css +++ b/style.css @@ -462,6 +462,15 @@ div.toprow-compact-tools{ padding: 4px; } +#settings > div.tab-nav .settings-category{ + display: block; + margin: 1em 0 0.25em 0; + font-weight: bold; + text-decoration: underline; + cursor: default; + user-select: none; +} + #settings_result{ height: 1.4em; margin: 0 1.2em; From 1f6844eb7e3a91639b2977d1e0cfbb9bf98baea7 Mon Sep 17 00:00:00 2001 From: Jabasukuriputo Wang Date: Sun, 26 Nov 2023 10:04:39 -0600 Subject: [PATCH 133/342] also consider extension url --- modules/ui_extensions.py | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/modules/ui_extensions.py b/modules/ui_extensions.py index b67088811..252e6ff2c 100644 --- a/modules/ui_extensions.py +++ b/modules/ui_extensions.py @@ -452,6 +452,7 @@ def get_date(info: dict, key): def refresh_available_extensions_from_data(hide_tags, sort_column, filter_text=""): extlist = available_extensions["extensions"] installed_extensions = {extension.name for extension in extensions.extensions} + installed_extension_urls = {normalize_git_url(extension.remote) for extension in extensions.extensions if extension.remote is not None} tags = available_extensions.get("tags", {}) tags_to_hide = set(hide_tags) @@ -484,7 +485,7 @@ def refresh_available_extensions_from_data(hide_tags, sort_column, filter_text=" if url is None: continue - existing = get_extension_dirname_from_url(url) in installed_extensions + existing = get_extension_dirname_from_url(url) in installed_extensions or normalize_git_url(url) in installed_extension_urls extension_tags = extension_tags + ["installed"] if existing else extension_tags if any(x for x in extension_tags if x in tags_to_hide): From b30cc87b786d32f2385cfecf40a2469ee3a96ab5 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Mon, 27 Nov 2023 13:15:17 +0900 Subject: [PATCH 134/342] add Block component creation callback --- modules/gradio_extensons.py | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/modules/gradio_extensons.py b/modules/gradio_extensons.py index e6b6835ad..7d88dc984 100644 --- a/modules/gradio_extensons.py +++ b/modules/gradio_extensons.py @@ -47,10 +47,20 @@ def Block_get_config(self): def BlockContext_init(self, *args, **kwargs): + if scripts.scripts_current is not None: + scripts.scripts_current.before_component(self, **kwargs) + + scripts.script_callbacks.before_component_callback(self, **kwargs) + res = original_BlockContext_init(self, *args, **kwargs) add_classes_to_gradio_component(self) + scripts.script_callbacks.after_component_callback(self, **kwargs) + + if scripts.scripts_current is not None: + scripts.scripts_current.after_component(self, **kwargs) + return res From 8a6e4bda21dddef3ab2e70a05d71b587b6c8b04b Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Mon, 27 Nov 2023 14:00:17 +0900 Subject: [PATCH 135/342] catch uncaught exception with ui creation scripts prevent total webui crash --- modules/scripts.py | 54 +++++++++++++++++++++++++--------------------- 1 file changed, 29 insertions(+), 25 deletions(-) diff --git a/modules/scripts.py b/modules/scripts.py index b0689a23d..961d032ce 100644 --- a/modules/scripts.py +++ b/modules/scripts.py @@ -570,40 +570,44 @@ class ScriptRunner: if controls is None: return - script.name = wrap_call(script.title, script.filename, "title", default=script.filename).lower() - api_args = [] + try: + script.name = wrap_call(script.title, script.filename, "title", default=script.filename).lower() + api_args = [] - for control in controls: - control.custom_script_source = os.path.basename(script.filename) + for control in controls: + control.custom_script_source = os.path.basename(script.filename) - arg_info = api_models.ScriptArg(label=control.label or "") + arg_info = api_models.ScriptArg(label=control.label or "") - for field in ("value", "minimum", "maximum", "step"): - v = getattr(control, field, None) - if v is not None: - setattr(arg_info, field, v) + for field in ("value", "minimum", "maximum", "step"): + v = getattr(control, field, None) + if v is not None: + setattr(arg_info, field, v) - choices = getattr(control, 'choices', None) # as of gradio 3.41, some items in choices are strings, and some are tuples where the first elem is the string - if choices is not None: - arg_info.choices = [x[0] if isinstance(x, tuple) else x for x in choices] + choices = getattr(control, 'choices', None) # as of gradio 3.41, some items in choices are strings, and some are tuples where the first elem is the string + if choices is not None: + arg_info.choices = [x[0] if isinstance(x, tuple) else x for x in choices] - api_args.append(arg_info) + api_args.append(arg_info) - script.api_info = api_models.ScriptInfo( - name=script.name, - is_img2img=script.is_img2img, - is_alwayson=script.alwayson, - args=api_args, - ) + script.api_info = api_models.ScriptInfo( + name=script.name, + is_img2img=script.is_img2img, + is_alwayson=script.alwayson, + args=api_args, + ) - if script.infotext_fields is not None: - self.infotext_fields += script.infotext_fields + if script.infotext_fields is not None: + self.infotext_fields += script.infotext_fields - if script.paste_field_names is not None: - self.paste_field_names += script.paste_field_names + if script.paste_field_names is not None: + self.paste_field_names += script.paste_field_names - self.inputs += controls - script.args_to = len(self.inputs) + self.inputs += controls + script.args_to = len(self.inputs) + + except Exception: + errors.report(f"Error creating UI for {script.name}: ", exc_info=True) def setup_ui_for_section(self, section, scriptlist=None): if scriptlist is None: From 9621ca4d64bbe59880d869b923e1572f1475a52b Mon Sep 17 00:00:00 2001 From: Charlie Joynt Date: Mon, 27 Nov 2023 11:39:50 +0000 Subject: [PATCH 136/342] Allow use of mutiple styles csv files --- modules/styles.py | 201 +++++++++++++++++++++++++++++++++++++++------- 1 file changed, 170 insertions(+), 31 deletions(-) diff --git a/modules/styles.py b/modules/styles.py index 0740fe1b1..974d3289b 100644 --- a/modules/styles.py +++ b/modules/styles.py @@ -1,4 +1,5 @@ import csv +import fnmatch import os import os.path import re @@ -10,6 +11,23 @@ class PromptStyle(typing.NamedTuple): name: str prompt: str negative_prompt: str + path: str = None + + +def clean_text(text: str) -> str: + """ + Iterating through a list of regular expressions and replacement strings, we + clean up the prompt and style text to make it easier to match against each + other. + """ + re_list = [ + ("multiple commas", re.compile("(,+\s+)+,?"), ", "), + ("multiple spaces", re.compile("\s{2,}"), " "), + ] + for _, regex, replace in re_list: + text = regex.sub(replace, text) + + return text.strip(", ") def merge_prompts(style_prompt: str, prompt: str) -> str: @@ -26,41 +44,64 @@ def apply_styles_to_prompt(prompt, styles): for style in styles: prompt = merge_prompts(style, prompt) - return prompt + return clean_text(prompt) -re_spaces = re.compile(" +") +def unwrap_style_text_from_prompt(style_text, prompt): + """ + Checks the prompt to see if the style text is wrapped around it. If so, + returns True plus the prompt text without the style text. Otherwise, returns + False with the original prompt. - -def extract_style_text_from_prompt(style_text, prompt): - stripped_prompt = re.sub(re_spaces, " ", prompt.strip()) - stripped_style_text = re.sub(re_spaces, " ", style_text.strip()) + Note that the "cleaned" version of the style text is only used for matching + purposes here. It isn't returned; the original style text is not modified. + """ + stripped_prompt = clean_text(prompt) + stripped_style_text = clean_text(style_text) if "{prompt}" in stripped_style_text: - left, right = stripped_style_text.split("{prompt}", 2) + # Work out whether the prompt is wrapped in the style text. If so, we + # return True and the "inner" prompt text that isn't part of the style. + try: + left, right = stripped_style_text.split("{prompt}", 2) + except ValueError as e: + # If the style text has multple "{prompt}"s, we can't split it into + # two parts. This is an error, but we can't do anything about it. + print("Unable to compare style text to prompt:`n{style_text}") + print(f"Error: {e}") + return False, prompt if stripped_prompt.startswith(left) and stripped_prompt.endswith(right): - prompt = stripped_prompt[len(left):len(stripped_prompt)-len(right)] + prompt = stripped_prompt[len(left) : len(stripped_prompt) - len(right)] return True, prompt else: + # Work out whether the given prompt ends with the style text. If so, we + # return True and the prompt text up to where the style text starts. if stripped_prompt.endswith(stripped_style_text): - prompt = stripped_prompt[:len(stripped_prompt)-len(stripped_style_text)] - - if prompt.endswith(', '): + prompt = stripped_prompt[: len(stripped_prompt) - len(stripped_style_text)] + if prompt.endswith(", "): prompt = prompt[:-2] - return True, prompt return False, prompt -def extract_style_from_prompts(style: PromptStyle, prompt, negative_prompt): +def extract_original_prompts(style: PromptStyle, prompt, negative_prompt): + """ + Takes a style and compares it to the prompt and negative prompt. If the style + matches, returns True plus the prompt and negative prompt with the style text + removed. Otherwise, returns False with the original prompt and negative prompt. + """ if not style.prompt and not style.negative_prompt: return False, prompt, negative_prompt - match_positive, extracted_positive = extract_style_text_from_prompt(style.prompt, prompt) + match_positive, extracted_positive = unwrap_style_text_from_prompt( + style.prompt, prompt + ) if not match_positive: return False, prompt, negative_prompt - match_negative, extracted_negative = extract_style_text_from_prompt(style.negative_prompt, negative_prompt) + match_negative, extracted_negative = unwrap_style_text_from_prompt( + style.negative_prompt, negative_prompt + ) if not match_negative: return False, prompt, negative_prompt @@ -69,25 +110,88 @@ def extract_style_from_prompts(style: PromptStyle, prompt, negative_prompt): class StyleDatabase: def __init__(self, path: str): - self.no_style = PromptStyle("None", "", "") + self.no_style = PromptStyle("None", "", "", None) self.styles = {} self.path = path + folder, file = os.path.split(self.path) + self.default_file = file.split("*")[0] + ".csv" + if self.default_file == ".csv": + self.default_file = "styles.csv" + self.default_path = os.path.join(folder, self.default_file) + + self.prompt_fields = [field for field in PromptStyle._fields if field != "path"] + self.reload() def reload(self): + """ + Clears the style database and reloads the styles from the CSV file(s) + matching the path used to initialize the database. + """ self.styles.clear() - if not os.path.exists(self.path): - return + path, filename = os.path.split(self.path) - with open(self.path, "r", encoding="utf-8-sig", newline='') as file: + if "*" in filename: + fileglob = filename.split("*")[0] + "*.csv" + filelist = [] + for file in os.listdir(path): + if fnmatch.fnmatch(file, fileglob): + filelist.append(file) + # Add a visible divider to the style list + half_len = round(len(file) / 2) + divider = f"{'-' * (20 - half_len)} {file.upper()}" + divider = f"{divider} {'-' * (40 - len(divider))}" + self.styles[divider] = PromptStyle( + f"{divider}", None, None, "do_not_save" + ) + # Add styles from this CSV file + self.load_from_csv(os.path.join(path, file)) + if len(filelist) == 0: + print(f"No styles found in {path} matching {fileglob}") + return + elif not os.path.exists(self.path): + print(f"Style database not found: {self.path}") + return + else: + self.load_from_csv(self.path) + + def load_from_csv(self, path: str): + with open(path, "r", encoding="utf-8-sig", newline="") as file: reader = csv.DictReader(file, skipinitialspace=True) for row in reader: + # Ignore empty rows or rows starting with a comment + if not row or row["name"].startswith("#"): + continue # Support loading old CSV format with "name, text"-columns prompt = row["prompt"] if "prompt" in row else row["text"] negative_prompt = row.get("negative_prompt", "") - self.styles[row["name"]] = PromptStyle(row["name"], prompt, negative_prompt) + # Add style to database + self.styles[row["name"]] = PromptStyle( + row["name"], prompt, negative_prompt, path + ) + + def get_style_paths(self) -> list(): + """ + Returns a list of all distinct paths, including the default path, of + files that styles are loaded from.""" + # Update any styles without a path to the default path + for style in list(self.styles.values()): + if not style.path: + self.styles[style.name] = style._replace(path=self.default_path) + + # Create a list of all distinct paths, including the default path + style_paths = set() + style_paths.add(self.default_path) + for _, style in self.styles.items(): + if style.path: + style_paths.add(style.path) + + # Remove any paths for styles that are just list dividers + style_paths.remove("do_not_save") + + return list(style_paths) def get_style_prompts(self, styles): return [self.styles.get(x, self.no_style).prompt for x in styles] @@ -96,20 +200,53 @@ class StyleDatabase: return [self.styles.get(x, self.no_style).negative_prompt for x in styles] def apply_styles_to_prompt(self, prompt, styles): - return apply_styles_to_prompt(prompt, [self.styles.get(x, self.no_style).prompt for x in styles]) + return apply_styles_to_prompt( + prompt, [self.styles.get(x, self.no_style).prompt for x in styles] + ) def apply_negative_styles_to_prompt(self, prompt, styles): - return apply_styles_to_prompt(prompt, [self.styles.get(x, self.no_style).negative_prompt for x in styles]) + return apply_styles_to_prompt( + prompt, [self.styles.get(x, self.no_style).negative_prompt for x in styles] + ) - def save_styles(self, path: str) -> None: - # Always keep a backup file around - if os.path.exists(path): - shutil.copy(path, f"{path}.bak") + def save_styles(self, path: str = None) -> None: + # The path argument is deprecated, but kept for backwards compatibility + _ = path - with open(path, "w", encoding="utf-8-sig", newline='') as file: - writer = csv.DictWriter(file, fieldnames=PromptStyle._fields) - writer.writeheader() - writer.writerows(style._asdict() for k, style in self.styles.items()) + # Update any styles without a path to the default path + for style in list(self.styles.values()): + if not style.path: + self.styles[style.name] = style._replace(path=self.default_path) + + # Create a list of all distinct paths, including the default path + style_paths = set() + style_paths.add(self.default_path) + for _, style in self.styles.items(): + if style.path: + style_paths.add(style.path) + + # Remove any paths for styles that are just list dividers + style_paths.remove("do_not_save") + + csv_names = [os.path.split(path)[1].lower() for path in style_paths] + + for style_path in style_paths: + # Always keep a backup file around + if os.path.exists(style_path): + shutil.copy(style_path, f"{style_path}.bak") + + # Write the styles to the CSV file + with open(style_path, "w", encoding="utf-8-sig", newline="") as file: + writer = csv.DictWriter(file, fieldnames=self.prompt_fields) + writer.writeheader() + for style in (s for s in self.styles.values() if s.path == style_path): + # Skip style list dividers, e.g. "STYLES.CSV" + if style.name.lower().strip("# ") in csv_names: + continue + # Write style fields, ignoring the path field + writer.writerow( + {k: v for k, v in style._asdict().items() if k != "path"} + ) def extract_styles_from_prompt(self, prompt, negative_prompt): extracted = [] @@ -120,7 +257,9 @@ class StyleDatabase: found_style = None for style in applicable_styles: - is_match, new_prompt, new_neg_prompt = extract_style_from_prompts(style, prompt, negative_prompt) + is_match, new_prompt, new_neg_prompt = extract_original_prompts( + style, prompt, negative_prompt + ) if is_match: found_style = style prompt = new_prompt From 1c64bb71402c2cd62ac98f936203437f0c4fcd02 Mon Sep 17 00:00:00 2001 From: MisterSeajay Date: Mon, 27 Nov 2023 11:57:27 +0000 Subject: [PATCH 137/342] bugfix for warning message (#6) --- modules/styles.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/styles.py b/modules/styles.py index 974d3289b..e73920c70 100644 --- a/modules/styles.py +++ b/modules/styles.py @@ -66,7 +66,7 @@ def unwrap_style_text_from_prompt(style_text, prompt): except ValueError as e: # If the style text has multple "{prompt}"s, we can't split it into # two parts. This is an error, but we can't do anything about it. - print("Unable to compare style text to prompt:`n{style_text}") + print(f"Unable to compare style text to prompt:`n{style_text}") print(f"Error: {e}") return False, prompt if stripped_prompt.startswith(left) and stripped_prompt.endswith(right): From a75314b41f938d1e598916ecdd0f14126ae1876b Mon Sep 17 00:00:00 2001 From: MisterSeajay Date: Mon, 27 Nov 2023 12:03:42 +0000 Subject: [PATCH 138/342] bugfix for warning message (#6) * bugfix for warning message * bugfix error message --- modules/styles.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/styles.py b/modules/styles.py index e73920c70..4d218cd7e 100644 --- a/modules/styles.py +++ b/modules/styles.py @@ -66,7 +66,7 @@ def unwrap_style_text_from_prompt(style_text, prompt): except ValueError as e: # If the style text has multple "{prompt}"s, we can't split it into # two parts. This is an error, but we can't do anything about it. - print(f"Unable to compare style text to prompt:`n{style_text}") + print(f"Unable to compare style text to prompt:\n{style_text}") print(f"Error: {e}") return False, prompt if stripped_prompt.startswith(left) and stripped_prompt.endswith(right): From 26a0c29587da428d27fd3e6a95491776ef66bbdd Mon Sep 17 00:00:00 2001 From: Charlie Joynt Date: Mon, 27 Nov 2023 11:39:50 +0000 Subject: [PATCH 139/342] Allow use of mutiple styles csv files * https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/14122 Fix edge case where style text has multiple {prompt} placeholders * https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/14005 --- modules/styles.py | 201 +++++++++++++++++++++++++++++++++++++++------- 1 file changed, 170 insertions(+), 31 deletions(-) diff --git a/modules/styles.py b/modules/styles.py index 0740fe1b1..4d218cd7e 100644 --- a/modules/styles.py +++ b/modules/styles.py @@ -1,4 +1,5 @@ import csv +import fnmatch import os import os.path import re @@ -10,6 +11,23 @@ class PromptStyle(typing.NamedTuple): name: str prompt: str negative_prompt: str + path: str = None + + +def clean_text(text: str) -> str: + """ + Iterating through a list of regular expressions and replacement strings, we + clean up the prompt and style text to make it easier to match against each + other. + """ + re_list = [ + ("multiple commas", re.compile("(,+\s+)+,?"), ", "), + ("multiple spaces", re.compile("\s{2,}"), " "), + ] + for _, regex, replace in re_list: + text = regex.sub(replace, text) + + return text.strip(", ") def merge_prompts(style_prompt: str, prompt: str) -> str: @@ -26,41 +44,64 @@ def apply_styles_to_prompt(prompt, styles): for style in styles: prompt = merge_prompts(style, prompt) - return prompt + return clean_text(prompt) -re_spaces = re.compile(" +") +def unwrap_style_text_from_prompt(style_text, prompt): + """ + Checks the prompt to see if the style text is wrapped around it. If so, + returns True plus the prompt text without the style text. Otherwise, returns + False with the original prompt. - -def extract_style_text_from_prompt(style_text, prompt): - stripped_prompt = re.sub(re_spaces, " ", prompt.strip()) - stripped_style_text = re.sub(re_spaces, " ", style_text.strip()) + Note that the "cleaned" version of the style text is only used for matching + purposes here. It isn't returned; the original style text is not modified. + """ + stripped_prompt = clean_text(prompt) + stripped_style_text = clean_text(style_text) if "{prompt}" in stripped_style_text: - left, right = stripped_style_text.split("{prompt}", 2) + # Work out whether the prompt is wrapped in the style text. If so, we + # return True and the "inner" prompt text that isn't part of the style. + try: + left, right = stripped_style_text.split("{prompt}", 2) + except ValueError as e: + # If the style text has multple "{prompt}"s, we can't split it into + # two parts. This is an error, but we can't do anything about it. + print(f"Unable to compare style text to prompt:\n{style_text}") + print(f"Error: {e}") + return False, prompt if stripped_prompt.startswith(left) and stripped_prompt.endswith(right): - prompt = stripped_prompt[len(left):len(stripped_prompt)-len(right)] + prompt = stripped_prompt[len(left) : len(stripped_prompt) - len(right)] return True, prompt else: + # Work out whether the given prompt ends with the style text. If so, we + # return True and the prompt text up to where the style text starts. if stripped_prompt.endswith(stripped_style_text): - prompt = stripped_prompt[:len(stripped_prompt)-len(stripped_style_text)] - - if prompt.endswith(', '): + prompt = stripped_prompt[: len(stripped_prompt) - len(stripped_style_text)] + if prompt.endswith(", "): prompt = prompt[:-2] - return True, prompt return False, prompt -def extract_style_from_prompts(style: PromptStyle, prompt, negative_prompt): +def extract_original_prompts(style: PromptStyle, prompt, negative_prompt): + """ + Takes a style and compares it to the prompt and negative prompt. If the style + matches, returns True plus the prompt and negative prompt with the style text + removed. Otherwise, returns False with the original prompt and negative prompt. + """ if not style.prompt and not style.negative_prompt: return False, prompt, negative_prompt - match_positive, extracted_positive = extract_style_text_from_prompt(style.prompt, prompt) + match_positive, extracted_positive = unwrap_style_text_from_prompt( + style.prompt, prompt + ) if not match_positive: return False, prompt, negative_prompt - match_negative, extracted_negative = extract_style_text_from_prompt(style.negative_prompt, negative_prompt) + match_negative, extracted_negative = unwrap_style_text_from_prompt( + style.negative_prompt, negative_prompt + ) if not match_negative: return False, prompt, negative_prompt @@ -69,25 +110,88 @@ def extract_style_from_prompts(style: PromptStyle, prompt, negative_prompt): class StyleDatabase: def __init__(self, path: str): - self.no_style = PromptStyle("None", "", "") + self.no_style = PromptStyle("None", "", "", None) self.styles = {} self.path = path + folder, file = os.path.split(self.path) + self.default_file = file.split("*")[0] + ".csv" + if self.default_file == ".csv": + self.default_file = "styles.csv" + self.default_path = os.path.join(folder, self.default_file) + + self.prompt_fields = [field for field in PromptStyle._fields if field != "path"] + self.reload() def reload(self): + """ + Clears the style database and reloads the styles from the CSV file(s) + matching the path used to initialize the database. + """ self.styles.clear() - if not os.path.exists(self.path): - return + path, filename = os.path.split(self.path) - with open(self.path, "r", encoding="utf-8-sig", newline='') as file: + if "*" in filename: + fileglob = filename.split("*")[0] + "*.csv" + filelist = [] + for file in os.listdir(path): + if fnmatch.fnmatch(file, fileglob): + filelist.append(file) + # Add a visible divider to the style list + half_len = round(len(file) / 2) + divider = f"{'-' * (20 - half_len)} {file.upper()}" + divider = f"{divider} {'-' * (40 - len(divider))}" + self.styles[divider] = PromptStyle( + f"{divider}", None, None, "do_not_save" + ) + # Add styles from this CSV file + self.load_from_csv(os.path.join(path, file)) + if len(filelist) == 0: + print(f"No styles found in {path} matching {fileglob}") + return + elif not os.path.exists(self.path): + print(f"Style database not found: {self.path}") + return + else: + self.load_from_csv(self.path) + + def load_from_csv(self, path: str): + with open(path, "r", encoding="utf-8-sig", newline="") as file: reader = csv.DictReader(file, skipinitialspace=True) for row in reader: + # Ignore empty rows or rows starting with a comment + if not row or row["name"].startswith("#"): + continue # Support loading old CSV format with "name, text"-columns prompt = row["prompt"] if "prompt" in row else row["text"] negative_prompt = row.get("negative_prompt", "") - self.styles[row["name"]] = PromptStyle(row["name"], prompt, negative_prompt) + # Add style to database + self.styles[row["name"]] = PromptStyle( + row["name"], prompt, negative_prompt, path + ) + + def get_style_paths(self) -> list(): + """ + Returns a list of all distinct paths, including the default path, of + files that styles are loaded from.""" + # Update any styles without a path to the default path + for style in list(self.styles.values()): + if not style.path: + self.styles[style.name] = style._replace(path=self.default_path) + + # Create a list of all distinct paths, including the default path + style_paths = set() + style_paths.add(self.default_path) + for _, style in self.styles.items(): + if style.path: + style_paths.add(style.path) + + # Remove any paths for styles that are just list dividers + style_paths.remove("do_not_save") + + return list(style_paths) def get_style_prompts(self, styles): return [self.styles.get(x, self.no_style).prompt for x in styles] @@ -96,20 +200,53 @@ class StyleDatabase: return [self.styles.get(x, self.no_style).negative_prompt for x in styles] def apply_styles_to_prompt(self, prompt, styles): - return apply_styles_to_prompt(prompt, [self.styles.get(x, self.no_style).prompt for x in styles]) + return apply_styles_to_prompt( + prompt, [self.styles.get(x, self.no_style).prompt for x in styles] + ) def apply_negative_styles_to_prompt(self, prompt, styles): - return apply_styles_to_prompt(prompt, [self.styles.get(x, self.no_style).negative_prompt for x in styles]) + return apply_styles_to_prompt( + prompt, [self.styles.get(x, self.no_style).negative_prompt for x in styles] + ) - def save_styles(self, path: str) -> None: - # Always keep a backup file around - if os.path.exists(path): - shutil.copy(path, f"{path}.bak") + def save_styles(self, path: str = None) -> None: + # The path argument is deprecated, but kept for backwards compatibility + _ = path - with open(path, "w", encoding="utf-8-sig", newline='') as file: - writer = csv.DictWriter(file, fieldnames=PromptStyle._fields) - writer.writeheader() - writer.writerows(style._asdict() for k, style in self.styles.items()) + # Update any styles without a path to the default path + for style in list(self.styles.values()): + if not style.path: + self.styles[style.name] = style._replace(path=self.default_path) + + # Create a list of all distinct paths, including the default path + style_paths = set() + style_paths.add(self.default_path) + for _, style in self.styles.items(): + if style.path: + style_paths.add(style.path) + + # Remove any paths for styles that are just list dividers + style_paths.remove("do_not_save") + + csv_names = [os.path.split(path)[1].lower() for path in style_paths] + + for style_path in style_paths: + # Always keep a backup file around + if os.path.exists(style_path): + shutil.copy(style_path, f"{style_path}.bak") + + # Write the styles to the CSV file + with open(style_path, "w", encoding="utf-8-sig", newline="") as file: + writer = csv.DictWriter(file, fieldnames=self.prompt_fields) + writer.writeheader() + for style in (s for s in self.styles.values() if s.path == style_path): + # Skip style list dividers, e.g. "STYLES.CSV" + if style.name.lower().strip("# ") in csv_names: + continue + # Write style fields, ignoring the path field + writer.writerow( + {k: v for k, v in style._asdict().items() if k != "path"} + ) def extract_styles_from_prompt(self, prompt, negative_prompt): extracted = [] @@ -120,7 +257,9 @@ class StyleDatabase: found_style = None for style in applicable_styles: - is_match, new_prompt, new_neg_prompt = extract_style_from_prompts(style, prompt, negative_prompt) + is_match, new_prompt, new_neg_prompt = extract_original_prompts( + style, prompt, negative_prompt + ) if is_match: found_style = style prompt = new_prompt From 23c36f59b4a423362d74f1ca2cc69871ae101e0e Mon Sep 17 00:00:00 2001 From: aria1th <35677394+aria1th@users.noreply.github.com> Date: Mon, 27 Nov 2023 21:10:26 +0900 Subject: [PATCH 140/342] Support XYZ scripts / split hires path from unet --- .../hypertile/scripts/hypertile_script.py | 11 ++-- .../hypertile/scripts/hypertile_xyz.py | 52 +++++++++++++++++++ 2 files changed, 60 insertions(+), 3 deletions(-) create mode 100644 extensions-builtin/hypertile/scripts/hypertile_xyz.py diff --git a/extensions-builtin/hypertile/scripts/hypertile_script.py b/extensions-builtin/hypertile/scripts/hypertile_script.py index 3cc29cd1f..b2413cc5f 100644 --- a/extensions-builtin/hypertile/scripts/hypertile_script.py +++ b/extensions-builtin/hypertile/scripts/hypertile_script.py @@ -1,5 +1,6 @@ import hypertile from modules import scripts, script_callbacks, shared +from scripts.hypertile_xyz import add_axis_options class ScriptHypertile(scripts.Script): @@ -17,7 +18,10 @@ class ScriptHypertile(scripts.Script): configure_hypertile(p.width, p.height, enable_unet=shared.opts.hypertile_enable_unet) def before_hr(self, p, *args): - configure_hypertile(p.hr_upscale_to_x, p.hr_upscale_to_y, enable_unet=shared.opts.hypertile_enable_unet_secondpass or shared.opts.hypertile_enable_unet) + # exclusive hypertile seed for the second pass + if not shared.opts.hypertile_enable_unet: + hypertile.set_hypertile_seed(p.all_seeds[0]) + configure_hypertile(p.hr_upscale_to_x, p.hr_upscale_to_y, enable_unet=shared.opts.hypertile_enable_unet_secondpass) def configure_hypertile(width, height, enable_unet=True): @@ -57,12 +61,12 @@ def on_ui_settings(): "hypertile_enable_unet_secondpass": shared.OptionInfo(False, "Enable Hypertile U-Net for hires fix second pass"), "hypertile_max_depth_unet": shared.OptionInfo(3, "Hypertile U-Net max depth", gr.Slider, {"minimum": 0, "maximum": 3, "step": 1}), "hypertile_max_tile_unet": shared.OptionInfo(256, "Hypertile U-net max tile size", gr.Slider, {"minimum": 0, "maximum": 512, "step": 16}), - "hypertile_swap_size_unet": shared.OptionInfo(3, "Hypertile U-net swap size", gr.Slider, {"minimum": 0, "maximum": 6, "step": 1}), + "hypertile_swap_size_unet": shared.OptionInfo(3, "Hypertile U-net swap size", gr.Slider, {"minimum": 0, "maximum": 64, "step": 1}), "hypertile_enable_vae": shared.OptionInfo(False, "Enable Hypertile VAE").info("minimal change in the generated picture"), "hypertile_max_depth_vae": shared.OptionInfo(3, "Hypertile VAE max depth", gr.Slider, {"minimum": 0, "maximum": 3, "step": 1}), "hypertile_max_tile_vae": shared.OptionInfo(128, "Hypertile VAE max tile size", gr.Slider, {"minimum": 0, "maximum": 512, "step": 16}), - "hypertile_swap_size_vae": shared.OptionInfo(3, "Hypertile VAE swap size ", gr.Slider, {"minimum": 0, "maximum": 6, "step": 1}), + "hypertile_swap_size_vae": shared.OptionInfo(3, "Hypertile VAE swap size ", gr.Slider, {"minimum": 0, "maximum": 64, "step": 1}), } for name, opt in options.items(): @@ -71,3 +75,4 @@ def on_ui_settings(): script_callbacks.on_ui_settings(on_ui_settings) +script_callbacks.on_before_ui(add_axis_options) \ No newline at end of file diff --git a/extensions-builtin/hypertile/scripts/hypertile_xyz.py b/extensions-builtin/hypertile/scripts/hypertile_xyz.py new file mode 100644 index 000000000..eaf7c8d7a --- /dev/null +++ b/extensions-builtin/hypertile/scripts/hypertile_xyz.py @@ -0,0 +1,52 @@ +from modules import scripts +xyz_grid = [x for x in scripts.scripts_data if x.script_class.__module__ == "xyz_grid.py"][0].module +from modules.shared import opts + +def int_applier(value_name:str, min_range:int = -1, max_range:int = -1): + """ + Returns a function that applies the given value to the given value_name in opts.data. + """ + # convert to int + def validate(value_name:str, value:str): + try: + value = int(value) + except: + raise ValueError(f"Value {value} for {value_name} is not an integer") + # validate value + if not min_range == -1: + assert value >= min_range, f"Value {value} for {value_name} must be greater than or equal to {min_range}" + if not max_range == -1: + assert value <= max_range, f"Value {value} for {value_name} must be less than or equal to {max_range}" + def apply_int(p, x, xs): + validate(value_name, x) + opts.data[value_name] = int(x) + return apply_int + +def bool_applier(value_name:str): + """ + Returns a function that applies the given value to the given value_name in opts.data. + """ + def validate(value_name:str, value:str): + assert value.lower() in ["true", "false"], f"Value {value} for {value_name} must be either true or false" + def apply_bool(p, x, xs): + validate(value_name, x) + value_boolean = x.lower() == "true" + opts.data[value_name] = value_boolean + return apply_bool + +def add_axis_options(): + extra_axis_options = [ + xyz_grid.AxisOption("[Hypertile] Unet First pass Enabled", str, bool_applier("hypertile_enable_unet"), choices=xyz_grid.boolean_choice(reverse=True)), + xyz_grid.AxisOption("[Hypertile] Unet Second pass Enabled", str, bool_applier("hypertile_enable_unet_secondpass"), choices=xyz_grid.boolean_choice(reverse=True)), + xyz_grid.AxisOption("[Hypertile] Unet Max Depth", int, int_applier("hypertile_max_depth_unet", 0, 3), choices=lambda: [str(x) for x in range(4)]), + xyz_grid.AxisOption("[Hypertile] Unet Max Tile Size", int, int_applier("hypertile_max_tile_unet", 0, 512)), + xyz_grid.AxisOption("[Hypertile] Unet Swap Size", int, int_applier("hypertile_swap_size_unet", 0, 64)), + xyz_grid.AxisOption("[Hypertile] VAE Enabled", str, bool_applier("hypertile_enable_vae"), choices=xyz_grid.boolean_choice(reverse=True)), + xyz_grid.AxisOption("[Hypertile] VAE Max Depth", int, int_applier("hypertile_max_depth_vae", 0, 3), choices=lambda: [str(x) for x in range(4)]), + xyz_grid.AxisOption("[Hypertile] VAE Max Tile Size", int, int_applier("hypertile_max_tile_vae", 0, 512)), + xyz_grid.AxisOption("[Hypertile] VAE Swap Size", int, int_applier("hypertile_swap_size_vae", 0, 64)), + ] + # check if the axis options have already been added + if any(set(opt.label for opt in extra_axis_options).intersection(set(opt.label for opt in xyz_grid.axis_options))): + return + xyz_grid.axis_options.extend(extra_axis_options) \ No newline at end of file From 601a7b4ce5b28efd29b1668c7b8b74fb6b62f6f3 Mon Sep 17 00:00:00 2001 From: aria1th <35677394+aria1th@users.noreply.github.com> Date: Mon, 27 Nov 2023 22:10:31 +0900 Subject: [PATCH 141/342] cache divisors / fix ruff --- extensions-builtin/hypertile/hypertile.py | 24 ++++++++++++------- .../hypertile/scripts/hypertile_script.py | 2 +- .../hypertile/scripts/hypertile_xyz.py | 18 +++++++------- 3 files changed, 26 insertions(+), 18 deletions(-) diff --git a/extensions-builtin/hypertile/hypertile.py b/extensions-builtin/hypertile/hypertile.py index feb02fd27..0f40e2d39 100644 --- a/extensions-builtin/hypertile/hypertile.py +++ b/extensions-builtin/hypertile/hypertile.py @@ -6,7 +6,6 @@ Original author: @tfernd Github: https://github.com/tfernd/HyperTile from __future__ import annotations -import functools from dataclasses import dataclass from typing import Callable @@ -189,6 +188,19 @@ DEPTH_LAYERS_XL = { RNG_INSTANCE = random.Random() +@cache +def get_divisors(value: int, min_value: int, /, max_options: int = 1) -> list[int]: + """ + Returns divisors of value that + x * min_value <= value + in big -> small order, amount of divisors is limited by max_options + """ + max_options = max(1, max_options) # at least 1 option should be returned + min_value = min(min_value, value) + divisors = [i for i in range(min_value, value + 1) if value % i == 0] # divisors in small -> big order + ns = [value // i for i in divisors[:max_options]] # has at least 1 element # big -> small order + return ns + def random_divisor(value: int, min_value: int, /, max_options: int = 1) -> int: """ @@ -196,13 +208,7 @@ def random_divisor(value: int, min_value: int, /, max_options: int = 1) -> int: x * min_value <= value if max_options is 1, the behavior is deterministic """ - min_value = min(min_value, value) - - # All big divisors of value (inclusive) - divisors = [i for i in range(min_value, value + 1) if value % i == 0] # divisors in small -> big order - - ns = [value // i for i in divisors[:max_options]] # has at least 1 element # big -> small order - + ns = get_divisors(value, min_value, max_options=max_options) # get cached divisors idx = RNG_INSTANCE.randint(0, len(ns) - 1) return ns[idx] @@ -212,7 +218,7 @@ def set_hypertile_seed(seed: int) -> None: RNG_INSTANCE.seed(seed) -@functools.cache +@cache def largest_tile_size_available(width: int, height: int) -> int: """ Calculates the largest tile size available for a given width and height diff --git a/extensions-builtin/hypertile/scripts/hypertile_script.py b/extensions-builtin/hypertile/scripts/hypertile_script.py index b2413cc5f..d3ab60915 100644 --- a/extensions-builtin/hypertile/scripts/hypertile_script.py +++ b/extensions-builtin/hypertile/scripts/hypertile_script.py @@ -75,4 +75,4 @@ def on_ui_settings(): script_callbacks.on_ui_settings(on_ui_settings) -script_callbacks.on_before_ui(add_axis_options) \ No newline at end of file +script_callbacks.on_before_ui(add_axis_options) diff --git a/extensions-builtin/hypertile/scripts/hypertile_xyz.py b/extensions-builtin/hypertile/scripts/hypertile_xyz.py index eaf7c8d7a..3007a0832 100644 --- a/extensions-builtin/hypertile/scripts/hypertile_xyz.py +++ b/extensions-builtin/hypertile/scripts/hypertile_xyz.py @@ -1,17 +1,17 @@ from modules import scripts -xyz_grid = [x for x in scripts.scripts_data if x.script_class.__module__ == "xyz_grid.py"][0].module from modules.shared import opts +xyz_grid = [x for x in scripts.scripts_data if x.script_class.__module__ == "xyz_grid.py"][0].module + def int_applier(value_name:str, min_range:int = -1, max_range:int = -1): """ Returns a function that applies the given value to the given value_name in opts.data. """ # convert to int def validate(value_name:str, value:str): - try: - value = int(value) - except: - raise ValueError(f"Value {value} for {value_name} is not an integer") + if not value.isnumeric(): + raise ValueError(f"Value {value} for {value_name} must be an integer") + value = int(value) # validate value if not min_range == -1: assert value >= min_range, f"Value {value} for {value_name} must be greater than or equal to {min_range}" @@ -46,7 +46,9 @@ def add_axis_options(): xyz_grid.AxisOption("[Hypertile] VAE Max Tile Size", int, int_applier("hypertile_max_tile_vae", 0, 512)), xyz_grid.AxisOption("[Hypertile] VAE Swap Size", int, int_applier("hypertile_swap_size_vae", 0, 64)), ] - # check if the axis options have already been added - if any(set(opt.label for opt in extra_axis_options).intersection(set(opt.label for opt in xyz_grid.axis_options))): + set_a = set([opt.label for opt in xyz_grid.axis_options]) + set_b = set([opt.label for opt in extra_axis_options]) + if set_a.intersection(set_b): return - xyz_grid.axis_options.extend(extra_axis_options) \ No newline at end of file + + xyz_grid.axis_options.extend(extra_axis_options) From f207eb7a0d8b4443dbe665df99c31f8ff91660fd Mon Sep 17 00:00:00 2001 From: aria1th <35677394+aria1th@users.noreply.github.com> Date: Mon, 27 Nov 2023 22:11:28 +0900 Subject: [PATCH 142/342] fix ruff in hypertile_xyz.py --- extensions-builtin/hypertile/scripts/hypertile_xyz.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/extensions-builtin/hypertile/scripts/hypertile_xyz.py b/extensions-builtin/hypertile/scripts/hypertile_xyz.py index 3007a0832..4055a9ead 100644 --- a/extensions-builtin/hypertile/scripts/hypertile_xyz.py +++ b/extensions-builtin/hypertile/scripts/hypertile_xyz.py @@ -46,8 +46,8 @@ def add_axis_options(): xyz_grid.AxisOption("[Hypertile] VAE Max Tile Size", int, int_applier("hypertile_max_tile_vae", 0, 512)), xyz_grid.AxisOption("[Hypertile] VAE Swap Size", int, int_applier("hypertile_swap_size_vae", 0, 64)), ] - set_a = set([opt.label for opt in xyz_grid.axis_options]) - set_b = set([opt.label for opt in extra_axis_options]) + set_a = set(opt.label for opt in xyz_grid.axis_options) + set_b = set(opt.label for opt in extra_axis_options) if set_a.intersection(set_b): return From 524d6a4dbae68bf557d9c5fe686707d96841e0b5 Mon Sep 17 00:00:00 2001 From: aria1th <35677394+aria1th@users.noreply.github.com> Date: Mon, 27 Nov 2023 22:13:18 +0900 Subject: [PATCH 143/342] fix ruff - set comprehension --- extensions-builtin/hypertile/scripts/hypertile_xyz.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/extensions-builtin/hypertile/scripts/hypertile_xyz.py b/extensions-builtin/hypertile/scripts/hypertile_xyz.py index 4055a9ead..928e99652 100644 --- a/extensions-builtin/hypertile/scripts/hypertile_xyz.py +++ b/extensions-builtin/hypertile/scripts/hypertile_xyz.py @@ -46,8 +46,8 @@ def add_axis_options(): xyz_grid.AxisOption("[Hypertile] VAE Max Tile Size", int, int_applier("hypertile_max_tile_vae", 0, 512)), xyz_grid.AxisOption("[Hypertile] VAE Swap Size", int, int_applier("hypertile_swap_size_vae", 0, 64)), ] - set_a = set(opt.label for opt in xyz_grid.axis_options) - set_b = set(opt.label for opt in extra_axis_options) + set_a = {opt.label for opt in xyz_grid.axis_options} + set_b = {opt.label for opt in extra_axis_options} if set_a.intersection(set_b): return From ec78354efa179b64e92d6b98d781f6572b4eb084 Mon Sep 17 00:00:00 2001 From: aria1th <35677394+aria1th@users.noreply.github.com> Date: Mon, 27 Nov 2023 22:25:28 +0900 Subject: [PATCH 144/342] hypertile_xyz: we don't need isnumeric check for AxisOption --- extensions-builtin/hypertile/scripts/hypertile_xyz.py | 3 --- 1 file changed, 3 deletions(-) diff --git a/extensions-builtin/hypertile/scripts/hypertile_xyz.py b/extensions-builtin/hypertile/scripts/hypertile_xyz.py index 928e99652..9e96ae3c5 100644 --- a/extensions-builtin/hypertile/scripts/hypertile_xyz.py +++ b/extensions-builtin/hypertile/scripts/hypertile_xyz.py @@ -7,10 +7,7 @@ def int_applier(value_name:str, min_range:int = -1, max_range:int = -1): """ Returns a function that applies the given value to the given value_name in opts.data. """ - # convert to int def validate(value_name:str, value:str): - if not value.isnumeric(): - raise ValueError(f"Value {value} for {value_name} must be an integer") value = int(value) # validate value if not min_range == -1: From 3cd6e1d0a0877e6f1ac931c8253e6eee09da3805 Mon Sep 17 00:00:00 2001 From: obsol <33932119+read-0nly@users.noreply.github.com> Date: Mon, 27 Nov 2023 19:21:43 -0500 Subject: [PATCH 145/342] Update devices.py fixes issue where "--use-cpu" all properly makes SD run on CPU but leaves ControlNet (and other extensions, I presume) pointed at GPU, causing a crash in ControlNet caused by a mismatch between devices between SD and CN https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/14097 --- modules/devices.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/devices.py b/modules/devices.py index c01f06024..65efcf1eb 100644 --- a/modules/devices.py +++ b/modules/devices.py @@ -38,7 +38,7 @@ def get_optimal_device(): def get_device_for(task): - if task in shared.cmd_opts.use_cpu: + if task in shared.cmd_opts.use_cpu or "all" in shared.cmd_opts.use_cpu: return cpu return get_optimal_device() From 03ee297aa22296ea12b965fc1cb11aa46375d372 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Mon, 27 Nov 2023 17:26:16 +0900 Subject: [PATCH 146/342] fix Auto focal point crop for opencv >= 4.8.x autocrop.download_and_cache_models in opencv >= 4.8 the face detection model was updated download the base on opencv version returns the model path or raise exception --- modules/textual_inversion/autocrop.py | 29 ++++++++++++++----------- modules/textual_inversion/preprocess.py | 4 ++-- 2 files changed, 18 insertions(+), 15 deletions(-) diff --git a/modules/textual_inversion/autocrop.py b/modules/textual_inversion/autocrop.py index 1675e39a5..051be1188 100644 --- a/modules/textual_inversion/autocrop.py +++ b/modules/textual_inversion/autocrop.py @@ -3,6 +3,8 @@ import requests import os import numpy as np from PIL import ImageDraw +from modules import paths_internal +from pkg_resources import parse_version GREEN = "#0F0" BLUE = "#00F" @@ -294,22 +296,23 @@ def is_square(w, h): return w == h -def download_and_cache_models(dirname): - download_url = 'https://github.com/opencv/opencv_zoo/blob/91fb0290f50896f38a0ab1e558b74b16bc009428/models/face_detection_yunet/face_detection_yunet_2022mar.onnx?raw=true' - model_file_name = 'face_detection_yunet.onnx' +model_dir_opencv = os.path.join(paths_internal.models_path, 'opencv') +if parse_version(cv2.__version__) >= parse_version('4.8'): + model_file_path = os.path.join(model_dir_opencv, 'face_detection_yunet_2023mar.onnx') + model_url = 'https://github.com/opencv/opencv_zoo/blob/b6e370b10f641879a87890d44e42173077154a05/models/face_detection_yunet/face_detection_yunet_2023mar.onnx?raw=true' +else: + model_file_path = os.path.join(model_dir_opencv, 'face_detection_yunet.onnx') + model_url = 'https://github.com/opencv/opencv_zoo/blob/91fb0290f50896f38a0ab1e558b74b16bc009428/models/face_detection_yunet/face_detection_yunet_2022mar.onnx?raw=true' - os.makedirs(dirname, exist_ok=True) - cache_file = os.path.join(dirname, model_file_name) - if not os.path.exists(cache_file): - print(f"downloading face detection model from '{download_url}' to '{cache_file}'") - response = requests.get(download_url) - with open(cache_file, "wb") as f: +def download_and_cache_models(): + if not os.path.exists(model_file_path): + os.makedirs(model_dir_opencv, exist_ok=True) + print(f"downloading face detection model from '{model_url}' to '{model_file_path}'") + response = requests.get(model_url) + with open(model_file_path, "wb") as f: f.write(response.content) - - if os.path.exists(cache_file): - return cache_file - return None + return model_file_path class PointOfInterest: diff --git a/modules/textual_inversion/preprocess.py b/modules/textual_inversion/preprocess.py index dbd856bd8..789fa0838 100644 --- a/modules/textual_inversion/preprocess.py +++ b/modules/textual_inversion/preprocess.py @@ -3,7 +3,7 @@ from PIL import Image, ImageOps import math import tqdm -from modules import paths, shared, images, deepbooru +from modules import shared, images, deepbooru from modules.textual_inversion import autocrop @@ -196,7 +196,7 @@ def preprocess_work(process_src, process_dst, process_width, process_height, pre dnn_model_path = None try: - dnn_model_path = autocrop.download_and_cache_models(os.path.join(paths.models_path, "opencv")) + dnn_model_path = autocrop.download_and_cache_models() except Exception as e: print("Unable to load face detection model for auto crop selection. Falling back to lower quality haar method.", e) From d608926f817b279d16b39a7875beec80d010a988 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Tue, 28 Nov 2023 12:12:27 +0900 Subject: [PATCH 147/342] reformat file with uniform indentation --- modules/textual_inversion/autocrop.py | 204 +++++++++++++------------- 1 file changed, 103 insertions(+), 101 deletions(-) diff --git a/modules/textual_inversion/autocrop.py b/modules/textual_inversion/autocrop.py index 051be1188..e223a2e0c 100644 --- a/modules/textual_inversion/autocrop.py +++ b/modules/textual_inversion/autocrop.py @@ -27,7 +27,6 @@ def crop_image(im, settings): elif is_portrait(settings.crop_width, settings.crop_height): scale_by = settings.crop_height / im.height - im = im.resize((int(im.width * scale_by), int(im.height * scale_by))) im_debug = im.copy() @@ -71,6 +70,7 @@ def crop_image(im, settings): return results + def focal_point(im, settings): corner_points = image_corner_points(im, settings) if settings.corner_points_weight > 0 else [] entropy_points = image_entropy_points(im, settings) if settings.entropy_points_weight > 0 else [] @@ -80,118 +80,120 @@ def focal_point(im, settings): weight_pref_total = 0 if corner_points: - weight_pref_total += settings.corner_points_weight + weight_pref_total += settings.corner_points_weight if entropy_points: - weight_pref_total += settings.entropy_points_weight + weight_pref_total += settings.entropy_points_weight if face_points: - weight_pref_total += settings.face_points_weight + weight_pref_total += settings.face_points_weight corner_centroid = None if corner_points: - corner_centroid = centroid(corner_points) - corner_centroid.weight = settings.corner_points_weight / weight_pref_total - pois.append(corner_centroid) + corner_centroid = centroid(corner_points) + corner_centroid.weight = settings.corner_points_weight / weight_pref_total + pois.append(corner_centroid) entropy_centroid = None if entropy_points: - entropy_centroid = centroid(entropy_points) - entropy_centroid.weight = settings.entropy_points_weight / weight_pref_total - pois.append(entropy_centroid) + entropy_centroid = centroid(entropy_points) + entropy_centroid.weight = settings.entropy_points_weight / weight_pref_total + pois.append(entropy_centroid) face_centroid = None if face_points: - face_centroid = centroid(face_points) - face_centroid.weight = settings.face_points_weight / weight_pref_total - pois.append(face_centroid) + face_centroid = centroid(face_points) + face_centroid.weight = settings.face_points_weight / weight_pref_total + pois.append(face_centroid) average_point = poi_average(pois, settings) if settings.annotate_image: - d = ImageDraw.Draw(im) - max_size = min(im.width, im.height) * 0.07 - if corner_centroid is not None: - color = BLUE - box = corner_centroid.bounding(max_size * corner_centroid.weight) - d.text((box[0], box[1]-15), f"Edge: {corner_centroid.weight:.02f}", fill=color) - d.ellipse(box, outline=color) - if len(corner_points) > 1: - for f in corner_points: - d.rectangle(f.bounding(4), outline=color) - if entropy_centroid is not None: - color = "#ff0" - box = entropy_centroid.bounding(max_size * entropy_centroid.weight) - d.text((box[0], box[1]-15), f"Entropy: {entropy_centroid.weight:.02f}", fill=color) - d.ellipse(box, outline=color) - if len(entropy_points) > 1: - for f in entropy_points: - d.rectangle(f.bounding(4), outline=color) - if face_centroid is not None: - color = RED - box = face_centroid.bounding(max_size * face_centroid.weight) - d.text((box[0], box[1]-15), f"Face: {face_centroid.weight:.02f}", fill=color) - d.ellipse(box, outline=color) - if len(face_points) > 1: - for f in face_points: - d.rectangle(f.bounding(4), outline=color) + d = ImageDraw.Draw(im) + max_size = min(im.width, im.height) * 0.07 + if corner_centroid is not None: + color = BLUE + box = corner_centroid.bounding(max_size * corner_centroid.weight) + d.text((box[0], box[1] - 15), f"Edge: {corner_centroid.weight:.02f}", fill=color) + d.ellipse(box, outline=color) + if len(corner_points) > 1: + for f in corner_points: + d.rectangle(f.bounding(4), outline=color) + if entropy_centroid is not None: + color = "#ff0" + box = entropy_centroid.bounding(max_size * entropy_centroid.weight) + d.text((box[0], box[1] - 15), f"Entropy: {entropy_centroid.weight:.02f}", fill=color) + d.ellipse(box, outline=color) + if len(entropy_points) > 1: + for f in entropy_points: + d.rectangle(f.bounding(4), outline=color) + if face_centroid is not None: + color = RED + box = face_centroid.bounding(max_size * face_centroid.weight) + d.text((box[0], box[1] - 15), f"Face: {face_centroid.weight:.02f}", fill=color) + d.ellipse(box, outline=color) + if len(face_points) > 1: + for f in face_points: + d.rectangle(f.bounding(4), outline=color) - d.ellipse(average_point.bounding(max_size), outline=GREEN) + d.ellipse(average_point.bounding(max_size), outline=GREEN) return average_point def image_face_points(im, settings): if settings.dnn_model_path is not None: - detector = cv2.FaceDetectorYN.create( - settings.dnn_model_path, - "", - (im.width, im.height), - 0.9, # score threshold - 0.3, # nms threshold - 5000 # keep top k before nms - ) - faces = detector.detect(np.array(im)) - results = [] - if faces[1] is not None: - for face in faces[1]: - x = face[0] - y = face[1] - w = face[2] - h = face[3] - results.append( - PointOfInterest( - int(x + (w * 0.5)), # face focus left/right is center - int(y + (h * 0.33)), # face focus up/down is close to the top of the head - size = w, - weight = 1/len(faces[1]) - ) - ) - return results + detector = cv2.FaceDetectorYN.create( + settings.dnn_model_path, + "", + (im.width, im.height), + 0.9, # score threshold + 0.3, # nms threshold + 5000 # keep top k before nms + ) + faces = detector.detect(np.array(im)) + results = [] + if faces[1] is not None: + for face in faces[1]: + x = face[0] + y = face[1] + w = face[2] + h = face[3] + results.append( + PointOfInterest( + int(x + (w * 0.5)), # face focus left/right is center + int(y + (h * 0.33)), # face focus up/down is close to the top of the head + size=w, + weight=1 / len(faces[1]) + ) + ) + return results else: - np_im = np.array(im) - gray = cv2.cvtColor(np_im, cv2.COLOR_BGR2GRAY) + np_im = np.array(im) + gray = cv2.cvtColor(np_im, cv2.COLOR_BGR2GRAY) - tries = [ - [ f'{cv2.data.haarcascades}haarcascade_eye.xml', 0.01 ], - [ f'{cv2.data.haarcascades}haarcascade_frontalface_default.xml', 0.05 ], - [ f'{cv2.data.haarcascades}haarcascade_profileface.xml', 0.05 ], - [ f'{cv2.data.haarcascades}haarcascade_frontalface_alt.xml', 0.05 ], - [ f'{cv2.data.haarcascades}haarcascade_frontalface_alt2.xml', 0.05 ], - [ f'{cv2.data.haarcascades}haarcascade_frontalface_alt_tree.xml', 0.05 ], - [ f'{cv2.data.haarcascades}haarcascade_eye_tree_eyeglasses.xml', 0.05 ], - [ f'{cv2.data.haarcascades}haarcascade_upperbody.xml', 0.05 ] - ] - for t in tries: - classifier = cv2.CascadeClassifier(t[0]) - minsize = int(min(im.width, im.height) * t[1]) # at least N percent of the smallest side - try: - faces = classifier.detectMultiScale(gray, scaleFactor=1.1, - minNeighbors=7, minSize=(minsize, minsize), flags=cv2.CASCADE_SCALE_IMAGE) - except Exception: - continue + tries = [ + [f'{cv2.data.haarcascades}haarcascade_eye.xml', 0.01], + [f'{cv2.data.haarcascades}haarcascade_frontalface_default.xml', 0.05], + [f'{cv2.data.haarcascades}haarcascade_profileface.xml', 0.05], + [f'{cv2.data.haarcascades}haarcascade_frontalface_alt.xml', 0.05], + [f'{cv2.data.haarcascades}haarcascade_frontalface_alt2.xml', 0.05], + [f'{cv2.data.haarcascades}haarcascade_frontalface_alt_tree.xml', 0.05], + [f'{cv2.data.haarcascades}haarcascade_eye_tree_eyeglasses.xml', 0.05], + [f'{cv2.data.haarcascades}haarcascade_upperbody.xml', 0.05] + ] + for t in tries: + classifier = cv2.CascadeClassifier(t[0]) + minsize = int(min(im.width, im.height) * t[1]) # at least N percent of the smallest side + try: + faces = classifier.detectMultiScale(gray, scaleFactor=1.1, + minNeighbors=7, minSize=(minsize, minsize), + flags=cv2.CASCADE_SCALE_IMAGE) + except Exception: + continue - if faces: - rects = [[f[0], f[1], f[0] + f[2], f[1] + f[3]] for f in faces] - return [PointOfInterest((r[0] +r[2]) // 2, (r[1] + r[3]) // 2, size=abs(r[0]-r[2]), weight=1/len(rects)) for r in rects] + if faces: + rects = [[f[0], f[1], f[0] + f[2], f[1] + f[3]] for f in faces] + return [PointOfInterest((r[0] + r[2]) // 2, (r[1] + r[3]) // 2, size=abs(r[0] - r[2]), + weight=1 / len(rects)) for r in rects] return [] @@ -200,7 +202,7 @@ def image_corner_points(im, settings): # naive attempt at preventing focal points from collecting at watermarks near the bottom gd = ImageDraw.Draw(grayscale) - gd.rectangle([0, im.height*.9, im.width, im.height], fill="#999") + gd.rectangle([0, im.height * .9, im.width, im.height], fill="#999") np_im = np.array(grayscale) @@ -208,7 +210,7 @@ def image_corner_points(im, settings): np_im, maxCorners=100, qualityLevel=0.04, - minDistance=min(grayscale.width, grayscale.height)*0.06, + minDistance=min(grayscale.width, grayscale.height) * 0.06, useHarrisDetector=False, ) @@ -217,8 +219,8 @@ def image_corner_points(im, settings): focal_points = [] for point in points: - x, y = point.ravel() - focal_points.append(PointOfInterest(x, y, size=4, weight=1/len(points))) + x, y = point.ravel() + focal_points.append(PointOfInterest(x, y, size=4, weight=1 / len(points))) return focal_points @@ -227,13 +229,13 @@ def image_entropy_points(im, settings): landscape = im.height < im.width portrait = im.height > im.width if landscape: - move_idx = [0, 2] - move_max = im.size[0] + move_idx = [0, 2] + move_max = im.size[0] elif portrait: - move_idx = [1, 3] - move_max = im.size[1] + move_idx = [1, 3] + move_max = im.size[1] else: - return [] + return [] e_max = 0 crop_current = [0, 0, settings.crop_width, settings.crop_height] @@ -243,14 +245,14 @@ def image_entropy_points(im, settings): e = image_entropy(crop) if (e > e_max): - e_max = e - crop_best = list(crop_current) + e_max = e + crop_best = list(crop_current) crop_current[move_idx[0]] += 4 crop_current[move_idx[1]] += 4 - x_mid = int(crop_best[0] + settings.crop_width/2) - y_mid = int(crop_best[1] + settings.crop_height/2) + x_mid = int(crop_best[0] + settings.crop_width / 2) + y_mid = int(crop_best[1] + settings.crop_height / 2) return [PointOfInterest(x_mid, y_mid, size=25, weight=1.0)] From 39eae9f009c8302eed77b0942e1e634f6125d53e Mon Sep 17 00:00:00 2001 From: hidenorly Date: Wed, 29 Nov 2023 04:07:48 +0900 Subject: [PATCH 148/342] Revert "Add FP32 fallback support on sd_vae_approx" This reverts commit 58c19545c83fa6925c9ce2216ee64964eb5129ce. Since the modification is expected to move to mac_specific.py (https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14046#issuecomment-1826731532) --- modules/sd_vae_approx.py | 8 +------- 1 file changed, 1 insertion(+), 7 deletions(-) diff --git a/modules/sd_vae_approx.py b/modules/sd_vae_approx.py index 8370493f9..3965e223e 100644 --- a/modules/sd_vae_approx.py +++ b/modules/sd_vae_approx.py @@ -21,13 +21,7 @@ class VAEApprox(nn.Module): def forward(self, x): extra = 11 - try: - x = nn.functional.interpolate(x, (x.shape[2] * 2, x.shape[3] * 2)) - except RuntimeError as e: - if "not implemented for" in str(e) and "Half" in str(e): - x = nn.functional.interpolate(x.to(torch.float32), (x.shape[2] * 2, x.shape[3] * 2)).to(x.dtype) - else: - print(f"An unexpected RuntimeError occurred: {str(e)}") + x = nn.functional.interpolate(x, (x.shape[2] * 2, x.shape[3] * 2)) x = nn.functional.pad(x, (extra, extra, extra, extra)) for layer in [self.conv1, self.conv2, self.conv3, self.conv4, self.conv5, self.conv6, self.conv7, self.conv8, ]: From a0096c58977c01ddc6a2b83a8a7b64da6fd4a51e Mon Sep 17 00:00:00 2001 From: hidenorly Date: Wed, 29 Nov 2023 04:45:04 +0900 Subject: [PATCH 149/342] Add FP32 fallback support on torch.nn.functional.interpolate This tries to execute interpolate with FP32 if it failed. Background is that on some environment such as Mx chip MacOS devices, we get error as follows: ``` "torch/nn/functional.py", line 3931, in interpolate return torch._C._nn.upsample_nearest2d(input, output_size, scale_factors) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ RuntimeError: "upsample_nearest2d_channels_last" not implemented for 'Half' ``` In this case, ```--no-half``` doesn't help to solve. Therefore this commits add the FP32 fallback execution to solve it. Note that the ```upsample_nearest2d``` is called from ```torch.nn.functional.interpolate```. And the fallback for torch.nn.functional.interpolate is necessary at ```modules/sd_vae_approx.py``` 's ```VAEApprox.forward``` ```repositories/stable-diffusion-stability-ai/ldm/modules/diffusionmodules/openaimodel.py``` 's ```Upsample.forward``` --- modules/mac_specific.py | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) diff --git a/modules/mac_specific.py b/modules/mac_specific.py index 89256c5b0..3538e659d 100644 --- a/modules/mac_specific.py +++ b/modules/mac_specific.py @@ -1,6 +1,8 @@ import logging import torch +from typing import Optional, List +from torch import Tensor import platform from modules.sd_hijack_utils import CondFunc from packaging import version @@ -51,6 +53,17 @@ def cumsum_fix(input, cumsum_func, *args, **kwargs): return cumsum_func(input, *args, **kwargs) +# MPS workaround for https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14046 +def interpolate_with_fp32_fallback(orig_func, *args, **kwargs) -> Tensor: + try: + return orig_func(*args, **kwargs) + except RuntimeError as e: + if "not implemented for" in str(e) and "Half" in str(e): + input_tensor = args[0] + return orig_func(input_tensor.to(torch.float32), *args[1:], **kwargs).to(input_tensor.dtype) + else: + print(f"An unexpected RuntimeError occurred: {str(e)}") + if has_mps: if platform.mac_ver()[0].startswith("13.2."): # MPS workaround for https://github.com/pytorch/pytorch/issues/95188, thanks to danieldk (https://github.com/explosion/curated-transformers/pull/124) @@ -77,6 +90,9 @@ if has_mps: # MPS workaround for https://github.com/pytorch/pytorch/issues/96113 CondFunc('torch.nn.functional.layer_norm', lambda orig_func, x, normalized_shape, weight, bias, eps, **kwargs: orig_func(x.float(), normalized_shape, weight.float() if weight is not None else None, bias.float() if bias is not None else bias, eps).to(x.dtype), lambda _, input, *args, **kwargs: len(args) == 4 and input.device.type == 'mps') + # MPS workaround for https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14046 + CondFunc('torch.nn.functional.interpolate', interpolate_with_fp32_fallback, None) + # MPS workaround for https://github.com/pytorch/pytorch/issues/92311 if platform.processor() == 'i386': for funcName in ['torch.argmax', 'torch.Tensor.argmax']: From 81c00728b8ec0b6c0e70ea10c7687aad065a95cb Mon Sep 17 00:00:00 2001 From: hidenorly Date: Wed, 29 Nov 2023 04:59:35 +0900 Subject: [PATCH 150/342] Fix the Ruff error about unused import --- modules/mac_specific.py | 1 - 1 file changed, 1 deletion(-) diff --git a/modules/mac_specific.py b/modules/mac_specific.py index 3538e659d..d96d86d79 100644 --- a/modules/mac_specific.py +++ b/modules/mac_specific.py @@ -1,7 +1,6 @@ import logging import torch -from typing import Optional, List from torch import Tensor import platform from modules.sd_hijack_utils import CondFunc From dec791d35ddcd02ca33563d3d0355e05e45de8ad Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Tue, 28 Nov 2023 15:05:01 -0700 Subject: [PATCH 151/342] Removed code which forces the inpainting mask to be 0 or 1. Now fractional values (e.g. 0.5) are accepted. --- modules/processing.py | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index e124e7f0d..317458f58 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -83,7 +83,7 @@ def apply_overlay(image, paste_loc, index, overlays): def create_binary_mask(image): if image.mode == 'RGBA' and image.getextrema()[-1] != (255, 255): - image = image.split()[-1].convert("L").point(lambda x: 255 if x > 128 else 0) + image = image.split()[-1].convert("L") else: image = image.convert('L') return image @@ -319,9 +319,6 @@ class StableDiffusionProcessing: conditioning_mask = np.array(image_mask.convert("L")) conditioning_mask = conditioning_mask.astype(np.float32) / 255.0 conditioning_mask = torch.from_numpy(conditioning_mask[None, None]) - - # Inpainting model uses a discretized mask as input, so we round to either 1.0 or 0.0 - conditioning_mask = torch.round(conditioning_mask) else: conditioning_mask = source_image.new_ones(1, 1, *source_image.shape[-2:]) @@ -1504,7 +1501,6 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): latmask = init_mask.convert('RGB').resize((self.init_latent.shape[3], self.init_latent.shape[2])) latmask = np.moveaxis(np.array(latmask, dtype=np.float32), 2, 0) / 255 latmask = latmask[0] - latmask = np.around(latmask) latmask = np.tile(latmask[None], (4, 1, 1)) self.mask = torch.asarray(1.0 - latmask).to(shared.device).type(self.sd_model.dtype) From bbba133f054706c3668b7d03b0e6d0afc15705db Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Tue, 28 Nov 2023 15:09:43 -0700 Subject: [PATCH 152/342] Removed conflicting step that replaces the softly inpainted latents with a naive blend with the original latents. --- modules/processing.py | 3 --- 1 file changed, 3 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index 317458f58..ae894f1a7 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -1523,9 +1523,6 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): samples = self.sampler.sample_img2img(self, self.init_latent, x, conditioning, unconditional_conditioning, image_conditioning=self.image_conditioning) - if self.mask is not None: - samples = samples * self.nmask + self.init_latent * self.mask - del x devices.torch_gc() From e715e46b6aa7f2e5e147cfa1fa2f49b1d926a074 Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Tue, 28 Nov 2023 16:10:22 -0700 Subject: [PATCH 153/342] Implements "scheduling" for blending of the original latents and a latent blending formula that preserves details in blend transition areas. --- modules/sd_samplers_cfg_denoiser.py | 61 ++++++++++++++++++++++++++++- 1 file changed, 59 insertions(+), 2 deletions(-) diff --git a/modules/sd_samplers_cfg_denoiser.py b/modules/sd_samplers_cfg_denoiser.py index b8101d38d..c4d6fda65 100644 --- a/modules/sd_samplers_cfg_denoiser.py +++ b/modules/sd_samplers_cfg_denoiser.py @@ -43,6 +43,9 @@ class CFGDenoiser(torch.nn.Module): self.model_wrap = None self.mask = None self.nmask = None + self.mask_blend_power = 1 + self.mask_blend_scale = 1 + self.mask_blend_offset = 0 self.init_latent = None self.steps = None """number of steps as specified by user in UI""" @@ -56,6 +59,9 @@ class CFGDenoiser(torch.nn.Module): self.sampler = sampler self.model_wrap = None self.p = None + + # NOTE: masking before denoising can cause the original latents to be oversmoothed + # as the original latents do not have noise self.mask_before_denoising = False @property @@ -89,6 +95,55 @@ class CFGDenoiser(torch.nn.Module): self.sampler.sampler_extra_args['uncond'] = uc def forward(self, x, sigma, uncond, cond, cond_scale, s_min_uncond, image_cond): + def latent_blend(a, b, t): + """ + Interpolates two latent image representations according to the parameter t, + where the interpolated vectors' magnitudes are also interpolated separately. + The "detail_preservation" factor biases the magnitude interpolation towards + the larger of the two magnitudes. + """ + # Record the original latent vector magnitudes. + # We bring them to a power so that larger magnitudes are favored over smaller ones. + # 64-bit operations are used here to allow large exponents. + detail_preservation = 32 + a_magnitude = torch.norm(a, p=2, dim=1).to(torch.float64) ** detail_preservation + b_magnitude = torch.norm(b, p=2, dim=1).to(torch.float64) ** detail_preservation + + one_minus_t = 1 - t + + # Interpolate the powered magnitudes, then un-power them (bring them back to a power of 1). + interp_magnitude = (a_magnitude * one_minus_t + b_magnitude * t) ** (1 / detail_preservation) + + # Linearly interpolate the image vectors. + image_interp = a * one_minus_t + b * t + + # Calculate the magnitude of the interpolated vectors. (We will remove this magnitude.) + # 64-bit operations are used here to allow large exponents. + image_interp_magnitude = torch.norm(image_interp, p=2, dim=1).to(torch.float64) + 0.0001 + + # Change the linearly interpolated image vectors' magnitudes to the value we want. + # This is the last 64-bit operation. + image_interp *= (interp_magnitude / image_interp_magnitude).to(image_interp.dtype) + + return image_interp + + def get_modified_nmask(nmask, _sigma): + """ + Converts a negative mask representing the transparency of the original latent vectors being overlayed + to a mask that is scaled according to the denoising strength for this step. + + Where: + 0 = fully opaque, infinite density, fully masked + 1 = fully transparent, zero density, fully unmasked + + We bring this transparency to a power, as this allows one to simulate N number of blending operations + where N can be any positive real value. Using this one can control the balance of influence between + the denoiser and the original latents according to the sigma value. + + NOTE: "mask" is not used + """ + return torch.pow(nmask, (_sigma ** self.mask_blend_power) * self.mask_blend_scale + self.mask_blend_offset) + if state.interrupted or state.skipped: raise sd_samplers_common.InterruptedException @@ -105,8 +160,9 @@ class CFGDenoiser(torch.nn.Module): assert not is_edit_model or all(len(conds) == 1 for conds in conds_list), "AND is not supported for InstructPix2Pix checkpoint (unless using Image CFG scale = 1.0)" + # Blend in the original latents (before) if self.mask_before_denoising and self.mask is not None: - x = self.init_latent * self.mask + self.nmask * x + x = latent_blend(self.init_latent, x, get_modified_nmask(self.nmask, sigma)) batch_size = len(conds_list) repeats = [len(conds_list[i]) for i in range(batch_size)] @@ -207,8 +263,9 @@ class CFGDenoiser(torch.nn.Module): else: denoised = self.combine_denoised(x_out, conds_list, uncond, cond_scale) + # Blend in the original latents (after) if not self.mask_before_denoising and self.mask is not None: - denoised = self.init_latent * self.mask + self.nmask * denoised + denoised = latent_blend(self.init_latent, denoised, get_modified_nmask(self.nmask, sigma)) self.sampler.last_latent = self.get_pred_x0(torch.cat([x_in[i:i + 1] for i in denoised_image_indexes]), torch.cat([x_out[i:i + 1] for i in denoised_image_indexes]), sigma) From a6e584645305c0a91a3d46f73546e191b249210f Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Tue, 28 Nov 2023 16:13:42 -0700 Subject: [PATCH 154/342] Nerfs the aggressive post-processing step of overlaying the original image. --- modules/processing.py | 14 +++++++++++--- 1 file changed, 11 insertions(+), 3 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index ae894f1a7..12e08e876 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -1412,7 +1412,12 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): image_mask = Image.fromarray(np_mask) if self.inpaint_full_res: - self.mask_for_overlay = image_mask + np_mask = np.array(image_mask).astype(np.float32) + np_mask /= 255 + np_mask = 1-pow(1-np_mask, 100) + np_mask *= 255 + np_mask = np.clip(np_mask, 0, 255).astype(np.uint8) + self.mask_for_overlay = Image.fromarray(np_mask) mask = image_mask.convert('L') crop_region = masking.get_crop_region(np.array(mask), self.inpaint_full_res_padding) crop_region = masking.expand_crop_region(crop_region, self.width, self.height, mask.width, mask.height) @@ -1423,8 +1428,11 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): self.paste_to = (x1, y1, x2-x1, y2-y1) else: image_mask = images.resize_image(self.resize_mode, image_mask, self.width, self.height) - np_mask = np.array(image_mask) - np_mask = np.clip((np_mask.astype(np.float32)) * 2, 0, 255).astype(np.uint8) + np_mask = np.array(image_mask).astype(np.float32) + np_mask /= 255 + np_mask = 1-pow(1-np_mask, 100) + np_mask *= 255 + np_mask = np.clip(np_mask, 0, 255).astype(np.uint8) self.mask_for_overlay = Image.fromarray(np_mask) self.overlay_images = [] From debf836fcc8d9becc3da8b1a29e33f40b0d9ef3e Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Tue, 28 Nov 2023 16:15:36 -0700 Subject: [PATCH 155/342] Added UI elements to control blending parameters. --- modules/img2img.py | 48 +++++++++++++++++++++++++++++++- modules/processing.py | 3 ++ modules/sd_samplers_common.py | 3 ++ modules/ui.py | 9 ++++++ scripts/outpainting_mk_2.py | 10 +++++-- scripts/poor_mans_outpainting.py | 11 ++++++-- test/test_img2img.py | 3 ++ 7 files changed, 82 insertions(+), 5 deletions(-) diff --git a/modules/img2img.py b/modules/img2img.py index 1519e132b..240d05884 100644 --- a/modules/img2img.py +++ b/modules/img2img.py @@ -116,7 +116,47 @@ def process_batch(p, input_dir, output_dir, inpaint_mask_dir, args, to_scale=Fal process_images(p) -def img2img(id_task: str, mode: int, prompt: str, negative_prompt: str, prompt_styles, init_img, sketch, init_img_with_mask, inpaint_color_sketch, inpaint_color_sketch_orig, init_img_inpaint, init_mask_inpaint, steps: int, sampler_name: str, mask_blur: int, mask_alpha: float, inpainting_fill: int, n_iter: int, batch_size: int, cfg_scale: float, image_cfg_scale: float, denoising_strength: float, selected_scale_tab: int, height: int, width: int, scale_by: float, resize_mode: int, inpaint_full_res: bool, inpaint_full_res_padding: int, inpainting_mask_invert: int, img2img_batch_input_dir: str, img2img_batch_output_dir: str, img2img_batch_inpaint_mask_dir: str, override_settings_texts, img2img_batch_use_png_info: bool, img2img_batch_png_info_props: list, img2img_batch_png_info_dir: str, request: gr.Request, *args): +def img2img(id_task: str, + mode: int, + prompt: str, + negative_prompt: str, + prompt_styles, + init_img, + sketch, + init_img_with_mask, + inpaint_color_sketch, + inpaint_color_sketch_orig, + init_img_inpaint, + init_mask_inpaint, + steps: int, + sampler_name: str, + mask_blur: int, + mask_alpha: float, + mask_blend_power: float, + mask_blend_scale: float, + mask_blend_offset: float, + inpainting_fill: int, + n_iter: int, + batch_size: int, + cfg_scale: float, + image_cfg_scale: float, + denoising_strength: float, + selected_scale_tab: int, + height: int, + width: int, + scale_by: float, + resize_mode: int, + inpaint_full_res: bool, + inpaint_full_res_padding: int, + inpainting_mask_invert: int, + img2img_batch_input_dir: str, + img2img_batch_output_dir: str, + img2img_batch_inpaint_mask_dir: str, + override_settings_texts, + img2img_batch_use_png_info: bool, + img2img_batch_png_info_props: list, + img2img_batch_png_info_dir: str, + request: gr.Request, *args): override_settings = create_override_settings_dict(override_settings_texts) is_batch = mode == 5 @@ -174,6 +214,9 @@ def img2img(id_task: str, mode: int, prompt: str, negative_prompt: str, prompt_s init_images=[image], mask=mask, mask_blur=mask_blur, + mask_blend_power=mask_blend_power, + mask_blend_scale=mask_blend_scale, + mask_blend_offset=mask_blend_offset, inpainting_fill=inpainting_fill, resize_mode=resize_mode, denoising_strength=denoising_strength, @@ -194,6 +237,9 @@ def img2img(id_task: str, mode: int, prompt: str, negative_prompt: str, prompt_s if mask: p.extra_generation_params["Mask blur"] = mask_blur + p.extra_generation_params["Mask blend power"] = mask_blend_power + p.extra_generation_params["Mask blend scale"] = mask_blend_scale + p.extra_generation_params["Mask blend offset"] = mask_blend_offset with closing(p): if is_batch: diff --git a/modules/processing.py b/modules/processing.py index 12e08e876..da4d6fda9 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -1349,6 +1349,9 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): mask_blur_x: int = 4 mask_blur_y: int = 4 mask_blur: int = None + mask_blend_power: float = 1 + mask_blend_scale: float = 1 + mask_blend_offset: float = 0 inpainting_fill: int = 0 inpaint_full_res: bool = True inpaint_full_res_padding: int = 0 diff --git a/modules/sd_samplers_common.py b/modules/sd_samplers_common.py index 58efcad23..8904da2fb 100644 --- a/modules/sd_samplers_common.py +++ b/modules/sd_samplers_common.py @@ -277,6 +277,9 @@ class Sampler: self.model_wrap_cfg.p = p self.model_wrap_cfg.mask = p.mask if hasattr(p, 'mask') else None self.model_wrap_cfg.nmask = p.nmask if hasattr(p, 'nmask') else None + self.model_wrap_cfg.mask_blend_power = p.mask_blend_power if hasattr(p, 'mask_blend_power') else None + self.model_wrap_cfg.mask_blend_scale = p.mask_blend_scale if hasattr(p, 'mask_blend_scale') else None + self.model_wrap_cfg.mask_blend_offset = p.mask_blend_offset if hasattr(p, 'mask_blend_offset') else None self.model_wrap_cfg.step = 0 self.model_wrap_cfg.image_cfg_scale = getattr(p, 'image_cfg_scale', None) self.eta = p.eta if p.eta is not None else getattr(opts, self.eta_option_field, 0.0) diff --git a/modules/ui.py b/modules/ui.py index 579bab980..86c130869 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -732,6 +732,9 @@ def create_ui(): with FormRow(): mask_blur = gr.Slider(label='Mask blur', minimum=0, maximum=64, step=1, value=4, elem_id="img2img_mask_blur") mask_alpha = gr.Slider(label="Mask transparency", visible=False, elem_id="img2img_mask_alpha") + mask_blend_power = gr.Slider(label='Mask blend power', minimum=0, maximum=8, step=0.1, value=1, elem_id="img2img_mask_blend_power") + mask_blend_scale = gr.Slider(label='Mask blend scale', minimum=0, maximum=8, step=0.1, value=1, elem_id="img2img_mask_blend_scale") + mask_blend_offset = gr.Slider(label='Mask blend offset', minimum=-4, maximum=4, step=0.1, value=0, elem_id="img2img_mask_blend_offset") with FormRow(): inpainting_mask_invert = gr.Radio(label='Mask mode', choices=['Inpaint masked', 'Inpaint not masked'], value='Inpaint masked', type="index", elem_id="img2img_mask_mode") @@ -781,6 +784,9 @@ def create_ui(): sampler_name, mask_blur, mask_alpha, + mask_blend_power, + mask_blend_scale, + mask_blend_offset, inpainting_fill, batch_count, batch_size, @@ -879,6 +885,9 @@ def create_ui(): (toprow.ui_styles.dropdown, lambda d: d["Styles array"] if isinstance(d.get("Styles array"), list) else gr.update()), (denoising_strength, "Denoising strength"), (mask_blur, "Mask blur"), + (mask_blend_power, "Mask blend power"), + (mask_blend_scale, "Mask blend scale"), + (mask_blend_offset, "Mask blend offset"), *scripts.scripts_img2img.infotext_fields ] parameters_copypaste.add_paste_fields("img2img", init_img, img2img_paste_fields, override_settings) diff --git a/scripts/outpainting_mk_2.py b/scripts/outpainting_mk_2.py index c98ab4809..6aa97edfa 100644 --- a/scripts/outpainting_mk_2.py +++ b/scripts/outpainting_mk_2.py @@ -133,13 +133,16 @@ class Script(scripts.Script): pixels = gr.Slider(label="Pixels to expand", minimum=8, maximum=256, step=8, value=128, elem_id=self.elem_id("pixels")) mask_blur = gr.Slider(label='Mask blur', minimum=0, maximum=64, step=1, value=8, elem_id=self.elem_id("mask_blur")) + mask_blend_power = gr.Slider(label='Mask blend power', minimum=0, maximum=8, step=0.1, value=1, elem_id=self.elem_id("mask_blend_power")) + mask_blend_scale = gr.Slider(label='Mask blend scale', minimum=0, maximum=8, step=0.1, value=1, elem_id=self.elem_id("mask_blend_scale")) + mask_blend_offset = gr.Slider(label='Mask blend scale', minimum=-4, maximum=4, step=0.1, value=1, elem_id=self.elem_id("mask_blend_offset")) direction = gr.CheckboxGroup(label="Outpainting direction", choices=['left', 'right', 'up', 'down'], value=['left', 'right', 'up', 'down'], elem_id=self.elem_id("direction")) noise_q = gr.Slider(label="Fall-off exponent (lower=higher detail)", minimum=0.0, maximum=4.0, step=0.01, value=1.0, elem_id=self.elem_id("noise_q")) color_variation = gr.Slider(label="Color variation", minimum=0.0, maximum=1.0, step=0.01, value=0.05, elem_id=self.elem_id("color_variation")) - return [info, pixels, mask_blur, direction, noise_q, color_variation] + return [info, pixels, mask_blur, mask_blend_power, mask_blend_scale, mask_blend_offset, direction, noise_q, color_variation] - def run(self, p, _, pixels, mask_blur, direction, noise_q, color_variation): + def run(self, p, _, pixels, mask_blur, mask_blend_power, mask_blend_scale, mask_blend_offset, direction, noise_q, color_variation): initial_seed_and_info = [None, None] process_width = p.width @@ -167,6 +170,9 @@ class Script(scripts.Script): p.mask_blur_x = mask_blur_x*4 p.mask_blur_y = mask_blur_y*4 + p.mask_blend_power = mask_blend_power + p.mask_blend_scale = mask_blend_scale + p.mask_blend_offset = mask_blend_offset init_img = p.init_images[0] target_w = math.ceil((init_img.width + left + right) / 64) * 64 diff --git a/scripts/poor_mans_outpainting.py b/scripts/poor_mans_outpainting.py index ea0632b68..b10140f14 100644 --- a/scripts/poor_mans_outpainting.py +++ b/scripts/poor_mans_outpainting.py @@ -22,16 +22,23 @@ class Script(scripts.Script): pixels = gr.Slider(label="Pixels to expand", minimum=8, maximum=256, step=8, value=128, elem_id=self.elem_id("pixels")) mask_blur = gr.Slider(label='Mask blur', minimum=0, maximum=64, step=1, value=4, elem_id=self.elem_id("mask_blur")) + mask_blend_power = gr.Slider(label='Mask blend power', minimum=0, maximum=8, step=0.1, value=1, elem_id=self.elem_id("mask_blend_power")) + mask_blend_scale = gr.Slider(label='Mask blend scale', minimum=0, maximum=8, step=0.1, value=1, elem_id=self.elem_id("mask_blend_scale")) + mask_blend_offset = gr.Slider(label='Mask blend offset', minimum=-4, maximum=4, step=0.1, value=0, elem_id=self.elem_id("mask_blend_offset")) inpainting_fill = gr.Radio(label='Masked content', choices=['fill', 'original', 'latent noise', 'latent nothing'], value='fill', type="index", elem_id=self.elem_id("inpainting_fill")) direction = gr.CheckboxGroup(label="Outpainting direction", choices=['left', 'right', 'up', 'down'], value=['left', 'right', 'up', 'down'], elem_id=self.elem_id("direction")) - return [pixels, mask_blur, inpainting_fill, direction] + return [pixels, mask_blur, mask_blend_power, mask_blend_scale, mask_blend_offset, inpainting_fill, direction] - def run(self, p, pixels, mask_blur, inpainting_fill, direction): + def run(self, p, pixels, mask_blur, mask_blend_power, mask_blend_scale, mask_blend_offset, inpainting_fill, direction): initial_seed = None initial_info = None p.mask_blur = mask_blur * 2 + p.mask_blend_power = mask_blend_power + p.mask_blend_scale = mask_blend_scale + p.mask_blend_offset = mask_blend_offset + p.inpainting_fill = inpainting_fill p.inpaint_full_res = False diff --git a/test/test_img2img.py b/test/test_img2img.py index 117d2d1eb..6289e59e1 100644 --- a/test/test_img2img.py +++ b/test/test_img2img.py @@ -24,6 +24,9 @@ def simple_img2img_request(img2img_basic_image_base64): "inpainting_mask_invert": False, "mask": None, "mask_blur": 4, + "mask_blend_power": 1, + "mask_blend_scale": 1, + "mask_blend_offset": 0, "n_iter": 1, "negative_prompt": "", "override_settings": {}, From c5c7fa06aae1ae9f8b6d29ae2da3874921d4729b Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Tue, 28 Nov 2023 22:35:07 -0700 Subject: [PATCH 156/342] Added slider for detail preservation strength, removed largely needless offset parameter, changed labels in UI and for saving to/pasting data from PNG files. --- modules/img2img.py | 10 +++++----- modules/processing.py | 2 +- modules/sd_samplers_cfg_denoiser.py | 11 +++++------ modules/sd_samplers_common.py | 2 +- modules/ui.py | 14 +++++++------- scripts/outpainting_mk_2.py | 12 ++++++------ scripts/poor_mans_outpainting.py | 12 ++++++------ test/test_img2img.py | 2 +- 8 files changed, 32 insertions(+), 33 deletions(-) diff --git a/modules/img2img.py b/modules/img2img.py index 240d05884..023808d6c 100644 --- a/modules/img2img.py +++ b/modules/img2img.py @@ -134,7 +134,7 @@ def img2img(id_task: str, mask_alpha: float, mask_blend_power: float, mask_blend_scale: float, - mask_blend_offset: float, + inpaint_detail_preservation: float, inpainting_fill: int, n_iter: int, batch_size: int, @@ -216,7 +216,7 @@ def img2img(id_task: str, mask_blur=mask_blur, mask_blend_power=mask_blend_power, mask_blend_scale=mask_blend_scale, - mask_blend_offset=mask_blend_offset, + inpaint_detail_preservation=inpaint_detail_preservation, inpainting_fill=inpainting_fill, resize_mode=resize_mode, denoising_strength=denoising_strength, @@ -237,9 +237,9 @@ def img2img(id_task: str, if mask: p.extra_generation_params["Mask blur"] = mask_blur - p.extra_generation_params["Mask blend power"] = mask_blend_power - p.extra_generation_params["Mask blend scale"] = mask_blend_scale - p.extra_generation_params["Mask blend offset"] = mask_blend_offset + p.extra_generation_params["Mask blending bias"] = mask_blend_power + p.extra_generation_params["Mask blending preservation"] = mask_blend_scale + p.extra_generation_params["Mask blending detail boost"] = inpaint_detail_preservation with closing(p): if is_batch: diff --git a/modules/processing.py b/modules/processing.py index da4d6fda9..361e8b05d 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -1351,7 +1351,7 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): mask_blur: int = None mask_blend_power: float = 1 mask_blend_scale: float = 1 - mask_blend_offset: float = 0 + inpaint_detail_preservation: float = 16 inpainting_fill: int = 0 inpaint_full_res: bool = True inpaint_full_res_padding: int = 0 diff --git a/modules/sd_samplers_cfg_denoiser.py b/modules/sd_samplers_cfg_denoiser.py index c4d6fda65..598cd4876 100644 --- a/modules/sd_samplers_cfg_denoiser.py +++ b/modules/sd_samplers_cfg_denoiser.py @@ -45,7 +45,7 @@ class CFGDenoiser(torch.nn.Module): self.nmask = None self.mask_blend_power = 1 self.mask_blend_scale = 1 - self.mask_blend_offset = 0 + self.inpaint_detail_preservation = 16 self.init_latent = None self.steps = None """number of steps as specified by user in UI""" @@ -105,14 +105,13 @@ class CFGDenoiser(torch.nn.Module): # Record the original latent vector magnitudes. # We bring them to a power so that larger magnitudes are favored over smaller ones. # 64-bit operations are used here to allow large exponents. - detail_preservation = 32 - a_magnitude = torch.norm(a, p=2, dim=1).to(torch.float64) ** detail_preservation - b_magnitude = torch.norm(b, p=2, dim=1).to(torch.float64) ** detail_preservation + a_magnitude = torch.norm(a, p=2, dim=1).to(torch.float64) ** self.inpaint_detail_preservation + b_magnitude = torch.norm(b, p=2, dim=1).to(torch.float64) ** self.inpaint_detail_preservation one_minus_t = 1 - t # Interpolate the powered magnitudes, then un-power them (bring them back to a power of 1). - interp_magnitude = (a_magnitude * one_minus_t + b_magnitude * t) ** (1 / detail_preservation) + interp_magnitude = (a_magnitude * one_minus_t + b_magnitude * t) ** (1 / self.inpaint_detail_preservation) # Linearly interpolate the image vectors. image_interp = a * one_minus_t + b * t @@ -142,7 +141,7 @@ class CFGDenoiser(torch.nn.Module): NOTE: "mask" is not used """ - return torch.pow(nmask, (_sigma ** self.mask_blend_power) * self.mask_blend_scale + self.mask_blend_offset) + return torch.pow(nmask, (_sigma ** self.mask_blend_power) * self.mask_blend_scale) if state.interrupted or state.skipped: raise sd_samplers_common.InterruptedException diff --git a/modules/sd_samplers_common.py b/modules/sd_samplers_common.py index 8904da2fb..ecd8ab0a0 100644 --- a/modules/sd_samplers_common.py +++ b/modules/sd_samplers_common.py @@ -279,7 +279,7 @@ class Sampler: self.model_wrap_cfg.nmask = p.nmask if hasattr(p, 'nmask') else None self.model_wrap_cfg.mask_blend_power = p.mask_blend_power if hasattr(p, 'mask_blend_power') else None self.model_wrap_cfg.mask_blend_scale = p.mask_blend_scale if hasattr(p, 'mask_blend_scale') else None - self.model_wrap_cfg.mask_blend_offset = p.mask_blend_offset if hasattr(p, 'mask_blend_offset') else None + self.model_wrap_cfg.inpaint_detail_preservation = p.inpaint_detail_preservation if hasattr(p, 'inpaint_detail_preservation') else None self.model_wrap_cfg.step = 0 self.model_wrap_cfg.image_cfg_scale = getattr(p, 'image_cfg_scale', None) self.eta = p.eta if p.eta is not None else getattr(opts, self.eta_option_field, 0.0) diff --git a/modules/ui.py b/modules/ui.py index 86c130869..f5e201477 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -732,9 +732,9 @@ def create_ui(): with FormRow(): mask_blur = gr.Slider(label='Mask blur', minimum=0, maximum=64, step=1, value=4, elem_id="img2img_mask_blur") mask_alpha = gr.Slider(label="Mask transparency", visible=False, elem_id="img2img_mask_alpha") - mask_blend_power = gr.Slider(label='Mask blend power', minimum=0, maximum=8, step=0.1, value=1, elem_id="img2img_mask_blend_power") - mask_blend_scale = gr.Slider(label='Mask blend scale', minimum=0, maximum=8, step=0.1, value=1, elem_id="img2img_mask_blend_scale") - mask_blend_offset = gr.Slider(label='Mask blend offset', minimum=-4, maximum=4, step=0.1, value=0, elem_id="img2img_mask_blend_offset") + mask_blend_power = gr.Slider(label='Blending bias', minimum=0, maximum=8, step=0.1, value=1, elem_id="img2img_mask_blend_power") + mask_blend_scale = gr.Slider(label='Blending preservation', minimum=0, maximum=8, step=0.05, value=1, elem_id="img2img_mask_blend_scale") + inpaint_detail_preservation = gr.Slider(label='Blending detail boost', minimum=1, maximum=32, step=0.5, value=16, elem_id="img2img_mask_blend_offset") with FormRow(): inpainting_mask_invert = gr.Radio(label='Mask mode', choices=['Inpaint masked', 'Inpaint not masked'], value='Inpaint masked', type="index", elem_id="img2img_mask_mode") @@ -786,7 +786,7 @@ def create_ui(): mask_alpha, mask_blend_power, mask_blend_scale, - mask_blend_offset, + inpaint_detail_preservation, inpainting_fill, batch_count, batch_size, @@ -885,9 +885,9 @@ def create_ui(): (toprow.ui_styles.dropdown, lambda d: d["Styles array"] if isinstance(d.get("Styles array"), list) else gr.update()), (denoising_strength, "Denoising strength"), (mask_blur, "Mask blur"), - (mask_blend_power, "Mask blend power"), - (mask_blend_scale, "Mask blend scale"), - (mask_blend_offset, "Mask blend offset"), + (mask_blend_power, "Mask blending bias"), + (mask_blend_scale, "Mask blending preservation"), + (inpaint_detail_preservation, "Mask blending detail boost"), *scripts.scripts_img2img.infotext_fields ] parameters_copypaste.add_paste_fields("img2img", init_img, img2img_paste_fields, override_settings) diff --git a/scripts/outpainting_mk_2.py b/scripts/outpainting_mk_2.py index 6aa97edfa..54d95825a 100644 --- a/scripts/outpainting_mk_2.py +++ b/scripts/outpainting_mk_2.py @@ -133,16 +133,16 @@ class Script(scripts.Script): pixels = gr.Slider(label="Pixels to expand", minimum=8, maximum=256, step=8, value=128, elem_id=self.elem_id("pixels")) mask_blur = gr.Slider(label='Mask blur', minimum=0, maximum=64, step=1, value=8, elem_id=self.elem_id("mask_blur")) - mask_blend_power = gr.Slider(label='Mask blend power', minimum=0, maximum=8, step=0.1, value=1, elem_id=self.elem_id("mask_blend_power")) - mask_blend_scale = gr.Slider(label='Mask blend scale', minimum=0, maximum=8, step=0.1, value=1, elem_id=self.elem_id("mask_blend_scale")) - mask_blend_offset = gr.Slider(label='Mask blend scale', minimum=-4, maximum=4, step=0.1, value=1, elem_id=self.elem_id("mask_blend_offset")) + mask_blend_power = gr.Slider(label='Blending bias', minimum=0, maximum=8, step=0.1, value=1, elem_id=self.elem_id("mask_blend_power")) + mask_blend_scale = gr.Slider(label='Blending preservation', minimum=0, maximum=8, step=0.1, value=1, elem_id=self.elem_id("mask_blend_scale")) + inpaint_detail_preservation = gr.Slider(label='Blending detail boost', minimum=1, maximum=32, step=0.5, value=16, elem_id=self.elem_id("inpaint_detail_preservation")) direction = gr.CheckboxGroup(label="Outpainting direction", choices=['left', 'right', 'up', 'down'], value=['left', 'right', 'up', 'down'], elem_id=self.elem_id("direction")) noise_q = gr.Slider(label="Fall-off exponent (lower=higher detail)", minimum=0.0, maximum=4.0, step=0.01, value=1.0, elem_id=self.elem_id("noise_q")) color_variation = gr.Slider(label="Color variation", minimum=0.0, maximum=1.0, step=0.01, value=0.05, elem_id=self.elem_id("color_variation")) - return [info, pixels, mask_blur, mask_blend_power, mask_blend_scale, mask_blend_offset, direction, noise_q, color_variation] + return [info, pixels, mask_blur, mask_blend_power, mask_blend_scale, inpaint_detail_preservation, direction, noise_q, color_variation] - def run(self, p, _, pixels, mask_blur, mask_blend_power, mask_blend_scale, mask_blend_offset, direction, noise_q, color_variation): + def run(self, p, _, pixels, mask_blur, mask_blend_power, mask_blend_scale, inpaint_detail_preservation, direction, noise_q, color_variation): initial_seed_and_info = [None, None] process_width = p.width @@ -172,7 +172,7 @@ class Script(scripts.Script): p.mask_blur_y = mask_blur_y*4 p.mask_blend_power = mask_blend_power p.mask_blend_scale = mask_blend_scale - p.mask_blend_offset = mask_blend_offset + p.inpaint_detail_preservation = inpaint_detail_preservation init_img = p.init_images[0] target_w = math.ceil((init_img.width + left + right) / 64) * 64 diff --git a/scripts/poor_mans_outpainting.py b/scripts/poor_mans_outpainting.py index b10140f14..e3acb3d47 100644 --- a/scripts/poor_mans_outpainting.py +++ b/scripts/poor_mans_outpainting.py @@ -22,22 +22,22 @@ class Script(scripts.Script): pixels = gr.Slider(label="Pixels to expand", minimum=8, maximum=256, step=8, value=128, elem_id=self.elem_id("pixels")) mask_blur = gr.Slider(label='Mask blur', minimum=0, maximum=64, step=1, value=4, elem_id=self.elem_id("mask_blur")) - mask_blend_power = gr.Slider(label='Mask blend power', minimum=0, maximum=8, step=0.1, value=1, elem_id=self.elem_id("mask_blend_power")) - mask_blend_scale = gr.Slider(label='Mask blend scale', minimum=0, maximum=8, step=0.1, value=1, elem_id=self.elem_id("mask_blend_scale")) - mask_blend_offset = gr.Slider(label='Mask blend offset', minimum=-4, maximum=4, step=0.1, value=0, elem_id=self.elem_id("mask_blend_offset")) + mask_blend_power = gr.Slider(label='Blending bias', minimum=0, maximum=8, step=0.1, value=1, elem_id=self.elem_id("mask_blend_power")) + mask_blend_scale = gr.Slider(label='Blending preservation', minimum=0, maximum=8, step=0.1, value=1, elem_id=self.elem_id("mask_blend_scale")) + inpaint_detail_preservation = gr.Slider(label='Blending detail boost', minimum=1, maximum=32, step=0.5, value=16, elem_id=self.elem_id("inpaint_detail_preservation")) inpainting_fill = gr.Radio(label='Masked content', choices=['fill', 'original', 'latent noise', 'latent nothing'], value='fill', type="index", elem_id=self.elem_id("inpainting_fill")) direction = gr.CheckboxGroup(label="Outpainting direction", choices=['left', 'right', 'up', 'down'], value=['left', 'right', 'up', 'down'], elem_id=self.elem_id("direction")) - return [pixels, mask_blur, mask_blend_power, mask_blend_scale, mask_blend_offset, inpainting_fill, direction] + return [pixels, mask_blur, mask_blend_power, mask_blend_scale, inpaint_detail_preservation, inpainting_fill, direction] - def run(self, p, pixels, mask_blur, mask_blend_power, mask_blend_scale, mask_blend_offset, inpainting_fill, direction): + def run(self, p, pixels, mask_blur, mask_blend_power, mask_blend_scale, inpaint_detail_preservation, inpainting_fill, direction): initial_seed = None initial_info = None p.mask_blur = mask_blur * 2 p.mask_blend_power = mask_blend_power p.mask_blend_scale = mask_blend_scale - p.mask_blend_offset = mask_blend_offset + p.inpaint_detail_preservation = inpaint_detail_preservation p.inpainting_fill = inpainting_fill p.inpaint_full_res = False diff --git a/test/test_img2img.py b/test/test_img2img.py index 6289e59e1..88b06eb8d 100644 --- a/test/test_img2img.py +++ b/test/test_img2img.py @@ -26,7 +26,7 @@ def simple_img2img_request(img2img_basic_image_base64): "mask_blur": 4, "mask_blend_power": 1, "mask_blend_scale": 1, - "mask_blend_offset": 0, + "inpaint_detail_preservation": 16, "n_iter": 1, "negative_prompt": "", "override_settings": {}, From 284fd8f415ec70e14ae5de0b7f5ce738007a6b7f Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Tue, 28 Nov 2023 23:03:50 -0700 Subject: [PATCH 157/342] Tweaked UI sliders and labels. --- modules/img2img.py | 2 +- modules/ui.py | 6 +++--- scripts/outpainting_mk_2.py | 4 ++-- scripts/poor_mans_outpainting.py | 4 ++-- 4 files changed, 8 insertions(+), 8 deletions(-) diff --git a/modules/img2img.py b/modules/img2img.py index 023808d6c..0ae163654 100644 --- a/modules/img2img.py +++ b/modules/img2img.py @@ -239,7 +239,7 @@ def img2img(id_task: str, p.extra_generation_params["Mask blur"] = mask_blur p.extra_generation_params["Mask blending bias"] = mask_blend_power p.extra_generation_params["Mask blending preservation"] = mask_blend_scale - p.extra_generation_params["Mask blending detail boost"] = inpaint_detail_preservation + p.extra_generation_params["Mask blending contrast boost"] = inpaint_detail_preservation with closing(p): if is_batch: diff --git a/modules/ui.py b/modules/ui.py index f5e201477..3a9038b22 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -733,8 +733,8 @@ def create_ui(): mask_blur = gr.Slider(label='Mask blur', minimum=0, maximum=64, step=1, value=4, elem_id="img2img_mask_blur") mask_alpha = gr.Slider(label="Mask transparency", visible=False, elem_id="img2img_mask_alpha") mask_blend_power = gr.Slider(label='Blending bias', minimum=0, maximum=8, step=0.1, value=1, elem_id="img2img_mask_blend_power") - mask_blend_scale = gr.Slider(label='Blending preservation', minimum=0, maximum=8, step=0.05, value=1, elem_id="img2img_mask_blend_scale") - inpaint_detail_preservation = gr.Slider(label='Blending detail boost', minimum=1, maximum=32, step=0.5, value=16, elem_id="img2img_mask_blend_offset") + mask_blend_scale = gr.Slider(label='Blending preservation', minimum=0, maximum=8, step=0.05, value=0.5, elem_id="img2img_mask_blend_scale") + inpaint_detail_preservation = gr.Slider(label='Blending contrast boost', minimum=1, maximum=32, step=0.5, value=4, elem_id="img2img_mask_blend_offset") with FormRow(): inpainting_mask_invert = gr.Radio(label='Mask mode', choices=['Inpaint masked', 'Inpaint not masked'], value='Inpaint masked', type="index", elem_id="img2img_mask_mode") @@ -887,7 +887,7 @@ def create_ui(): (mask_blur, "Mask blur"), (mask_blend_power, "Mask blending bias"), (mask_blend_scale, "Mask blending preservation"), - (inpaint_detail_preservation, "Mask blending detail boost"), + (inpaint_detail_preservation, "Mask blending contrast boost"), *scripts.scripts_img2img.infotext_fields ] parameters_copypaste.add_paste_fields("img2img", init_img, img2img_paste_fields, override_settings) diff --git a/scripts/outpainting_mk_2.py b/scripts/outpainting_mk_2.py index 54d95825a..bd9cb61bf 100644 --- a/scripts/outpainting_mk_2.py +++ b/scripts/outpainting_mk_2.py @@ -134,8 +134,8 @@ class Script(scripts.Script): pixels = gr.Slider(label="Pixels to expand", minimum=8, maximum=256, step=8, value=128, elem_id=self.elem_id("pixels")) mask_blur = gr.Slider(label='Mask blur', minimum=0, maximum=64, step=1, value=8, elem_id=self.elem_id("mask_blur")) mask_blend_power = gr.Slider(label='Blending bias', minimum=0, maximum=8, step=0.1, value=1, elem_id=self.elem_id("mask_blend_power")) - mask_blend_scale = gr.Slider(label='Blending preservation', minimum=0, maximum=8, step=0.1, value=1, elem_id=self.elem_id("mask_blend_scale")) - inpaint_detail_preservation = gr.Slider(label='Blending detail boost', minimum=1, maximum=32, step=0.5, value=16, elem_id=self.elem_id("inpaint_detail_preservation")) + mask_blend_scale = gr.Slider(label='Blending preservation', minimum=0, maximum=8, step=0.05, value=0.5, elem_id=self.elem_id("mask_blend_scale")) + inpaint_detail_preservation = gr.Slider(label='Blending contrast boost', minimum=1, maximum=32, step=0.5, value=4, elem_id=self.elem_id("inpaint_detail_preservation")) direction = gr.CheckboxGroup(label="Outpainting direction", choices=['left', 'right', 'up', 'down'], value=['left', 'right', 'up', 'down'], elem_id=self.elem_id("direction")) noise_q = gr.Slider(label="Fall-off exponent (lower=higher detail)", minimum=0.0, maximum=4.0, step=0.01, value=1.0, elem_id=self.elem_id("noise_q")) color_variation = gr.Slider(label="Color variation", minimum=0.0, maximum=1.0, step=0.01, value=0.05, elem_id=self.elem_id("color_variation")) diff --git a/scripts/poor_mans_outpainting.py b/scripts/poor_mans_outpainting.py index e3acb3d47..5388f5db4 100644 --- a/scripts/poor_mans_outpainting.py +++ b/scripts/poor_mans_outpainting.py @@ -23,8 +23,8 @@ class Script(scripts.Script): pixels = gr.Slider(label="Pixels to expand", minimum=8, maximum=256, step=8, value=128, elem_id=self.elem_id("pixels")) mask_blur = gr.Slider(label='Mask blur', minimum=0, maximum=64, step=1, value=4, elem_id=self.elem_id("mask_blur")) mask_blend_power = gr.Slider(label='Blending bias', minimum=0, maximum=8, step=0.1, value=1, elem_id=self.elem_id("mask_blend_power")) - mask_blend_scale = gr.Slider(label='Blending preservation', minimum=0, maximum=8, step=0.1, value=1, elem_id=self.elem_id("mask_blend_scale")) - inpaint_detail_preservation = gr.Slider(label='Blending detail boost', minimum=1, maximum=32, step=0.5, value=16, elem_id=self.elem_id("inpaint_detail_preservation")) + mask_blend_scale = gr.Slider(label='Blending preservation', minimum=0, maximum=8, step=0.05, value=0.5, elem_id=self.elem_id("mask_blend_scale")) + inpaint_detail_preservation = gr.Slider(label='Blending contrast boost', minimum=1, maximum=32, step=0.5, value=4, elem_id=self.elem_id("inpaint_detail_preservation")) inpainting_fill = gr.Radio(label='Masked content', choices=['fill', 'original', 'latent noise', 'latent nothing'], value='fill', type="index", elem_id=self.elem_id("inpainting_fill")) direction = gr.CheckboxGroup(label="Outpainting direction", choices=['left', 'right', 'up', 'down'], value=['left', 'right', 'up', 'down'], elem_id=self.elem_id("direction")) From c7a1ff87207544dd4bcf3aefffa67a4a38678c16 Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Tue, 28 Nov 2023 23:31:10 -0700 Subject: [PATCH 158/342] Tweaked default values. --- modules/processing.py | 4 ++-- modules/sd_samplers_cfg_denoiser.py | 4 ++-- test/test_img2img.py | 4 ++-- 3 files changed, 6 insertions(+), 6 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index 361e8b05d..92fdebadd 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -1350,8 +1350,8 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): mask_blur_y: int = 4 mask_blur: int = None mask_blend_power: float = 1 - mask_blend_scale: float = 1 - inpaint_detail_preservation: float = 16 + mask_blend_scale: float = 0.5 + inpaint_detail_preservation: float = 4 inpainting_fill: int = 0 inpaint_full_res: bool = True inpaint_full_res_padding: int = 0 diff --git a/modules/sd_samplers_cfg_denoiser.py b/modules/sd_samplers_cfg_denoiser.py index 598cd4876..ceb612d79 100644 --- a/modules/sd_samplers_cfg_denoiser.py +++ b/modules/sd_samplers_cfg_denoiser.py @@ -44,8 +44,8 @@ class CFGDenoiser(torch.nn.Module): self.mask = None self.nmask = None self.mask_blend_power = 1 - self.mask_blend_scale = 1 - self.inpaint_detail_preservation = 16 + self.mask_blend_scale = 0.5 + self.inpaint_detail_preservation = 4 self.init_latent = None self.steps = None """number of steps as specified by user in UI""" diff --git a/test/test_img2img.py b/test/test_img2img.py index 88b06eb8d..5cda2dbae 100644 --- a/test/test_img2img.py +++ b/test/test_img2img.py @@ -25,8 +25,8 @@ def simple_img2img_request(img2img_basic_image_base64): "mask": None, "mask_blur": 4, "mask_blend_power": 1, - "mask_blend_scale": 1, - "inpaint_detail_preservation": 16, + "mask_blend_scale": 0.5, + "inpaint_detail_preservation": 4, "n_iter": 1, "negative_prompt": "", "override_settings": {}, From b25c126ccdbc4da22ade46597a9addf808998989 Mon Sep 17 00:00:00 2001 From: drhead <1313496+drhead@users.noreply.github.com> Date: Wed, 29 Nov 2023 17:38:53 -0500 Subject: [PATCH 159/342] Protect alphas_cumprod from downcasting --- modules/sd_models.py | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/modules/sd_models.py b/modules/sd_models.py index 841402e86..de80a493a 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -387,7 +387,11 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer if shared.cmd_opts.upcast_sampling and depth_model: model.depth_model = None + alphas_cumprod = model.alphas_cumprod + model.alphas_cumprod = None model.half() + model.alphas_cumprod = alphas_cumprod + model.alphas_cumprod_original = alphas_cumprod model.first_stage_model = vae if depth_model: model.depth_model = depth_model @@ -642,6 +646,7 @@ def load_model(checkpoint_info=None, already_loaded_state_dict=None): else: weight_dtype_conversion = { 'first_stage_model': None, + 'alphas_cumprod': None, '': torch.float16, } From 588a52891dca4d030ca7028dd9c0b56022a68b57 Mon Sep 17 00:00:00 2001 From: drhead <1313496+drhead@users.noreply.github.com> Date: Wed, 29 Nov 2023 17:40:23 -0500 Subject: [PATCH 160/342] Add options for zero terminal SNR --- modules/shared_options.py | 2 ++ 1 file changed, 2 insertions(+) diff --git a/modules/shared_options.py b/modules/shared_options.py index 04e68a712..515967770 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -218,6 +218,7 @@ options_templates.update(options_section(('compatibility', "Compatibility", "sd" "dont_fix_second_order_samplers_schedule": OptionInfo(False, "Do not fix prompt schedule for second order samplers."), "hires_fix_use_firstpass_conds": OptionInfo(False, "For hires fix, calculate conds of second pass using extra networks of first pass."), "use_old_scheduling": OptionInfo(False, "Use old prompt editing timelines.", infotext="Old prompt editing timelines").info("For [red:green:N]; old: If N < 1, it's a fraction of steps (and hires fix uses range from 0 to 1), if N >= 1, it's an absolute number of steps; new: If N has a decimal point in it, it's a fraction of steps (and hires fix uses range from 1 to 2), othewrwise it's an absolute number of steps"), + "use_downcasted_alpha_bar": OptionInfo(False, "Downcast model alphas_cumprod to fp16 before sampling. For reproducing old seeds.") })) options_templates.update(options_section(('interrogate', "Interrogate"), { @@ -335,6 +336,7 @@ options_templates.update(options_section(('sampler-params', "Sampler parameters" 'uni_pc_skip_type': OptionInfo("time_uniform", "UniPC skip type", gr.Radio, {"choices": ["time_uniform", "time_quadratic", "logSNR"]}, infotext='UniPC skip type'), 'uni_pc_order': OptionInfo(3, "UniPC order", gr.Slider, {"minimum": 1, "maximum": 50, "step": 1}, infotext='UniPC order').info("must be < sampling steps"), 'uni_pc_lower_order_final': OptionInfo(True, "UniPC lower order final", infotext='UniPC lower order final'), + 'sd_noise_schedule': OptionInfo("Default", "Noise schedule for sampling", gr.Radio, {"choices": ["Default", "Zero Terminal SNR"]}, infotext="Noise schedule for sampling").info("for use with zero terminal SNR trained models") })) options_templates.update(options_section(('postprocessing', "Postprocessing", "postprocessing"), { From 6d0a8dcd892f7ad9b399fed6edbad6ede13c5f69 Mon Sep 17 00:00:00 2001 From: drhead <1313496+drhead@users.noreply.github.com> Date: Wed, 29 Nov 2023 17:42:07 -0500 Subject: [PATCH 161/342] Implement zero terminal SNR schedule option --- modules/processing.py | 28 ++++++++++++++++++++++++++++ 1 file changed, 28 insertions(+) diff --git a/modules/processing.py b/modules/processing.py index ac58ef869..c88eec705 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -863,6 +863,34 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: if p.n_iter > 1: shared.state.job = f"Batch {n+1} out of {p.n_iter}" + + def rescale_zero_terminal_snr_abar(alphas_cumprod): + alphas_bar_sqrt = alphas_cumprod.sqrt() + + # Store old values. + alphas_bar_sqrt_0 = alphas_bar_sqrt[0].clone() + alphas_bar_sqrt_T = alphas_bar_sqrt[-1].clone() + + # Shift so the last timestep is zero. + alphas_bar_sqrt -= (alphas_bar_sqrt_T) + + # Scale so the first timestep is back to the old value. + alphas_bar_sqrt *= alphas_bar_sqrt_0 / (alphas_bar_sqrt_0 - alphas_bar_sqrt_T) + + # Convert alphas_bar_sqrt to betas + alphas_bar = alphas_bar_sqrt**2 # Revert sqrt + alphas_bar[-1] = 4.8973451890853435e-08 + return alphas_bar + + p.sd_model.alphas_cumprod = p.sd_model.alphas_cumprod_original.to(shared.device) + + if opts.use_downcasted_alpha_bar: + p.extra_generation_params['Downcast alphas_cumprod'] = opts.use_downcasted_alpha_bar + p.sd_model.alphas_cumprod = p.sd_model.alphas_cumprod.half().to(shared.device) + if opts.sd_noise_schedule == "Zero Terminal SNR": + p.extra_generation_params['Noise Schedule'] = opts.sd_noise_schedule + print("rescaling noise schedule for zero snr") + p.sd_model.alphas_cumprod = rescale_zero_terminal_snr_abar(p.sd_model.alphas_cumprod).to(shared.device) with devices.without_autocast() if devices.unet_needs_upcast else devices.autocast(): samples_ddim = p.sample(conditioning=p.c, unconditional_conditioning=p.uc, seeds=p.seeds, subseeds=p.subseeds, subseed_strength=p.subseed_strength, prompts=p.prompts) From ec6ee5c13bf3453f8703e225a191333a9bbcf10a Mon Sep 17 00:00:00 2001 From: catboxanon <122327233+catboxanon@users.noreply.github.com> Date: Wed, 29 Nov 2023 18:10:27 -0500 Subject: [PATCH 162/342] Fix infotext for ztSNR --- modules/shared_options.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/modules/shared_options.py b/modules/shared_options.py index 515967770..bc3d56dec 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -218,7 +218,7 @@ options_templates.update(options_section(('compatibility', "Compatibility", "sd" "dont_fix_second_order_samplers_schedule": OptionInfo(False, "Do not fix prompt schedule for second order samplers."), "hires_fix_use_firstpass_conds": OptionInfo(False, "For hires fix, calculate conds of second pass using extra networks of first pass."), "use_old_scheduling": OptionInfo(False, "Use old prompt editing timelines.", infotext="Old prompt editing timelines").info("For [red:green:N]; old: If N < 1, it's a fraction of steps (and hires fix uses range from 0 to 1), if N >= 1, it's an absolute number of steps; new: If N has a decimal point in it, it's a fraction of steps (and hires fix uses range from 1 to 2), othewrwise it's an absolute number of steps"), - "use_downcasted_alpha_bar": OptionInfo(False, "Downcast model alphas_cumprod to fp16 before sampling. For reproducing old seeds.") + "use_downcasted_alpha_bar": OptionInfo(False, "Downcast model alphas_cumprod to fp16 before sampling. For reproducing old seeds.", infotext="Downcast alphas_cumprod") })) options_templates.update(options_section(('interrogate', "Interrogate"), { @@ -336,7 +336,7 @@ options_templates.update(options_section(('sampler-params', "Sampler parameters" 'uni_pc_skip_type': OptionInfo("time_uniform", "UniPC skip type", gr.Radio, {"choices": ["time_uniform", "time_quadratic", "logSNR"]}, infotext='UniPC skip type'), 'uni_pc_order': OptionInfo(3, "UniPC order", gr.Slider, {"minimum": 1, "maximum": 50, "step": 1}, infotext='UniPC order').info("must be < sampling steps"), 'uni_pc_lower_order_final': OptionInfo(True, "UniPC lower order final", infotext='UniPC lower order final'), - 'sd_noise_schedule': OptionInfo("Default", "Noise schedule for sampling", gr.Radio, {"choices": ["Default", "Zero Terminal SNR"]}, infotext="Noise schedule for sampling").info("for use with zero terminal SNR trained models") + 'sd_noise_schedule': OptionInfo("Default", "Noise schedule for sampling", gr.Radio, {"choices": ["Default", "Zero Terminal SNR"]}, infotext="Noise Schedule").info("for use with zero terminal SNR trained models") })) options_templates.update(options_section(('postprocessing', "Postprocessing", "postprocessing"), { From ffa7f8201d849636bb327b3b40298e7c169ff204 Mon Sep 17 00:00:00 2001 From: catboxanon <122327233+catboxanon@users.noreply.github.com> Date: Wed, 29 Nov 2023 18:10:43 -0500 Subject: [PATCH 163/342] Lint --- modules/processing.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index c88eec705..f3883d5b3 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -863,7 +863,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: if p.n_iter > 1: shared.state.job = f"Batch {n+1} out of {p.n_iter}" - + def rescale_zero_terminal_snr_abar(alphas_cumprod): alphas_bar_sqrt = alphas_cumprod.sqrt() @@ -881,7 +881,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: alphas_bar = alphas_bar_sqrt**2 # Revert sqrt alphas_bar[-1] = 4.8973451890853435e-08 return alphas_bar - + p.sd_model.alphas_cumprod = p.sd_model.alphas_cumprod_original.to(shared.device) if opts.use_downcasted_alpha_bar: From de79597ab9894965e3702939b8536ec3dcc3c859 Mon Sep 17 00:00:00 2001 From: catboxanon <122327233+catboxanon@users.noreply.github.com> Date: Wed, 29 Nov 2023 18:33:32 -0500 Subject: [PATCH 164/342] Only apply ztSNR related code if alphas_cumprod exists --- modules/processing.py | 17 +++++++++-------- 1 file changed, 9 insertions(+), 8 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index f3883d5b3..7e73d7e29 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -882,15 +882,16 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: alphas_bar[-1] = 4.8973451890853435e-08 return alphas_bar - p.sd_model.alphas_cumprod = p.sd_model.alphas_cumprod_original.to(shared.device) + if hasattr(p.sd_model, 'alphas_cumprod') and hasattr(p.sd_model, 'alphas_cumprod_original'): + p.sd_model.alphas_cumprod = p.sd_model.alphas_cumprod_original.to(shared.device) - if opts.use_downcasted_alpha_bar: - p.extra_generation_params['Downcast alphas_cumprod'] = opts.use_downcasted_alpha_bar - p.sd_model.alphas_cumprod = p.sd_model.alphas_cumprod.half().to(shared.device) - if opts.sd_noise_schedule == "Zero Terminal SNR": - p.extra_generation_params['Noise Schedule'] = opts.sd_noise_schedule - print("rescaling noise schedule for zero snr") - p.sd_model.alphas_cumprod = rescale_zero_terminal_snr_abar(p.sd_model.alphas_cumprod).to(shared.device) + if opts.use_downcasted_alpha_bar: + p.extra_generation_params['Downcast alphas_cumprod'] = opts.use_downcasted_alpha_bar + p.sd_model.alphas_cumprod = p.sd_model.alphas_cumprod.half().to(shared.device) + if opts.sd_noise_schedule == "Zero Terminal SNR": + p.extra_generation_params['Noise Schedule'] = opts.sd_noise_schedule + print("rescaling noise schedule for zero snr") + p.sd_model.alphas_cumprod = rescale_zero_terminal_snr_abar(p.sd_model.alphas_cumprod).to(shared.device) with devices.without_autocast() if devices.unet_needs_upcast else devices.autocast(): samples_ddim = p.sample(conditioning=p.c, unconditional_conditioning=p.uc, seeds=p.seeds, subseeds=p.subseeds, subseed_strength=p.subseed_strength, prompts=p.prompts) From 668ae34e21df848ef4909b8b49c4142a3674701b Mon Sep 17 00:00:00 2001 From: drhead <1313496+drhead@users.noreply.github.com> Date: Wed, 29 Nov 2023 22:48:31 -0500 Subject: [PATCH 165/342] remove debug print --- modules/processing.py | 1 - 1 file changed, 1 deletion(-) diff --git a/modules/processing.py b/modules/processing.py index 7e73d7e29..d73c8bfc1 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -890,7 +890,6 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: p.sd_model.alphas_cumprod = p.sd_model.alphas_cumprod.half().to(shared.device) if opts.sd_noise_schedule == "Zero Terminal SNR": p.extra_generation_params['Noise Schedule'] = opts.sd_noise_schedule - print("rescaling noise schedule for zero snr") p.sd_model.alphas_cumprod = rescale_zero_terminal_snr_abar(p.sd_model.alphas_cumprod).to(shared.device) with devices.without_autocast() if devices.unet_needs_upcast else devices.autocast(): From 8b40f475a31109cc6ecbdc0d14a0cee9e0303291 Mon Sep 17 00:00:00 2001 From: Nuullll Date: Fri, 10 Nov 2023 11:06:26 +0800 Subject: [PATCH 166/342] Initial IPEX support --- modules/devices.py | 11 +++++++++-- modules/xpu_specific.py | 42 +++++++++++++++++++++++++++++++++++++++++ 2 files changed, 51 insertions(+), 2 deletions(-) create mode 100644 modules/xpu_specific.py diff --git a/modules/devices.py b/modules/devices.py index 1d4eb5635..be599736c 100644 --- a/modules/devices.py +++ b/modules/devices.py @@ -3,7 +3,7 @@ import contextlib from functools import lru_cache import torch -from modules import errors, shared +from modules import errors, shared, xpu_specific if sys.platform == "darwin": from modules import mac_specific @@ -30,6 +30,9 @@ def get_optimal_device_name(): if has_mps(): return "mps" + if xpu_specific.has_ipex: + return xpu_specific.get_xpu_device_string() + return "cpu" @@ -100,11 +103,15 @@ def autocast(disable=False): if dtype == torch.float32 or shared.cmd_opts.precision == "full": return contextlib.nullcontext() + if xpu_specific.has_xpu: + return torch.autocast("xpu") + return torch.autocast("cuda") def without_autocast(disable=False): - return torch.autocast("cuda", enabled=False) if torch.is_autocast_enabled() and not disable else contextlib.nullcontext() + device_type = "xpu" if xpu_specific.has_xpu else "cuda" + return torch.autocast(device_type, enabled=False) if torch.is_autocast_enabled() and not disable else contextlib.nullcontext() class NansException(Exception): diff --git a/modules/xpu_specific.py b/modules/xpu_specific.py new file mode 100644 index 000000000..6417dd2d6 --- /dev/null +++ b/modules/xpu_specific.py @@ -0,0 +1,42 @@ +import contextlib +from modules import shared +from modules.sd_hijack_utils import CondFunc + +has_ipex = False +try: + import torch + import intel_extension_for_pytorch as ipex + has_ipex = True +except Exception: + pass + +def check_for_xpu(): + if not has_ipex: + return False + + return hasattr(torch, 'xpu') and torch.xpu.is_available() + +has_xpu = check_for_xpu() + +def get_xpu_device_string(): + if shared.cmd_opts.device_id is not None: + return f"xpu:{shared.cmd_opts.device_id}" + return "xpu" + +def return_null_context(*args, **kwargs): # pylint: disable=unused-argument + return contextlib.nullcontext() + +if has_xpu: + CondFunc('torch.Generator', + lambda orig_func, device=None: torch.xpu.Generator(device), + lambda orig_func, device=None: device is not None and device != torch.device("cpu") and device != "cpu") + + CondFunc('torch.nn.functional.layer_norm', + lambda orig_func, input, normalized_shape=None, weight=None, *args, **kwargs: + orig_func(input.to(weight.data.dtype), normalized_shape, weight, *args, **kwargs), + lambda orig_func, input, normalized_shape=None, weight=None, *args, **kwargs: + weight is not None and input.dtype != weight.data.dtype) + + CondFunc('torch.nn.modules.GroupNorm.forward', + lambda orig_func, self, input: orig_func(self, input.to(self.weight.data.dtype)), + lambda orig_func, self, input: input.dtype != self.weight.data.dtype) From c2ed4132037a32cda856e8ba6e2cda32b44b9784 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Fri, 1 Dec 2023 02:59:41 +0900 Subject: [PATCH 167/342] add max-heigh/width to global-popup-inner prevent the pop-up from being too big as to making exiting the pop-up impossible --- style.css | 2 ++ 1 file changed, 2 insertions(+) diff --git a/style.css b/style.css index 6e3ca8411..ee39a57b7 100644 --- a/style.css +++ b/style.css @@ -646,6 +646,8 @@ table.popup-table .link{ margin: auto; padding: 2em; z-index: 1001; + max-height: 90%; + max-width: 90%; } /* fullpage image viewer */ From 01c8f1803a77c63b2ebfd3cbbd41659fb914f274 Mon Sep 17 00:00:00 2001 From: missionfloyd Date: Thu, 30 Nov 2023 22:36:12 -0700 Subject: [PATCH 168/342] Close popups with escape key --- javascript/extraNetworks.js | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/javascript/extraNetworks.js b/javascript/extraNetworks.js index a787372cf..98a7abb74 100644 --- a/javascript/extraNetworks.js +++ b/javascript/extraNetworks.js @@ -392,3 +392,9 @@ function extraNetworksRefreshSingleCard(page, tabname, name) { } }); } + +window.addEventListener("keydown", function(event) { + if (event.key == "Escape") { + closePopup(); + } +}); From 293f44e6c1de7bbf744a4236db81ac4559bdb82a Mon Sep 17 00:00:00 2001 From: MrCheeze Date: Fri, 1 Dec 2023 22:56:08 -0500 Subject: [PATCH 169/342] Fix bug where is_using_v_parameterization_for_sd2 fails because the sd_hijack is only partially undone --- modules/sd_hijack.py | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/modules/sd_hijack.py b/modules/sd_hijack.py index 0157e19f0..3d340fc9b 100644 --- a/modules/sd_hijack.py +++ b/modules/sd_hijack.py @@ -38,9 +38,6 @@ ldm.models.diffusion.ddpm.print = shared.ldm_print optimizers = [] current_optimizer: sd_hijack_optimizations.SdOptimization = None -ldm_original_forward = patches.patch(__file__, ldm.modules.diffusionmodules.openaimodel.UNetModel, "forward", sd_unet.UNetModel_forward) -sgm_original_forward = patches.patch(__file__, sgm.modules.diffusionmodules.openaimodel.UNetModel, "forward", sd_unet.UNetModel_forward) - def list_optimizers(): new_optimizers = script_callbacks.list_optimizers_callback() @@ -258,6 +255,9 @@ class StableDiffusionModelHijack: import modules.models.diffusion.ddpm_edit + ldm_original_forward = patches.patch(__file__, ldm.modules.diffusionmodules.openaimodel.UNetModel, "forward", sd_unet.UNetModel_forward) + sgm_original_forward = patches.patch(__file__, sgm.modules.diffusionmodules.openaimodel.UNetModel, "forward", sd_unet.UNetModel_forward) + if isinstance(m, ldm.models.diffusion.ddpm.LatentDiffusion): sd_unet.original_forward = ldm_original_forward elif isinstance(m, modules.models.diffusion.ddpm_edit.LatentDiffusion): @@ -303,6 +303,9 @@ class StableDiffusionModelHijack: self.layers = None self.clip = None + patches.undo(__file__, ldm.modules.diffusionmodules.openaimodel.UNetModel, "forward") + patches.undo(__file__, sgm.modules.diffusionmodules.openaimodel.UNetModel, "forward") + sd_unet.original_forward = None From 6080045b2a0964e63bdcd33dd26015f8a51411f6 Mon Sep 17 00:00:00 2001 From: MrCheeze Date: Fri, 1 Dec 2023 22:58:05 -0500 Subject: [PATCH 170/342] Add support for SD 2.1 Turbo, by converting the state dict from SGM to LDM on load --- modules/sd_models.py | 17 +++++++++++++---- 1 file changed, 13 insertions(+), 4 deletions(-) diff --git a/modules/sd_models.py b/modules/sd_models.py index 841402e86..9355f1e16 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -230,15 +230,19 @@ def select_checkpoint(): return checkpoint_info -checkpoint_dict_replacements = { +checkpoint_dict_replacements_sd1 = { 'cond_stage_model.transformer.embeddings.': 'cond_stage_model.transformer.text_model.embeddings.', 'cond_stage_model.transformer.encoder.': 'cond_stage_model.transformer.text_model.encoder.', 'cond_stage_model.transformer.final_layer_norm.': 'cond_stage_model.transformer.text_model.final_layer_norm.', } +checkpoint_dict_replacements_sd2_turbo = { # Converts SD 2.1 Turbo from SGM to LDM format. + 'conditioner.embedders.0.': 'cond_stage_model.', +} -def transform_checkpoint_dict_key(k): - for text, replacement in checkpoint_dict_replacements.items(): + +def transform_checkpoint_dict_key(k, replacements): + for text, replacement in replacements.items(): if k.startswith(text): k = replacement + k[len(text):] @@ -249,9 +253,14 @@ def get_state_dict_from_checkpoint(pl_sd): pl_sd = pl_sd.pop("state_dict", pl_sd) pl_sd.pop("state_dict", None) + is_sd2_turbo = 'conditioner.embedders.0.model.ln_final.weight' in pl_sd and pl_sd['conditioner.embedders.0.model.ln_final.weight'].size()[0] == 1024 + sd = {} for k, v in pl_sd.items(): - new_key = transform_checkpoint_dict_key(k) + if is_sd2_turbo: + new_key = transform_checkpoint_dict_key(k, checkpoint_dict_replacements_sd2_turbo) + else: + new_key = transform_checkpoint_dict_key(k, checkpoint_dict_replacements_sd1) if new_key is not None: sd[new_key] = v From b58d061e41cba6fb91910d310d53e175d0511650 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sat, 2 Dec 2023 08:33:28 +0300 Subject: [PATCH 171/342] infotext updates: add option to disregard certain infotext fields, add option to not include VAE in infotext, add explanation to infotext settings page, move some options to infotext settings page --- modules/generation_parameters_copypaste.py | 13 +++++++++---- modules/processing.py | 4 ++-- modules/shared_items.py | 16 ++++++++++++++++ modules/shared_options.py | 20 ++++++++++++++------ 4 files changed, 41 insertions(+), 12 deletions(-) diff --git a/modules/generation_parameters_copypaste.py b/modules/generation_parameters_copypaste.py index 0a606515b..4efe53e0c 100644 --- a/modules/generation_parameters_copypaste.py +++ b/modules/generation_parameters_copypaste.py @@ -1,3 +1,4 @@ +from __future__ import annotations import base64 import io import json @@ -15,9 +16,6 @@ re_imagesize = re.compile(r"^(\d+)x(\d+)$") re_hypernet_hash = re.compile("\(([0-9a-f]+)\)$") type_of_gr_update = type(gr.update()) -paste_fields = {} -registered_param_bindings = [] - class ParamBinding: def __init__(self, paste_button, tabname, source_text_component=None, source_image_component=None, source_tabname=None, override_settings_component=None, paste_field_names=None): @@ -30,6 +28,10 @@ class ParamBinding: self.paste_field_names = paste_field_names or [] +paste_fields: dict[str, dict] = {} +registered_param_bindings: list[ParamBinding] = [] + + def reset(): paste_fields.clear() registered_param_bindings.clear() @@ -113,7 +115,6 @@ def register_paste_params_button(binding: ParamBinding): def connect_paste_params_buttons(): - binding: ParamBinding for binding in registered_param_bindings: destination_image_component = paste_fields[binding.tabname]["init_img"] fields = paste_fields[binding.tabname]["fields"] @@ -313,6 +314,9 @@ Steps: 20, Sampler: Euler a, CFG scale: 7, Seed: 965400086, Size: 512x512, Model if "VAE Decoder" not in res: res["VAE Decoder"] = "Full" + skip = set(shared.opts.infotext_skip_pasting) + res = {k: v for k, v in res.items() if k not in skip} + return res @@ -443,3 +447,4 @@ def connect_paste(button, paste_fields, input_comp, override_settings_component, outputs=[], show_progress=False, ) + diff --git a/modules/processing.py b/modules/processing.py index ac58ef869..5ab6dddef 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -679,8 +679,8 @@ def create_infotext(p, all_prompts, all_seeds, all_subseeds, comments=None, iter "Size": f"{p.width}x{p.height}", "Model hash": p.sd_model_hash if opts.add_model_hash_to_info else None, "Model": p.sd_model_name if opts.add_model_name_to_info else None, - "VAE hash": p.sd_vae_hash if opts.add_model_hash_to_info else None, - "VAE": p.sd_vae_name if opts.add_model_name_to_info else None, + "VAE hash": p.sd_vae_hash if opts.add_vae_hash_to_info else None, + "VAE": p.sd_vae_name if opts.add_vae_name_to_info else None, "Variation seed": (None if p.subseed_strength == 0 else (p.all_subseeds[0] if use_main_prompt else all_subseeds[index])), "Variation seed strength": (None if p.subseed_strength == 0 else p.subseed_strength), "Seed resize from": (None if p.seed_resize_from_w <= 0 or p.seed_resize_from_h <= 0 else f"{p.seed_resize_from_w}x{p.seed_resize_from_h}"), diff --git a/modules/shared_items.py b/modules/shared_items.py index 5024b4268..991971ad0 100644 --- a/modules/shared_items.py +++ b/modules/shared_items.py @@ -66,6 +66,22 @@ def reload_hypernetworks(): shared.hypernetworks = hypernetwork.list_hypernetworks(cmd_opts.hypernetwork_dir) +def get_infotext_names(): + from modules import generation_parameters_copypaste, shared + res = {} + + for info in shared.opts.data_labels.values(): + if info.infotext: + res[info.infotext] = 1 + + for tab_data in generation_parameters_copypaste.paste_fields.values(): + for _, name in tab_data.get("fields") or []: + if isinstance(name, str): + res[name] = 1 + + return list(res) + + ui_reorder_categories_builtin_items = [ "prompt", "image", diff --git a/modules/shared_options.py b/modules/shared_options.py index 04e68a712..df45fc0a0 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -46,8 +46,6 @@ options_templates.update(options_section(('saving-images', "Saving images/grids" "grid_text_inactive_color": OptionInfo("#999999", "Inactive text color for image grids", ui_components.FormColorPicker, {}), "grid_background_color": OptionInfo("#ffffff", "Background color for image grids", ui_components.FormColorPicker, {}), - "enable_pnginfo": OptionInfo(True, "Save text information about generation parameters as chunks to png files"), - "save_txt": OptionInfo(False, "Create a text file next to every image with generation parameters."), "save_images_before_face_restoration": OptionInfo(False, "Save a copy of image before doing face restoration."), "save_images_before_highres_fix": OptionInfo(False, "Save a copy of image before applying highres fix."), "save_images_before_color_correction": OptionInfo(False, "Save a copy of image before applying color correction to img2img results"), @@ -288,11 +286,21 @@ options_templates.update(options_section(('ui', "User interface", "ui"), { options_templates.update(options_section(('infotext', "Infotext", "ui"), { - "add_model_hash_to_info": OptionInfo(True, "Add model hash to generation information"), - "add_model_name_to_info": OptionInfo(True, "Add model name to generation information"), - "add_user_name_to_info": OptionInfo(False, "Add user name to generation information when authenticated"), - "add_version_to_infotext": OptionInfo(True, "Add program version to generation information"), + "infotext_explanation": OptionHTML(""" +Infotext is what this software calls the text that contains generation parameters and can be used to generate the same picture again. +It is displayed in UI below the image. To use infotext, paste it into the prompt and click the ↙️ paste button. +"""), + "enable_pnginfo": OptionInfo(True, "Write infotext to metadata of the generated image"), + "save_txt": OptionInfo(False, "Create a text file with infotext next to every generated image"), + + "add_model_name_to_info": OptionInfo(True, "Add model name to infotext"), + "add_model_hash_to_info": OptionInfo(True, "Add model hash to infotext"), + "add_vae_name_to_info": OptionInfo(True, "Add VAE name to infotext"), + "add_vae_hash_to_info": OptionInfo(True, "Add VAE hash to infotext"), + "add_user_name_to_info": OptionInfo(False, "Add user name to infotext when authenticated"), + "add_version_to_infotext": OptionInfo(True, "Add program version to infotext"), "disable_weights_auto_swap": OptionInfo(True, "Disregard checkpoint information from pasted infotext").info("when reading generation parameters from text into UI"), + "infotext_skip_pasting": OptionInfo([], "Disregard fields from pasted infotext", ui_components.DropdownMulti, lambda: {"choices": shared_items.get_infotext_names()}), "infotext_styles": OptionInfo("Apply if any", "Infer styles from prompts of pasted infotext", gr.Radio, {"choices": ["Ignore", "Apply", "Discard", "Apply if any"]}).info("when reading generation parameters from text into UI)").html("""
  • Ignore: keep prompt and styles dropdown as it is.
  • Apply: remove style text from prompt, always replace styles dropdown value with found styles (even if none are found).
    • From 7499148ad4dbd3444215c843d02453f68c459707 Mon Sep 17 00:00:00 2001 From: Nuullll Date: Sat, 2 Dec 2023 14:00:46 +0800 Subject: [PATCH 172/342] Disable ipex autocast due to its bad perf --- modules/cmd_args.py | 1 + modules/devices.py | 20 +++++++++++++------- modules/xpu_specific.py | 28 ++++++++++++++++++---------- webui-ipex-user.bat | 19 +++++++++++++++++++ 4 files changed, 51 insertions(+), 17 deletions(-) create mode 100644 webui-ipex-user.bat diff --git a/modules/cmd_args.py b/modules/cmd_args.py index a9fb9bfa3..da93eb266 100644 --- a/modules/cmd_args.py +++ b/modules/cmd_args.py @@ -70,6 +70,7 @@ parser.add_argument("--opt-sdp-no-mem-attention", action='store_true', help="pre parser.add_argument("--disable-opt-split-attention", action='store_true', help="prefer no cross-attention layer optimization for automatic choice of optimization") parser.add_argument("--disable-nan-check", action='store_true', help="do not check if produced images/latent spaces have nans; useful for running without a checkpoint in CI") parser.add_argument("--use-cpu", nargs='+', help="use CPU as torch device for specified modules", default=[], type=str.lower) +parser.add_argument("--use-ipex", action="store_true", help="use Intel XPU as torch device") parser.add_argument("--disable-model-loading-ram-optimization", action='store_true', help="disable an optimization that reduces RAM use when loading a model") parser.add_argument("--listen", action='store_true', help="launch gradio with 0.0.0.0 as server name, allowing to respond to network requests") parser.add_argument("--port", type=int, help="launch gradio with given server port, you need root/admin rights for ports < 1024, defaults to 7860 if available", default=None) diff --git a/modules/devices.py b/modules/devices.py index be599736c..37ecca784 100644 --- a/modules/devices.py +++ b/modules/devices.py @@ -3,11 +3,18 @@ import contextlib from functools import lru_cache import torch -from modules import errors, shared, xpu_specific +from modules import errors, shared if sys.platform == "darwin": from modules import mac_specific +if shared.cmd_opts.use_ipex: + from modules import xpu_specific + + +def has_xpu() -> bool: + return shared.cmd_opts.use_ipex and xpu_specific.has_xpu + def has_mps() -> bool: if sys.platform != "darwin": @@ -30,7 +37,7 @@ def get_optimal_device_name(): if has_mps(): return "mps" - if xpu_specific.has_ipex: + if has_xpu(): return xpu_specific.get_xpu_device_string() return "cpu" @@ -57,6 +64,9 @@ def torch_gc(): if has_mps(): mac_specific.torch_mps_gc() + if has_xpu(): + xpu_specific.torch_xpu_gc() + def enable_tf32(): if torch.cuda.is_available(): @@ -103,15 +113,11 @@ def autocast(disable=False): if dtype == torch.float32 or shared.cmd_opts.precision == "full": return contextlib.nullcontext() - if xpu_specific.has_xpu: - return torch.autocast("xpu") - return torch.autocast("cuda") def without_autocast(disable=False): - device_type = "xpu" if xpu_specific.has_xpu else "cuda" - return torch.autocast(device_type, enabled=False) if torch.is_autocast_enabled() and not disable else contextlib.nullcontext() + return torch.autocast("cuda", enabled=False) if torch.is_autocast_enabled() and not disable else contextlib.nullcontext() class NansException(Exception): diff --git a/modules/xpu_specific.py b/modules/xpu_specific.py index 6417dd2d6..2df68665a 100644 --- a/modules/xpu_specific.py +++ b/modules/xpu_specific.py @@ -1,4 +1,3 @@ -import contextlib from modules import shared from modules.sd_hijack_utils import CondFunc @@ -10,33 +9,42 @@ try: except Exception: pass + def check_for_xpu(): - if not has_ipex: - return False + return has_ipex and hasattr(torch, 'xpu') and torch.xpu.is_available() - return hasattr(torch, 'xpu') and torch.xpu.is_available() - -has_xpu = check_for_xpu() def get_xpu_device_string(): if shared.cmd_opts.device_id is not None: return f"xpu:{shared.cmd_opts.device_id}" return "xpu" -def return_null_context(*args, **kwargs): # pylint: disable=unused-argument - return contextlib.nullcontext() + +def torch_xpu_gc(): + with torch.xpu.device(get_xpu_device_string()): + torch.xpu.empty_cache() + + +has_xpu = check_for_xpu() if has_xpu: + # W/A for https://github.com/intel/intel-extension-for-pytorch/issues/452: torch.Generator API doesn't support XPU device CondFunc('torch.Generator', lambda orig_func, device=None: torch.xpu.Generator(device), - lambda orig_func, device=None: device is not None and device != torch.device("cpu") and device != "cpu") + lambda orig_func, device=None: device is not None and device.type == "xpu") + # W/A for some OPs that could not handle different input dtypes CondFunc('torch.nn.functional.layer_norm', lambda orig_func, input, normalized_shape=None, weight=None, *args, **kwargs: orig_func(input.to(weight.data.dtype), normalized_shape, weight, *args, **kwargs), lambda orig_func, input, normalized_shape=None, weight=None, *args, **kwargs: weight is not None and input.dtype != weight.data.dtype) - CondFunc('torch.nn.modules.GroupNorm.forward', lambda orig_func, self, input: orig_func(self, input.to(self.weight.data.dtype)), lambda orig_func, self, input: input.dtype != self.weight.data.dtype) + CondFunc('torch.nn.modules.linear.Linear.forward', + lambda orig_func, self, input: orig_func(self, input.to(self.weight.data.dtype)), + lambda orig_func, self, input: input.dtype != self.weight.data.dtype) + CondFunc('torch.nn.modules.conv.Conv2d.forward', + lambda orig_func, self, input: orig_func(self, input.to(self.weight.data.dtype)), + lambda orig_func, self, input: input.dtype != self.weight.data.dtype) diff --git a/webui-ipex-user.bat b/webui-ipex-user.bat new file mode 100644 index 000000000..ab25a0400 --- /dev/null +++ b/webui-ipex-user.bat @@ -0,0 +1,19 @@ +@echo off + +set PYTHON= +@REM The "Nuullll/intel-extension-for-pytorch" wheels were built from IPEX source for Intel Arc GPU: https://github.com/intel/intel-extension-for-pytorch/tree/xpu-main +@REM This is NOT an Intel official release so please use it at your own risk!! +@REM See https://github.com/Nuullll/intel-extension-for-pytorch/releases/tag/v2.0.110%2Bxpu-master%2Bdll-bundle for details. +@REM +@REM Strengths (over official IPEX 2.0.110 windows release): +@REM - AOT build (for Arc GPU only) to eliminate JIT compilation overhead: https://github.com/intel/intel-extension-for-pytorch/issues/399 +@REM - Bundles minimal oneAPI 2023.2 dependencies into the python wheels, so users don't need to install oneAPI for the whole system. +@REM - Provides a compatible torchvision wheel: https://github.com/intel/intel-extension-for-pytorch/issues/465 +@REM Limitation: +@REM - Only works for python 3.10 +set "TORCH_COMMAND=pip install https://github.com/Nuullll/intel-extension-for-pytorch/releases/download/v2.0.110%%2Bxpu-master%%2Bdll-bundle/torch-2.0.0a0+gite9ebda2-cp310-cp310-win_amd64.whl https://github.com/Nuullll/intel-extension-for-pytorch/releases/download/v2.0.110%%2Bxpu-master%%2Bdll-bundle/torchvision-0.15.2a0+fa99a53-cp310-cp310-win_amd64.whl https://github.com/Nuullll/intel-extension-for-pytorch/releases/download/v2.0.110%%2Bxpu-master%%2Bdll-bundle/intel_extension_for_pytorch-2.0.110+gitc6ea20b-cp310-cp310-win_amd64.whl" +set GIT= +set VENV_DIR= +set "COMMANDLINE_ARGS=--use-ipex --skip-torch-cuda-test --skip-version-check --opt-sdp-attention" + +call webui.bat From e294e46d46a814457fc77af13c17128bd6075d45 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sat, 2 Dec 2023 09:26:38 +0300 Subject: [PATCH 173/342] split UI settings page into many --- .../scripts/extra_options_section.py | 13 +++-- modules/shared_options.py | 57 +++++++++++-------- 2 files changed, 40 insertions(+), 30 deletions(-) diff --git a/extensions-builtin/extra-options-section/scripts/extra_options_section.py b/extensions-builtin/extra-options-section/scripts/extra_options_section.py index 983f87ff0..a903df625 100644 --- a/extensions-builtin/extra-options-section/scripts/extra_options_section.py +++ b/extensions-builtin/extra-options-section/scripts/extra_options_section.py @@ -64,11 +64,14 @@ class ExtraOptionsSection(scripts.Script): p.override_settings[name] = value -shared.options_templates.update(shared.options_section(('ui', "User interface"), { - "extra_options_txt2img": shared.OptionInfo([], "Options in main UI - txt2img", ui_components.DropdownMulti, lambda: {"choices": list(shared.opts.data_labels.keys())}).js("info", "settingsHintsShowQuicksettings").info("setting entries that also appear in txt2img interfaces").needs_reload_ui(), - "extra_options_img2img": shared.OptionInfo([], "Options in main UI - img2img", ui_components.DropdownMulti, lambda: {"choices": list(shared.opts.data_labels.keys())}).js("info", "settingsHintsShowQuicksettings").info("setting entries that also appear in img2img interfaces").needs_reload_ui(), - "extra_options_cols": shared.OptionInfo(1, "Options in main UI - number of columns", gr.Number, {"precision": 0}).needs_reload_ui(), - "extra_options_accordion": shared.OptionInfo(False, "Options in main UI - place into an accordion").needs_reload_ui() +shared.options_templates.update(shared.options_section(('settings_in_ui', "Settings in UI", "ui"), { + "settings_in_ui": shared.OptionHTML(""" +This page allows you to add some settings to the main interface of txt2img and img2img tabs. +"""), + "extra_options_txt2img": shared.OptionInfo([], "Settings for txt2img", ui_components.DropdownMulti, lambda: {"choices": list(shared.opts.data_labels.keys())}).js("info", "settingsHintsShowQuicksettings").info("setting entries that also appear in txt2img interfaces").needs_reload_ui(), + "extra_options_img2img": shared.OptionInfo([], "Settings for img2img", ui_components.DropdownMulti, lambda: {"choices": list(shared.opts.data_labels.keys())}).js("info", "settingsHintsShowQuicksettings").info("setting entries that also appear in img2img interfaces").needs_reload_ui(), + "extra_options_cols": shared.OptionInfo(1, "Number of columns for added settings", gr.Number, {"precision": 0}).needs_reload_ui(), + "extra_options_accordion": shared.OptionInfo(False, "Place added settings into an accordion").needs_reload_ui() })) diff --git a/modules/shared_options.py b/modules/shared_options.py index df45fc0a0..1390152db 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -250,38 +250,45 @@ options_templates.update(options_section(('extra_networks', "Extra Networks", "s "sd_hypernetwork": OptionInfo("None", "Add hypernetwork to prompt", gr.Dropdown, lambda: {"choices": ["None", *shared.hypernetworks]}, refresh=shared_items.reload_hypernetworks), })) -options_templates.update(options_section(('ui', "User interface", "ui"), { - "localization": OptionInfo("None", "Localization", gr.Dropdown, lambda: {"choices": ["None"] + list(localization.localizations.keys())}, refresh=lambda: localization.list_localizations(cmd_opts.localizations_dir)).needs_reload_ui(), - "gradio_theme": OptionInfo("Default", "Gradio theme", ui_components.DropdownEditable, lambda: {"choices": ["Default"] + shared_gradio_themes.gradio_hf_hub_themes}).info("you can also manually enter any of themes from the gallery.").needs_reload_ui(), - "gradio_themes_cache": OptionInfo(True, "Cache gradio themes locally").info("disable to update the selected Gradio theme"), - "gallery_height": OptionInfo("", "Gallery height", gr.Textbox).info("an be any valid CSS value").needs_reload_ui(), - "return_grid": OptionInfo(True, "Show grid in results for web"), - "do_not_show_images": OptionInfo(False, "Do not show any images in results for web"), - "send_seed": OptionInfo(True, "Send seed when sending prompt or image to other interface"), - "send_size": OptionInfo(True, "Send size when sending prompt or image to another interface"), - "js_modal_lightbox": OptionInfo(True, "Enable full page image viewer"), - "js_modal_lightbox_initially_zoomed": OptionInfo(True, "Show images zoomed in by default in full page image viewer"), - "js_modal_lightbox_gamepad": OptionInfo(False, "Navigate image viewer with gamepad"), - "js_modal_lightbox_gamepad_repeat": OptionInfo(250, "Gamepad repeat period, in milliseconds"), - "show_progress_in_title": OptionInfo(True, "Show generation progress in window title."), +options_templates.update(options_section(('ui_prompt_editing', "Prompt editing", "ui"), { + "keyedit_precision_attention": OptionInfo(0.1, "Precision for (attention:1.1) when editing the prompt with Ctrl+up/down", gr.Slider, {"minimum": 0.01, "maximum": 0.2, "step": 0.001}), + "keyedit_precision_extra": OptionInfo(0.05, "Precision for when editing the prompt with Ctrl+up/down", gr.Slider, {"minimum": 0.01, "maximum": 0.2, "step": 0.001}), + "keyedit_delimiters": OptionInfo(r".,\/!?%^*;:{}=`~() ", "Word delimiters when editing the prompt with Ctrl+up/down"), + "disable_token_counters": OptionInfo(False, "Disable prompt token counters").needs_reload_ui(), +})) + +options_templates.update(options_section(('ui_gallery', "Gallery", "ui"), { + "return_grid": OptionInfo(True, "Show grid in gallery"), + "do_not_show_images": OptionInfo(False, "Do not show any images in gallery"), + "js_modal_lightbox": OptionInfo(True, "Full page image viewer: enable"), + "js_modal_lightbox_initially_zoomed": OptionInfo(True, "Full page image viewer: show images zoomed in by default"), + "js_modal_lightbox_gamepad": OptionInfo(False, "Full page image viewer: navigate with gamepad"), + "js_modal_lightbox_gamepad_repeat": OptionInfo(250, "Full page image viewer: gamepad repeat period").info("in milliseconds"), + "gallery_height": OptionInfo("", "Gallery height", gr.Textbox).info("can be any valid CSS value, for example 768px or 20em").needs_reload_ui(), +})) + +options_templates.update(options_section(('ui_alternatives', "UI alternatives", "ui"), { + "compact_prompt_box": OptionInfo(False, "Compact prompt layout").info("puts prompt and negative prompt inside the Generate tab, leaving more vertical space for the image on the right").needs_reload_ui(), "samplers_in_dropdown": OptionInfo(True, "Use dropdown for sampler selection instead of radio group").needs_reload_ui(), "dimensions_and_batch_together": OptionInfo(True, "Show Width/Height and Batch sliders in same row").needs_reload_ui(), - "keyedit_precision_attention": OptionInfo(0.1, "Ctrl+up/down precision when editing (attention:1.1)", gr.Slider, {"minimum": 0.01, "maximum": 0.2, "step": 0.001}), - "keyedit_precision_extra": OptionInfo(0.05, "Ctrl+up/down precision when editing ", gr.Slider, {"minimum": 0.01, "maximum": 0.2, "step": 0.001}), - "keyedit_delimiters": OptionInfo(r".,\/!?%^*;:{}=`~() ", "Ctrl+up/down word delimiters"), - "keyedit_delimiters_whitespace": OptionInfo(["Tab", "Carriage Return", "Line Feed"], "Ctrl+up/down whitespace delimiters", gr.CheckboxGroup, lambda: {"choices": ["Tab", "Carriage Return", "Line Feed"]}), - "keyedit_move": OptionInfo(True, "Alt+left/right moves prompt elements"), - "quicksettings_list": OptionInfo(["sd_model_checkpoint"], "Quicksettings list", ui_components.DropdownMulti, lambda: {"choices": list(shared.opts.data_labels.keys())}).js("info", "settingsHintsShowQuicksettings").info("setting entries that appear at the top of page rather than in settings tab").needs_reload_ui(), - "ui_tab_order": OptionInfo([], "UI tab order", ui_components.DropdownMulti, lambda: {"choices": list(shared.tab_names)}).needs_reload_ui(), - "hidden_tabs": OptionInfo([], "Hidden UI tabs", ui_components.DropdownMulti, lambda: {"choices": list(shared.tab_names)}).needs_reload_ui(), - "ui_reorder_list": OptionInfo([], "txt2img/img2img UI item order", ui_components.DropdownMulti, lambda: {"choices": list(shared_items.ui_reorder_categories())}).info("selected items appear first").needs_reload_ui(), "sd_checkpoint_dropdown_use_short": OptionInfo(False, "Checkpoint dropdown: use filenames without paths").info("models in subdirectories like photo/sd15.ckpt will be listed as just sd15.ckpt"), "hires_fix_show_sampler": OptionInfo(False, "Hires fix: show hires checkpoint and sampler selection").needs_reload_ui(), "hires_fix_show_prompts": OptionInfo(False, "Hires fix: show hires prompt and negative prompt").needs_reload_ui(), - "disable_token_counters": OptionInfo(False, "Disable prompt token counters").needs_reload_ui(), "txt2img_settings_accordion": OptionInfo(False, "Settings in txt2img hidden under Accordion").needs_reload_ui(), "img2img_settings_accordion": OptionInfo(False, "Settings in img2img hidden under Accordion").needs_reload_ui(), - "compact_prompt_box": OptionInfo(False, "Compact prompt layout").info("puts prompt and negative prompt inside the Generate tab, leaving more vertical space for the image on the right").needs_reload_ui(), +})) + +options_templates.update(options_section(('ui', "User interface", "ui"), { + "localization": OptionInfo("None", "Localization", gr.Dropdown, lambda: {"choices": ["None"] + list(localization.localizations.keys())}, refresh=lambda: localization.list_localizations(cmd_opts.localizations_dir)).needs_reload_ui(), + "quicksettings_list": OptionInfo(["sd_model_checkpoint"], "Quicksettings list", ui_components.DropdownMulti, lambda: {"choices": list(shared.opts.data_labels.keys())}).js("info", "settingsHintsShowQuicksettings").info("setting entries that appear at the top of page rather than in settings tab").needs_reload_ui(), + "ui_tab_order": OptionInfo([], "UI tab order", ui_components.DropdownMulti, lambda: {"choices": list(shared.tab_names)}).needs_reload_ui(), + "hidden_tabs": OptionInfo([], "Hidden UI tabs", ui_components.DropdownMulti, lambda: {"choices": list(shared.tab_names)}).needs_reload_ui(), + "ui_reorder_list": OptionInfo([], "UI item order for txt2img/img2img tabs", ui_components.DropdownMulti, lambda: {"choices": list(shared_items.ui_reorder_categories())}).info("selected items appear first").needs_reload_ui(), + "gradio_theme": OptionInfo("Default", "Gradio theme", ui_components.DropdownEditable, lambda: {"choices": ["Default"] + shared_gradio_themes.gradio_hf_hub_themes}).info("you can also manually enter any of themes from the gallery.").needs_reload_ui(), + "gradio_themes_cache": OptionInfo(True, "Cache gradio themes locally").info("disable to update the selected Gradio theme"), + "show_progress_in_title": OptionInfo(True, "Show generation progress in window title."), + "send_seed": OptionInfo(True, "Send seed when sending prompt or image to other interface"), + "send_size": OptionInfo(True, "Send size when sending prompt or image to another interface"), })) From ef6b8123dc57e4e4bd5e08d9f3e3dbdfdf6b4c4a Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sat, 2 Dec 2023 09:57:39 +0300 Subject: [PATCH 174/342] put code that can cause an exception into its own function for #14120 --- modules/scripts.py | 62 ++++++++++++++++++++++++---------------------- 1 file changed, 33 insertions(+), 29 deletions(-) diff --git a/modules/scripts.py b/modules/scripts.py index 961d032ce..7f9454eb5 100644 --- a/modules/scripts.py +++ b/modules/scripts.py @@ -560,54 +560,58 @@ class ScriptRunner: on_after.clear() def create_script_ui(self, script): - import modules.api.models as api_models script.args_from = len(self.inputs) script.args_to = len(self.inputs) + try: + self.create_script_ui_inner(script) + except Exception: + errors.report(f"Error creating UI for {script.name}: ", exc_info=True) + + def create_script_ui_inner(self, script): + import modules.api.models as api_models + controls = wrap_call(script.ui, script.filename, "ui", script.is_img2img) if controls is None: return - try: - script.name = wrap_call(script.title, script.filename, "title", default=script.filename).lower() - api_args = [] + script.name = wrap_call(script.title, script.filename, "title", default=script.filename).lower() - for control in controls: - control.custom_script_source = os.path.basename(script.filename) + api_args = [] - arg_info = api_models.ScriptArg(label=control.label or "") + for control in controls: + control.custom_script_source = os.path.basename(script.filename) - for field in ("value", "minimum", "maximum", "step"): - v = getattr(control, field, None) - if v is not None: - setattr(arg_info, field, v) + arg_info = api_models.ScriptArg(label=control.label or "") - choices = getattr(control, 'choices', None) # as of gradio 3.41, some items in choices are strings, and some are tuples where the first elem is the string - if choices is not None: - arg_info.choices = [x[0] if isinstance(x, tuple) else x for x in choices] + for field in ("value", "minimum", "maximum", "step"): + v = getattr(control, field, None) + if v is not None: + setattr(arg_info, field, v) - api_args.append(arg_info) + choices = getattr(control, 'choices', None) # as of gradio 3.41, some items in choices are strings, and some are tuples where the first elem is the string + if choices is not None: + arg_info.choices = [x[0] if isinstance(x, tuple) else x for x in choices] - script.api_info = api_models.ScriptInfo( - name=script.name, - is_img2img=script.is_img2img, - is_alwayson=script.alwayson, - args=api_args, - ) + api_args.append(arg_info) - if script.infotext_fields is not None: - self.infotext_fields += script.infotext_fields + script.api_info = api_models.ScriptInfo( + name=script.name, + is_img2img=script.is_img2img, + is_alwayson=script.alwayson, + args=api_args, + ) - if script.paste_field_names is not None: - self.paste_field_names += script.paste_field_names + if script.infotext_fields is not None: + self.infotext_fields += script.infotext_fields - self.inputs += controls - script.args_to = len(self.inputs) + if script.paste_field_names is not None: + self.paste_field_names += script.paste_field_names - except Exception: - errors.report(f"Error creating UI for {script.name}: ", exc_info=True) + self.inputs += controls + script.args_to = len(self.inputs) def setup_ui_for_section(self, section, scriptlist=None): if scriptlist is None: From 87cd07b3af74c447b02570bf3963ba83ade2e203 Mon Sep 17 00:00:00 2001 From: Nuullll Date: Sat, 2 Dec 2023 15:54:25 +0800 Subject: [PATCH 175/342] Fix fp64 --- modules/sd_samplers_timesteps_impl.py | 4 ++-- modules/xpu_specific.py | 2 +- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/modules/sd_samplers_timesteps_impl.py b/modules/sd_samplers_timesteps_impl.py index a72daafd4..930a64af5 100644 --- a/modules/sd_samplers_timesteps_impl.py +++ b/modules/sd_samplers_timesteps_impl.py @@ -11,7 +11,7 @@ from modules.models.diffusion.uni_pc import uni_pc def ddim(model, x, timesteps, extra_args=None, callback=None, disable=None, eta=0.0): alphas_cumprod = model.inner_model.inner_model.alphas_cumprod alphas = alphas_cumprod[timesteps] - alphas_prev = alphas_cumprod[torch.nn.functional.pad(timesteps[:-1], pad=(1, 0))].to(torch.float64 if x.device.type != 'mps' else torch.float32) + alphas_prev = alphas_cumprod[torch.nn.functional.pad(timesteps[:-1], pad=(1, 0))].to(torch.float64 if x.device.type != 'mps' and x.device.type != 'xpu' else torch.float32) sqrt_one_minus_alphas = torch.sqrt(1 - alphas) sigmas = eta * np.sqrt((1 - alphas_prev.cpu().numpy()) / (1 - alphas.cpu()) * (1 - alphas.cpu() / alphas_prev.cpu().numpy())) @@ -43,7 +43,7 @@ def ddim(model, x, timesteps, extra_args=None, callback=None, disable=None, eta= def plms(model, x, timesteps, extra_args=None, callback=None, disable=None): alphas_cumprod = model.inner_model.inner_model.alphas_cumprod alphas = alphas_cumprod[timesteps] - alphas_prev = alphas_cumprod[torch.nn.functional.pad(timesteps[:-1], pad=(1, 0))].to(torch.float64 if x.device.type != 'mps' else torch.float32) + alphas_prev = alphas_cumprod[torch.nn.functional.pad(timesteps[:-1], pad=(1, 0))].to(torch.float64 if x.device.type != 'mps' and x.device.type != 'xpu' else torch.float32) sqrt_one_minus_alphas = torch.sqrt(1 - alphas) extra_args = {} if extra_args is None else extra_args diff --git a/modules/xpu_specific.py b/modules/xpu_specific.py index 2df68665a..d933c7903 100644 --- a/modules/xpu_specific.py +++ b/modules/xpu_specific.py @@ -4,7 +4,7 @@ from modules.sd_hijack_utils import CondFunc has_ipex = False try: import torch - import intel_extension_for_pytorch as ipex + import intel_extension_for_pytorch as ipex # noqa: F401 has_ipex = True except Exception: pass From 4a666381bf98333ba4512db0f0033df5f6a08771 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sat, 2 Dec 2023 12:11:21 +0300 Subject: [PATCH 176/342] extras tab batch: actually use original filename preprocessing upscale: do not do an extra upscale step if it's not needed --- modules/postprocessing.py | 4 +++- modules/upscaler.py | 6 +++--- 2 files changed, 6 insertions(+), 4 deletions(-) diff --git a/modules/postprocessing.py b/modules/postprocessing.py index fd0c0cc99..0a134ee43 100644 --- a/modules/postprocessing.py +++ b/modules/postprocessing.py @@ -60,8 +60,10 @@ def run_postprocessing(extras_mode, image, image_folder, input_dir, output_dir, if opts.use_original_name_batch and name is not None: basename = os.path.splitext(os.path.basename(name))[0] + forced_filename = basename else: basename = '' + forced_filename = None infotext = ", ".join([k if k == v else f'{k}: {generation_parameters_copypaste.quote(v)}' for k, v in pp.info.items() if v is not None]) @@ -70,7 +72,7 @@ def run_postprocessing(extras_mode, image, image_folder, input_dir, output_dir, pp.image.info["postprocessing"] = infotext if save_output: - images.save_image(pp.image, path=outpath, basename=basename, seed=None, prompt=None, extension=opts.samples_format, info=infotext, short_filename=True, no_prompt=True, grid=False, pnginfo_section_name="extras", existing_info=existing_pnginfo, forced_filename=None) + images.save_image(pp.image, path=outpath, basename=basename, extension=opts.samples_format, info=infotext, short_filename=True, no_prompt=True, grid=False, pnginfo_section_name="extras", existing_info=existing_pnginfo, forced_filename=forced_filename) if extras_mode != 2 or show_extras_results: outputs.append(pp.image) diff --git a/modules/upscaler.py b/modules/upscaler.py index e682bbaa2..b256e085b 100644 --- a/modules/upscaler.py +++ b/modules/upscaler.py @@ -57,6 +57,9 @@ class Upscaler: dest_h = int((img.height * scale) // 8 * 8) for _ in range(3): + if img.width >= dest_w and img.height >= dest_h: + break + shape = (img.width, img.height) img = self.do_upscale(img, selected_model) @@ -64,9 +67,6 @@ class Upscaler: if shape == (img.width, img.height): break - if img.width >= dest_w and img.height >= dest_h: - break - if img.width != dest_w or img.height != dest_h: img = img.resize((int(dest_w), int(dest_h)), resample=LANCZOS) From 96871e4f744471177d97e01c49f8587d7f67c125 Mon Sep 17 00:00:00 2001 From: Nuullll Date: Sat, 2 Dec 2023 17:11:11 +0800 Subject: [PATCH 177/342] Remove webui-ipex-user.bat --- modules/launch_utils.py | 22 ++++++++++++++++++++++ webui-ipex-user.bat | 19 ------------------- 2 files changed, 22 insertions(+), 19 deletions(-) delete mode 100644 webui-ipex-user.bat diff --git a/modules/launch_utils.py b/modules/launch_utils.py index 264ec9ca6..586cdc7eb 100644 --- a/modules/launch_utils.py +++ b/modules/launch_utils.py @@ -310,6 +310,26 @@ def requirements_met(requirements_file): def prepare_environment(): torch_index_url = os.environ.get('TORCH_INDEX_URL', "https://download.pytorch.org/whl/cu118") torch_command = os.environ.get('TORCH_COMMAND', f"pip install torch==2.0.1 torchvision==0.15.2 --extra-index-url {torch_index_url}") + if args.use_ipex: + if platform.system() == "Windows": + # The "Nuullll/intel-extension-for-pytorch" wheels were built from IPEX source for Intel Arc GPU: https://github.com/intel/intel-extension-for-pytorch/tree/xpu-main + # This is NOT an Intel official release so please use it at your own risk!! + # See https://github.com/Nuullll/intel-extension-for-pytorch/releases/tag/v2.0.110%2Bxpu-master%2Bdll-bundle for details. + # + # Strengths (over official IPEX 2.0.110 windows release): + # - AOT build (for Arc GPU only) to eliminate JIT compilation overhead: https://github.com/intel/intel-extension-for-pytorch/issues/399 + # - Bundles minimal oneAPI 2023.2 dependencies into the python wheels, so users don't need to install oneAPI for the whole system. + # - Provides a compatible torchvision wheel: https://github.com/intel/intel-extension-for-pytorch/issues/465 + # Limitation: + # - Only works for python 3.10 + url_prefix = "https://github.com/Nuullll/intel-extension-for-pytorch/releases/download/v2.0.110%2Bxpu-master%2Bdll-bundle" + torch_command = os.environ.get('TORCH_COMMAND', f"pip install {url_prefix}/torch-2.0.0a0+gite9ebda2-cp310-cp310-win_amd64.whl {url_prefix}/torchvision-0.15.2a0+fa99a53-cp310-cp310-win_amd64.whl {url_prefix}/intel_extension_for_pytorch-2.0.110+gitc6ea20b-cp310-cp310-win_amd64.whl") + else: + # Using official IPEX release for linux since it's already an AOT build. + # However, users still have to install oneAPI toolkit and activate oneAPI environment manually. + # See https://intel.github.io/intel-extension-for-pytorch/index.html#installation for details. + torch_index_url = os.environ.get('TORCH_INDEX_URL', "https://pytorch-extension.intel.com/release-whl/stable/xpu/us/") + torch_command = os.environ.get('TORCH_COMMAND', f"pip install torch==2.0.0a0 intel-extension-for-pytorch==2.0.110+gitba7f6c1 --extra-index-url {torch_index_url}") requirements_file = os.environ.get('REQS_FILE', "requirements_versions.txt") xformers_package = os.environ.get('XFORMERS_PACKAGE', 'xformers==0.0.20') @@ -352,6 +372,8 @@ def prepare_environment(): run(f'"{python}" -m {torch_command}', "Installing torch and torchvision", "Couldn't install torch", live=True) startup_timer.record("install torch") + if args.use_ipex: + args.skip_torch_cuda_test = True if not args.skip_torch_cuda_test and not check_run_python("import torch; assert torch.cuda.is_available()"): raise RuntimeError( 'Torch is not able to use GPU; ' diff --git a/webui-ipex-user.bat b/webui-ipex-user.bat deleted file mode 100644 index ab25a0400..000000000 --- a/webui-ipex-user.bat +++ /dev/null @@ -1,19 +0,0 @@ -@echo off - -set PYTHON= -@REM The "Nuullll/intel-extension-for-pytorch" wheels were built from IPEX source for Intel Arc GPU: https://github.com/intel/intel-extension-for-pytorch/tree/xpu-main -@REM This is NOT an Intel official release so please use it at your own risk!! -@REM See https://github.com/Nuullll/intel-extension-for-pytorch/releases/tag/v2.0.110%2Bxpu-master%2Bdll-bundle for details. -@REM -@REM Strengths (over official IPEX 2.0.110 windows release): -@REM - AOT build (for Arc GPU only) to eliminate JIT compilation overhead: https://github.com/intel/intel-extension-for-pytorch/issues/399 -@REM - Bundles minimal oneAPI 2023.2 dependencies into the python wheels, so users don't need to install oneAPI for the whole system. -@REM - Provides a compatible torchvision wheel: https://github.com/intel/intel-extension-for-pytorch/issues/465 -@REM Limitation: -@REM - Only works for python 3.10 -set "TORCH_COMMAND=pip install https://github.com/Nuullll/intel-extension-for-pytorch/releases/download/v2.0.110%%2Bxpu-master%%2Bdll-bundle/torch-2.0.0a0+gite9ebda2-cp310-cp310-win_amd64.whl https://github.com/Nuullll/intel-extension-for-pytorch/releases/download/v2.0.110%%2Bxpu-master%%2Bdll-bundle/torchvision-0.15.2a0+fa99a53-cp310-cp310-win_amd64.whl https://github.com/Nuullll/intel-extension-for-pytorch/releases/download/v2.0.110%%2Bxpu-master%%2Bdll-bundle/intel_extension_for_pytorch-2.0.110+gitc6ea20b-cp310-cp310-win_amd64.whl" -set GIT= -set VENV_DIR= -set "COMMANDLINE_ARGS=--use-ipex --skip-torch-cuda-test --skip-version-check --opt-sdp-attention" - -call webui.bat From 50a21cb09fe3e9ea2d4fe058e0484e192c8a86e3 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Sat, 2 Dec 2023 22:06:47 +0800 Subject: [PATCH 178/342] Ensure the cached weight will not be affected --- modules/sd_models.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/modules/sd_models.py b/modules/sd_models.py index 4b8a9ae65..dcf816b3a 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -435,9 +435,9 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer for module in model.modules(): if isinstance(module, (torch.nn.Conv2d, torch.nn.Linear)): if shared.opts.cache_fp16_weight: - module.fp16_weight = module.weight.clone().half() + module.fp16_weight = module.weight.data.clone().cpu().half() if module.bias is not None: - module.fp16_bias = module.bias.clone().half() + module.fp16_bias = module.bias.data.clone().cpu().half() module.to(torch.float8_e4m3fn) model.first_stage_model = first_stage timer.record("apply fp8") From 11d23e8ca55c097ecfa255a05b63f194e25f08be Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sat, 2 Dec 2023 18:01:11 +0300 Subject: [PATCH 179/342] remove Train/Preprocessing tab and put all its functionality into extras batch images mode --- javascript/ui.js | 17 ++ modules/api/api.py | 15 -- modules/api/models.py | 3 - modules/postprocessing.py | 92 +++++-- modules/scripts_postprocessing.py | 86 ++++++- modules/shared_options.py | 1 + modules/textual_inversion/preprocess.py | 232 ------------------ modules/textual_inversion/ui.py | 7 - modules/ui.py | 107 -------- modules/ui_postprocessing.py | 16 +- modules/ui_toprow.py | 6 +- scripts/postprocessing_caption.py | 30 +++ scripts/postprocessing_codeformer.py | 16 +- .../postprocessing_create_flipped_copies.py | 32 +++ scripts/postprocessing_focal_crop.py | 54 ++++ scripts/postprocessing_gfpgan.py | 13 +- scripts/postprocessing_split_oversized.py | 71 ++++++ scripts/postprocessing_upscale.py | 12 + scripts/processing_autosized_crop.py | 64 +++++ 19 files changed, 460 insertions(+), 414 deletions(-) delete mode 100644 modules/textual_inversion/preprocess.py create mode 100644 scripts/postprocessing_caption.py create mode 100644 scripts/postprocessing_create_flipped_copies.py create mode 100644 scripts/postprocessing_focal_crop.py create mode 100644 scripts/postprocessing_split_oversized.py create mode 100644 scripts/processing_autosized_crop.py diff --git a/javascript/ui.js b/javascript/ui.js index 2e2626020..410fc44e3 100644 --- a/javascript/ui.js +++ b/javascript/ui.js @@ -170,6 +170,23 @@ function submit_img2img() { return res; } +function submit_extras() { + showSubmitButtons('extras', false); + + var id = randomId(); + + requestProgress(id, gradioApp().getElementById('extras_gallery_container'), gradioApp().getElementById('extras_gallery'), function() { + showSubmitButtons('extras', true); + }); + + var res = create_submit_args(arguments); + + res[0] = id; + + console.log(res); + return res; +} + function restoreProgressTxt2img() { showRestoreProgressButton("txt2img", false); var id = localGet("txt2img_task_id"); diff --git a/modules/api/api.py b/modules/api/api.py index 090838747..b3d74e513 100644 --- a/modules/api/api.py +++ b/modules/api/api.py @@ -22,7 +22,6 @@ from modules.api import models from modules.shared import opts from modules.processing import StableDiffusionProcessingTxt2Img, StableDiffusionProcessingImg2Img, process_images from modules.textual_inversion.textual_inversion import create_embedding, train_embedding -from modules.textual_inversion.preprocess import preprocess from modules.hypernetworks.hypernetwork import create_hypernetwork, train_hypernetwork from PIL import PngImagePlugin, Image from modules.sd_models_config import find_checkpoint_config_near_filename @@ -235,7 +234,6 @@ class Api: self.add_api_route("/sdapi/v1/refresh-vae", self.refresh_vae, methods=["POST"]) self.add_api_route("/sdapi/v1/create/embedding", self.create_embedding, methods=["POST"], response_model=models.CreateResponse) self.add_api_route("/sdapi/v1/create/hypernetwork", self.create_hypernetwork, methods=["POST"], response_model=models.CreateResponse) - self.add_api_route("/sdapi/v1/preprocess", self.preprocess, methods=["POST"], response_model=models.PreprocessResponse) self.add_api_route("/sdapi/v1/train/embedding", self.train_embedding, methods=["POST"], response_model=models.TrainResponse) self.add_api_route("/sdapi/v1/train/hypernetwork", self.train_hypernetwork, methods=["POST"], response_model=models.TrainResponse) self.add_api_route("/sdapi/v1/memory", self.get_memory, methods=["GET"], response_model=models.MemoryResponse) @@ -675,19 +673,6 @@ class Api: finally: shared.state.end() - def preprocess(self, args: dict): - try: - shared.state.begin(job="preprocess") - preprocess(**args) # quick operation unless blip/booru interrogation is enabled - shared.state.end() - return models.PreprocessResponse(info='preprocess complete') - except KeyError as e: - return models.PreprocessResponse(info=f"preprocess error: invalid token: {e}") - except Exception as e: - return models.PreprocessResponse(info=f"preprocess error: {e}") - finally: - shared.state.end() - def train_embedding(self, args: dict): try: shared.state.begin(job="train_embedding") diff --git a/modules/api/models.py b/modules/api/models.py index a0d80af8c..33894b3e6 100644 --- a/modules/api/models.py +++ b/modules/api/models.py @@ -202,9 +202,6 @@ class TrainResponse(BaseModel): class CreateResponse(BaseModel): info: str = Field(title="Create info", description="Response string from create embedding or hypernetwork task.") -class PreprocessResponse(BaseModel): - info: str = Field(title="Preprocess info", description="Response string from preprocessing task.") - fields = {} for key, metadata in opts.data_labels.items(): value = opts.data.get(key) diff --git a/modules/postprocessing.py b/modules/postprocessing.py index 0a134ee43..3c85a74c1 100644 --- a/modules/postprocessing.py +++ b/modules/postprocessing.py @@ -6,7 +6,7 @@ from modules import shared, images, devices, scripts, scripts_postprocessing, ui from modules.shared import opts -def run_postprocessing(extras_mode, image, image_folder, input_dir, output_dir, show_extras_results, *args, save_output: bool = True): +def run_postprocessing(id_task, extras_mode, image, image_folder, input_dir, output_dir, show_extras_results, *args, save_output: bool = True): devices.torch_gc() shared.state.begin(job="extras") @@ -29,11 +29,7 @@ def run_postprocessing(extras_mode, image, image_folder, input_dir, output_dir, image_list = shared.listfiles(input_dir) for filename in image_list: - try: - image = Image.open(filename) - except Exception: - continue - yield image, filename + yield filename, filename else: assert image, 'image not selected' yield image, None @@ -45,37 +41,85 @@ def run_postprocessing(extras_mode, image, image_folder, input_dir, output_dir, infotext = '' - for image_data, name in get_images(extras_mode, image, image_folder, input_dir): + data_to_process = list(get_images(extras_mode, image, image_folder, input_dir)) + shared.state.job_count = len(data_to_process) + + for image_placeholder, name in data_to_process: image_data: Image.Image + shared.state.nextjob() shared.state.textinfo = name + shared.state.skipped = False + + if shared.state.interrupted: + break + + if isinstance(image_placeholder, str): + try: + image_data = Image.open(image_placeholder) + except Exception: + continue + else: + image_data = image_placeholder + + shared.state.assign_current_image(image_data) parameters, existing_pnginfo = images.read_info_from_image(image_data) if parameters: existing_pnginfo["parameters"] = parameters - pp = scripts_postprocessing.PostprocessedImage(image_data.convert("RGB")) + initial_pp = scripts_postprocessing.PostprocessedImage(image_data.convert("RGB")) - scripts.scripts_postproc.run(pp, args) + scripts.scripts_postproc.run(initial_pp, args) - if opts.use_original_name_batch and name is not None: - basename = os.path.splitext(os.path.basename(name))[0] - forced_filename = basename - else: - basename = '' - forced_filename = None + if shared.state.skipped: + continue - infotext = ", ".join([k if k == v else f'{k}: {generation_parameters_copypaste.quote(v)}' for k, v in pp.info.items() if v is not None]) + used_suffixes = {} + for pp in [initial_pp, *initial_pp.extra_images]: + suffix = pp.get_suffix(used_suffixes) - if opts.enable_pnginfo: - pp.image.info = existing_pnginfo - pp.image.info["postprocessing"] = infotext + if opts.use_original_name_batch and name is not None: + basename = os.path.splitext(os.path.basename(name))[0] + forced_filename = basename + suffix + else: + basename = '' + forced_filename = None - if save_output: - images.save_image(pp.image, path=outpath, basename=basename, extension=opts.samples_format, info=infotext, short_filename=True, no_prompt=True, grid=False, pnginfo_section_name="extras", existing_info=existing_pnginfo, forced_filename=forced_filename) + infotext = ", ".join([k if k == v else f'{k}: {generation_parameters_copypaste.quote(v)}' for k, v in pp.info.items() if v is not None]) - if extras_mode != 2 or show_extras_results: - outputs.append(pp.image) + if opts.enable_pnginfo: + pp.image.info = existing_pnginfo + pp.image.info["postprocessing"] = infotext + + if save_output: + fullfn, _ = images.save_image(pp.image, path=outpath, basename=basename, extension=opts.samples_format, info=infotext, short_filename=True, no_prompt=True, grid=False, pnginfo_section_name="extras", existing_info=existing_pnginfo, forced_filename=forced_filename, suffix=suffix) + + if pp.caption: + caption_filename = os.path.splitext(fullfn)[0] + ".txt" + if os.path.isfile(caption_filename): + with open(caption_filename, encoding="utf8") as file: + existing_caption = file.read().strip() + else: + existing_caption = "" + + action = shared.opts.postprocessing_existing_caption_action + if action == 'Prepend' and existing_caption: + caption = f"{existing_caption} {pp.caption}" + elif action == 'Append' and existing_caption: + caption = f"{pp.caption} {existing_caption}" + elif action == 'Keep' and existing_caption: + caption = existing_caption + else: + caption = pp.caption + + caption = caption.strip() + if caption: + with open(caption_filename, "w", encoding="utf8") as file: + file.write(caption) + + if extras_mode != 2 or show_extras_results: + outputs.append(pp.image) image_data.close() @@ -99,9 +143,11 @@ def run_extras(extras_mode, resize_mode, image, image_folder, input_dir, output_ "upscaler_2_visibility": extras_upscaler_2_visibility, }, "GFPGAN": { + "enable": True, "gfpgan_visibility": gfpgan_visibility, }, "CodeFormer": { + "enable": True, "codeformer_visibility": codeformer_visibility, "codeformer_weight": codeformer_weight, }, diff --git a/modules/scripts_postprocessing.py b/modules/scripts_postprocessing.py index bac1335dc..901cad080 100644 --- a/modules/scripts_postprocessing.py +++ b/modules/scripts_postprocessing.py @@ -1,13 +1,56 @@ +import dataclasses import os import gradio as gr from modules import errors, shared +@dataclasses.dataclass +class PostprocessedImageSharedInfo: + target_width: int = None + target_height: int = None + + class PostprocessedImage: def __init__(self, image): self.image = image self.info = {} + self.shared = PostprocessedImageSharedInfo() + self.extra_images = [] + self.nametags = [] + self.disable_processing = False + self.caption = None + + def get_suffix(self, used_suffixes=None): + used_suffixes = {} if used_suffixes is None else used_suffixes + suffix = "-".join(self.nametags) + if suffix: + suffix = "-" + suffix + + if suffix not in used_suffixes: + used_suffixes[suffix] = 1 + return suffix + + for i in range(1, 100): + proposed_suffix = suffix + "-" + str(i) + + if proposed_suffix not in used_suffixes: + used_suffixes[proposed_suffix] = 1 + return proposed_suffix + + return suffix + + def create_copy(self, new_image, *, nametags=None, disable_processing=False): + pp = PostprocessedImage(new_image) + pp.shared = self.shared + pp.nametags = self.nametags.copy() + pp.info = self.info.copy() + pp.disable_processing = disable_processing + + if nametags is not None: + pp.nametags += nametags + + return pp class ScriptPostprocessing: @@ -42,10 +85,17 @@ class ScriptPostprocessing: pass - def image_changed(self): + def process_firstpass(self, pp: PostprocessedImage, **args): + """ + Called for all scripts before calling process(). Scripts can examine the image here and set fields + of the pp object to communicate things to other scripts. + args contains a dictionary with all values returned by components from ui() + """ + pass - + def image_changed(self): + pass def wrap_call(func, filename, funcname, *args, default=None, **kwargs): @@ -118,16 +168,42 @@ class ScriptPostprocessingRunner: return inputs def run(self, pp: PostprocessedImage, args): - for script in self.scripts_in_preferred_order(): - shared.state.job = script.name + scripts = [] + for script in self.scripts_in_preferred_order(): script_args = args[script.args_from:script.args_to] process_args = {} for (name, _component), value in zip(script.controls.items(), script_args): process_args[name] = value - script.process(pp, **process_args) + scripts.append((script, process_args)) + + for script, process_args in scripts: + script.process_firstpass(pp, **process_args) + + all_images = [pp] + + for script, process_args in scripts: + if shared.state.skipped: + break + + shared.state.job = script.name + + for single_image in all_images.copy(): + + if not single_image.disable_processing: + script.process(single_image, **process_args) + + for extra_image in single_image.extra_images: + if not isinstance(extra_image, PostprocessedImage): + extra_image = single_image.create_copy(extra_image) + + all_images.append(extra_image) + + single_image.extra_images.clear() + + pp.extra_images = all_images[1:] def create_args_for_run(self, scripts_args): if not self.ui_created: diff --git a/modules/shared_options.py b/modules/shared_options.py index d8a27180e..859dee404 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -357,6 +357,7 @@ options_templates.update(options_section(('postprocessing', "Postprocessing", "p 'postprocessing_enable_in_main_ui': OptionInfo([], "Enable postprocessing operations in txt2img and img2img tabs", ui_components.DropdownMulti, lambda: {"choices": [x.name for x in shared_items.postprocessing_scripts()]}), 'postprocessing_operation_order': OptionInfo([], "Postprocessing operation order", ui_components.DropdownMulti, lambda: {"choices": [x.name for x in shared_items.postprocessing_scripts()]}), 'upscaling_max_images_in_cache': OptionInfo(5, "Maximum number of images in upscaling cache", gr.Slider, {"minimum": 0, "maximum": 10, "step": 1}), + 'postprocessing_existing_caption_action': OptionInfo("Ignore", "Action for existing captions", gr.Radio, {"choices": ["Ignore", "Keep", "Prepend", "Append"]}).info("when generating captions using postprocessing; Ignore = use generated; Keep = use original; Prepend/Append = combine both"), })) options_templates.update(options_section((None, "Hidden options"), { diff --git a/modules/textual_inversion/preprocess.py b/modules/textual_inversion/preprocess.py deleted file mode 100644 index 789fa0838..000000000 --- a/modules/textual_inversion/preprocess.py +++ /dev/null @@ -1,232 +0,0 @@ -import os -from PIL import Image, ImageOps -import math -import tqdm - -from modules import shared, images, deepbooru -from modules.textual_inversion import autocrop - - -def preprocess(id_task, process_src, process_dst, process_width, process_height, preprocess_txt_action, process_keep_original_size, process_flip, process_split, process_caption, process_caption_deepbooru=False, split_threshold=0.5, overlap_ratio=0.2, process_focal_crop=False, process_focal_crop_face_weight=0.9, process_focal_crop_entropy_weight=0.15, process_focal_crop_edges_weight=0.5, process_focal_crop_debug=False, process_multicrop=None, process_multicrop_mindim=None, process_multicrop_maxdim=None, process_multicrop_minarea=None, process_multicrop_maxarea=None, process_multicrop_objective=None, process_multicrop_threshold=None): - try: - if process_caption: - shared.interrogator.load() - - if process_caption_deepbooru: - deepbooru.model.start() - - preprocess_work(process_src, process_dst, process_width, process_height, preprocess_txt_action, process_keep_original_size, process_flip, process_split, process_caption, process_caption_deepbooru, split_threshold, overlap_ratio, process_focal_crop, process_focal_crop_face_weight, process_focal_crop_entropy_weight, process_focal_crop_edges_weight, process_focal_crop_debug, process_multicrop, process_multicrop_mindim, process_multicrop_maxdim, process_multicrop_minarea, process_multicrop_maxarea, process_multicrop_objective, process_multicrop_threshold) - - finally: - - if process_caption: - shared.interrogator.send_blip_to_ram() - - if process_caption_deepbooru: - deepbooru.model.stop() - - -def listfiles(dirname): - return os.listdir(dirname) - - -class PreprocessParams: - src = None - dstdir = None - subindex = 0 - flip = False - process_caption = False - process_caption_deepbooru = False - preprocess_txt_action = None - - -def save_pic_with_caption(image, index, params: PreprocessParams, existing_caption=None): - caption = "" - - if params.process_caption: - caption += shared.interrogator.generate_caption(image) - - if params.process_caption_deepbooru: - if caption: - caption += ", " - caption += deepbooru.model.tag_multi(image) - - filename_part = params.src - filename_part = os.path.splitext(filename_part)[0] - filename_part = os.path.basename(filename_part) - - basename = f"{index:05}-{params.subindex}-{filename_part}" - image.save(os.path.join(params.dstdir, f"{basename}.png")) - - if params.preprocess_txt_action == 'prepend' and existing_caption: - caption = f"{existing_caption} {caption}" - elif params.preprocess_txt_action == 'append' and existing_caption: - caption = f"{caption} {existing_caption}" - elif params.preprocess_txt_action == 'copy' and existing_caption: - caption = existing_caption - - caption = caption.strip() - - if caption: - with open(os.path.join(params.dstdir, f"{basename}.txt"), "w", encoding="utf8") as file: - file.write(caption) - - params.subindex += 1 - - -def save_pic(image, index, params, existing_caption=None): - save_pic_with_caption(image, index, params, existing_caption=existing_caption) - - if params.flip: - save_pic_with_caption(ImageOps.mirror(image), index, params, existing_caption=existing_caption) - - -def split_pic(image, inverse_xy, width, height, overlap_ratio): - if inverse_xy: - from_w, from_h = image.height, image.width - to_w, to_h = height, width - else: - from_w, from_h = image.width, image.height - to_w, to_h = width, height - h = from_h * to_w // from_w - if inverse_xy: - image = image.resize((h, to_w)) - else: - image = image.resize((to_w, h)) - - split_count = math.ceil((h - to_h * overlap_ratio) / (to_h * (1.0 - overlap_ratio))) - y_step = (h - to_h) / (split_count - 1) - for i in range(split_count): - y = int(y_step * i) - if inverse_xy: - splitted = image.crop((y, 0, y + to_h, to_w)) - else: - splitted = image.crop((0, y, to_w, y + to_h)) - yield splitted - -# not using torchvision.transforms.CenterCrop because it doesn't allow float regions -def center_crop(image: Image, w: int, h: int): - iw, ih = image.size - if ih / h < iw / w: - sw = w * ih / h - box = (iw - sw) / 2, 0, iw - (iw - sw) / 2, ih - else: - sh = h * iw / w - box = 0, (ih - sh) / 2, iw, ih - (ih - sh) / 2 - return image.resize((w, h), Image.Resampling.LANCZOS, box) - - -def multicrop_pic(image: Image, mindim, maxdim, minarea, maxarea, objective, threshold): - iw, ih = image.size - err = lambda w, h: 1-(lambda x: x if x < 1 else 1/x)(iw/ih/(w/h)) - wh = max(((w, h) for w in range(mindim, maxdim+1, 64) for h in range(mindim, maxdim+1, 64) - if minarea <= w * h <= maxarea and err(w, h) <= threshold), - key= lambda wh: (wh[0]*wh[1], -err(*wh))[::1 if objective=='Maximize area' else -1], - default=None - ) - return wh and center_crop(image, *wh) - - -def preprocess_work(process_src, process_dst, process_width, process_height, preprocess_txt_action, process_keep_original_size, process_flip, process_split, process_caption, process_caption_deepbooru=False, split_threshold=0.5, overlap_ratio=0.2, process_focal_crop=False, process_focal_crop_face_weight=0.9, process_focal_crop_entropy_weight=0.3, process_focal_crop_edges_weight=0.5, process_focal_crop_debug=False, process_multicrop=None, process_multicrop_mindim=None, process_multicrop_maxdim=None, process_multicrop_minarea=None, process_multicrop_maxarea=None, process_multicrop_objective=None, process_multicrop_threshold=None): - width = process_width - height = process_height - src = os.path.abspath(process_src) - dst = os.path.abspath(process_dst) - split_threshold = max(0.0, min(1.0, split_threshold)) - overlap_ratio = max(0.0, min(0.9, overlap_ratio)) - - assert src != dst, 'same directory specified as source and destination' - - os.makedirs(dst, exist_ok=True) - - files = listfiles(src) - - shared.state.job = "preprocess" - shared.state.textinfo = "Preprocessing..." - shared.state.job_count = len(files) - - params = PreprocessParams() - params.dstdir = dst - params.flip = process_flip - params.process_caption = process_caption - params.process_caption_deepbooru = process_caption_deepbooru - params.preprocess_txt_action = preprocess_txt_action - - pbar = tqdm.tqdm(files) - for index, imagefile in enumerate(pbar): - params.subindex = 0 - filename = os.path.join(src, imagefile) - try: - img = Image.open(filename) - img = ImageOps.exif_transpose(img) - img = img.convert("RGB") - except Exception: - continue - - description = f"Preprocessing [Image {index}/{len(files)}]" - pbar.set_description(description) - shared.state.textinfo = description - - params.src = filename - - existing_caption = None - existing_caption_filename = f"{os.path.splitext(filename)[0]}.txt" - if os.path.exists(existing_caption_filename): - with open(existing_caption_filename, 'r', encoding="utf8") as file: - existing_caption = file.read() - - if shared.state.interrupted: - break - - if img.height > img.width: - ratio = (img.width * height) / (img.height * width) - inverse_xy = False - else: - ratio = (img.height * width) / (img.width * height) - inverse_xy = True - - process_default_resize = True - - if process_split and ratio < 1.0 and ratio <= split_threshold: - for splitted in split_pic(img, inverse_xy, width, height, overlap_ratio): - save_pic(splitted, index, params, existing_caption=existing_caption) - process_default_resize = False - - if process_focal_crop and img.height != img.width: - - dnn_model_path = None - try: - dnn_model_path = autocrop.download_and_cache_models() - except Exception as e: - print("Unable to load face detection model for auto crop selection. Falling back to lower quality haar method.", e) - - autocrop_settings = autocrop.Settings( - crop_width = width, - crop_height = height, - face_points_weight = process_focal_crop_face_weight, - entropy_points_weight = process_focal_crop_entropy_weight, - corner_points_weight = process_focal_crop_edges_weight, - annotate_image = process_focal_crop_debug, - dnn_model_path = dnn_model_path, - ) - for focal in autocrop.crop_image(img, autocrop_settings): - save_pic(focal, index, params, existing_caption=existing_caption) - process_default_resize = False - - if process_multicrop: - cropped = multicrop_pic(img, process_multicrop_mindim, process_multicrop_maxdim, process_multicrop_minarea, process_multicrop_maxarea, process_multicrop_objective, process_multicrop_threshold) - if cropped is not None: - save_pic(cropped, index, params, existing_caption=existing_caption) - else: - print(f"skipped {img.width}x{img.height} image {filename} (can't find suitable size within error threshold)") - process_default_resize = False - - if process_keep_original_size: - save_pic(img, index, params, existing_caption=existing_caption) - process_default_resize = False - - if process_default_resize: - img = images.resize_image(1, img, width, height) - save_pic(img, index, params, existing_caption=existing_caption) - - shared.state.nextjob() diff --git a/modules/textual_inversion/ui.py b/modules/textual_inversion/ui.py index 35c4feeff..f149ad1f0 100644 --- a/modules/textual_inversion/ui.py +++ b/modules/textual_inversion/ui.py @@ -3,7 +3,6 @@ import html import gradio as gr import modules.textual_inversion.textual_inversion -import modules.textual_inversion.preprocess from modules import sd_hijack, shared @@ -15,12 +14,6 @@ def create_embedding(name, initialization_text, nvpt, overwrite_old): return gr.Dropdown.update(choices=sorted(sd_hijack.model_hijack.embedding_db.word_embeddings.keys())), f"Created: {filename}", "" -def preprocess(*args): - modules.textual_inversion.preprocess.preprocess(*args) - - return f"Preprocessing {'interrupted' if shared.state.interrupted else 'finished'}.", "" - - def train_embedding(*args): assert not shared.cmd_opts.lowvram, 'Training models with lowvram not possible' diff --git a/modules/ui.py b/modules/ui.py index 08e0ad775..d80486dd4 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -912,71 +912,6 @@ def create_ui(): with gr.Column(): create_hypernetwork = gr.Button(value="Create hypernetwork", variant='primary', elem_id="train_create_hypernetwork") - with gr.Tab(label="Preprocess images", id="preprocess_images"): - process_src = gr.Textbox(label='Source directory', elem_id="train_process_src") - process_dst = gr.Textbox(label='Destination directory', elem_id="train_process_dst") - process_width = gr.Slider(minimum=64, maximum=2048, step=8, label="Width", value=512, elem_id="train_process_width") - process_height = gr.Slider(minimum=64, maximum=2048, step=8, label="Height", value=512, elem_id="train_process_height") - preprocess_txt_action = gr.Dropdown(label='Existing Caption txt Action', value="ignore", choices=["ignore", "copy", "prepend", "append"], elem_id="train_preprocess_txt_action") - - with gr.Row(): - process_keep_original_size = gr.Checkbox(label='Keep original size', elem_id="train_process_keep_original_size") - process_flip = gr.Checkbox(label='Create flipped copies', elem_id="train_process_flip") - process_split = gr.Checkbox(label='Split oversized images', elem_id="train_process_split") - process_focal_crop = gr.Checkbox(label='Auto focal point crop', elem_id="train_process_focal_crop") - process_multicrop = gr.Checkbox(label='Auto-sized crop', elem_id="train_process_multicrop") - process_caption = gr.Checkbox(label='Use BLIP for caption', elem_id="train_process_caption") - process_caption_deepbooru = gr.Checkbox(label='Use deepbooru for caption', visible=True, elem_id="train_process_caption_deepbooru") - - with gr.Row(visible=False) as process_split_extra_row: - process_split_threshold = gr.Slider(label='Split image threshold', value=0.5, minimum=0.0, maximum=1.0, step=0.05, elem_id="train_process_split_threshold") - process_overlap_ratio = gr.Slider(label='Split image overlap ratio', value=0.2, minimum=0.0, maximum=0.9, step=0.05, elem_id="train_process_overlap_ratio") - - with gr.Row(visible=False) as process_focal_crop_row: - process_focal_crop_face_weight = gr.Slider(label='Focal point face weight', value=0.9, minimum=0.0, maximum=1.0, step=0.05, elem_id="train_process_focal_crop_face_weight") - process_focal_crop_entropy_weight = gr.Slider(label='Focal point entropy weight', value=0.15, minimum=0.0, maximum=1.0, step=0.05, elem_id="train_process_focal_crop_entropy_weight") - process_focal_crop_edges_weight = gr.Slider(label='Focal point edges weight', value=0.5, minimum=0.0, maximum=1.0, step=0.05, elem_id="train_process_focal_crop_edges_weight") - process_focal_crop_debug = gr.Checkbox(label='Create debug image', elem_id="train_process_focal_crop_debug") - - with gr.Column(visible=False) as process_multicrop_col: - gr.Markdown('Each image is center-cropped with an automatically chosen width and height.') - with gr.Row(): - process_multicrop_mindim = gr.Slider(minimum=64, maximum=2048, step=8, label="Dimension lower bound", value=384, elem_id="train_process_multicrop_mindim") - process_multicrop_maxdim = gr.Slider(minimum=64, maximum=2048, step=8, label="Dimension upper bound", value=768, elem_id="train_process_multicrop_maxdim") - with gr.Row(): - process_multicrop_minarea = gr.Slider(minimum=64*64, maximum=2048*2048, step=1, label="Area lower bound", value=64*64, elem_id="train_process_multicrop_minarea") - process_multicrop_maxarea = gr.Slider(minimum=64*64, maximum=2048*2048, step=1, label="Area upper bound", value=640*640, elem_id="train_process_multicrop_maxarea") - with gr.Row(): - process_multicrop_objective = gr.Radio(["Maximize area", "Minimize error"], value="Maximize area", label="Resizing objective", elem_id="train_process_multicrop_objective") - process_multicrop_threshold = gr.Slider(minimum=0, maximum=1, step=0.01, label="Error threshold", value=0.1, elem_id="train_process_multicrop_threshold") - - with gr.Row(): - with gr.Column(scale=3): - gr.HTML(value="") - - with gr.Column(): - with gr.Row(): - interrupt_preprocessing = gr.Button("Interrupt", elem_id="train_interrupt_preprocessing") - run_preprocess = gr.Button(value="Preprocess", variant='primary', elem_id="train_run_preprocess") - - process_split.change( - fn=lambda show: gr_show(show), - inputs=[process_split], - outputs=[process_split_extra_row], - ) - - process_focal_crop.change( - fn=lambda show: gr_show(show), - inputs=[process_focal_crop], - outputs=[process_focal_crop_row], - ) - - process_multicrop.change( - fn=lambda show: gr_show(show), - inputs=[process_multicrop], - outputs=[process_multicrop_col], - ) - def get_textual_inversion_template_names(): return sorted(textual_inversion.textual_inversion_templates) @@ -1077,42 +1012,6 @@ def create_ui(): ] ) - run_preprocess.click( - fn=wrap_gradio_gpu_call(textual_inversion_ui.preprocess, extra_outputs=[gr.update()]), - _js="start_training_textual_inversion", - inputs=[ - dummy_component, - process_src, - process_dst, - process_width, - process_height, - preprocess_txt_action, - process_keep_original_size, - process_flip, - process_split, - process_caption, - process_caption_deepbooru, - process_split_threshold, - process_overlap_ratio, - process_focal_crop, - process_focal_crop_face_weight, - process_focal_crop_entropy_weight, - process_focal_crop_edges_weight, - process_focal_crop_debug, - process_multicrop, - process_multicrop_mindim, - process_multicrop_maxdim, - process_multicrop_minarea, - process_multicrop_maxarea, - process_multicrop_objective, - process_multicrop_threshold, - ], - outputs=[ - ti_output, - ti_outcome, - ], - ) - train_embedding.click( fn=wrap_gradio_gpu_call(textual_inversion_ui.train_embedding, extra_outputs=[gr.update()]), _js="start_training_textual_inversion", @@ -1186,12 +1085,6 @@ def create_ui(): outputs=[], ) - interrupt_preprocessing.click( - fn=lambda: shared.state.interrupt(), - inputs=[], - outputs=[], - ) - loadsave = ui_loadsave.UiLoadsave(cmd_opts.ui_config_file) settings = ui_settings.UiSettings() diff --git a/modules/ui_postprocessing.py b/modules/ui_postprocessing.py index 802e1ce71..fbad0800a 100644 --- a/modules/ui_postprocessing.py +++ b/modules/ui_postprocessing.py @@ -1,9 +1,10 @@ import gradio as gr -from modules import scripts, shared, ui_common, postprocessing, call_queue +from modules import scripts, shared, ui_common, postprocessing, call_queue, ui_toprow import modules.generation_parameters_copypaste as parameters_copypaste def create_ui(): + dummy_component = gr.Label(visible=False) tab_index = gr.State(value=0) with gr.Row(equal_height=False, variant='compact'): @@ -20,11 +21,13 @@ def create_ui(): extras_batch_output_dir = gr.Textbox(label="Output directory", **shared.hide_dirs, placeholder="Leave blank to save images to the default path.", elem_id="extras_batch_output_dir") show_extras_results = gr.Checkbox(label='Show result images', value=True, elem_id="extras_show_extras_results") - submit = gr.Button('Generate', elem_id="extras_generate", variant='primary') - script_inputs = scripts.scripts_postproc.setup_ui() with gr.Column(): + toprow = ui_toprow.Toprow(is_compact=True, is_img2img=False, id_part="extras") + toprow.create_inline_toprow_image() + submit = toprow.submit + result_images, html_info_x, html_info, html_log = ui_common.create_output_panel("extras", shared.opts.outdir_extras_samples) tab_single.select(fn=lambda: 0, inputs=[], outputs=[tab_index]) @@ -33,7 +36,9 @@ def create_ui(): submit.click( fn=call_queue.wrap_gradio_gpu_call(postprocessing.run_postprocessing, extra_outputs=[None, '']), + _js="submit_extras", inputs=[ + dummy_component, tab_index, extras_image, image_batch, @@ -45,8 +50,9 @@ def create_ui(): outputs=[ result_images, html_info_x, - html_info, - ] + html_log, + ], + show_progress=False, ) parameters_copypaste.add_paste_fields("extras", extras_image, None) diff --git a/modules/ui_toprow.py b/modules/ui_toprow.py index 985b5a2dd..88838f977 100644 --- a/modules/ui_toprow.py +++ b/modules/ui_toprow.py @@ -34,8 +34,10 @@ class Toprow: submit_box = None - def __init__(self, is_img2img, is_compact=False): - id_part = "img2img" if is_img2img else "txt2img" + def __init__(self, is_img2img, is_compact=False, id_part=None): + if id_part is None: + id_part = "img2img" if is_img2img else "txt2img" + self.id_part = id_part self.is_img2img = is_img2img self.is_compact = is_compact diff --git a/scripts/postprocessing_caption.py b/scripts/postprocessing_caption.py new file mode 100644 index 000000000..243e3ad9c --- /dev/null +++ b/scripts/postprocessing_caption.py @@ -0,0 +1,30 @@ +from modules import scripts_postprocessing, ui_components, deepbooru, shared +import gradio as gr + + +class ScriptPostprocessingCeption(scripts_postprocessing.ScriptPostprocessing): + name = "Caption" + order = 4000 + + def ui(self): + with ui_components.InputAccordion(False, label="Caption") as enable: + option = gr.CheckboxGroup(value=["Deepbooru"], choices=["Deepbooru", "BLIP"], show_label=False) + + return { + "enable": enable, + "option": option, + } + + def process(self, pp: scripts_postprocessing.PostprocessedImage, enable, option): + if not enable: + return + + captions = [pp.caption] + + if "Deepbooru" in option: + captions.append(deepbooru.model.tag(pp.image)) + + if "BLIP" in option: + captions.append(shared.interrogator.generate_caption(pp.image)) + + pp.caption = ", ".join([x for x in captions if x]) diff --git a/scripts/postprocessing_codeformer.py b/scripts/postprocessing_codeformer.py index a7d80d40e..e1e156ddc 100644 --- a/scripts/postprocessing_codeformer.py +++ b/scripts/postprocessing_codeformer.py @@ -1,28 +1,28 @@ from PIL import Image import numpy as np -from modules import scripts_postprocessing, codeformer_model +from modules import scripts_postprocessing, codeformer_model, ui_components import gradio as gr -from modules.ui_components import FormRow - class ScriptPostprocessingCodeFormer(scripts_postprocessing.ScriptPostprocessing): name = "CodeFormer" order = 3000 def ui(self): - with FormRow(): - codeformer_visibility = gr.Slider(minimum=0.0, maximum=1.0, step=0.001, label="CodeFormer visibility", value=0, elem_id="extras_codeformer_visibility") - codeformer_weight = gr.Slider(minimum=0.0, maximum=1.0, step=0.001, label="CodeFormer weight (0 = maximum effect, 1 = minimum effect)", value=0, elem_id="extras_codeformer_weight") + with ui_components.InputAccordion(False, label="CodeFormer") as enable: + with gr.Row(): + codeformer_visibility = gr.Slider(minimum=0.0, maximum=1.0, step=0.001, label="Visibility", value=1.0, elem_id="extras_codeformer_visibility") + codeformer_weight = gr.Slider(minimum=0.0, maximum=1.0, step=0.001, label="Weight (0 = maximum effect, 1 = minimum effect)", value=0, elem_id="extras_codeformer_weight") return { + "enable": enable, "codeformer_visibility": codeformer_visibility, "codeformer_weight": codeformer_weight, } - def process(self, pp: scripts_postprocessing.PostprocessedImage, codeformer_visibility, codeformer_weight): - if codeformer_visibility == 0: + def process(self, pp: scripts_postprocessing.PostprocessedImage, enable, codeformer_visibility, codeformer_weight): + if codeformer_visibility == 0 or not enable: return restored_img = codeformer_model.codeformer.restore(np.array(pp.image, dtype=np.uint8), w=codeformer_weight) diff --git a/scripts/postprocessing_create_flipped_copies.py b/scripts/postprocessing_create_flipped_copies.py new file mode 100644 index 000000000..3425571dc --- /dev/null +++ b/scripts/postprocessing_create_flipped_copies.py @@ -0,0 +1,32 @@ +from PIL import ImageOps, Image + +from modules import scripts_postprocessing, ui_components +import gradio as gr + + +class ScriptPostprocessingCreateFlippedCopies(scripts_postprocessing.ScriptPostprocessing): + name = "Create flipped copies" + order = 4000 + + def ui(self): + with ui_components.InputAccordion(False, label="Create flipped copies") as enable: + with gr.Row(): + option = gr.CheckboxGroup(value=["Horizontal"], choices=["Horizontal", "Vertical", "Both"], show_label=False) + + return { + "enable": enable, + "option": option, + } + + def process(self, pp: scripts_postprocessing.PostprocessedImage, enable, option): + if not enable: + return + + if "Horizontal" in option: + pp.extra_images.append(ImageOps.mirror(pp.image)) + + if "Vertical" in option: + pp.extra_images.append(pp.image.transpose(Image.Transpose.FLIP_TOP_BOTTOM)) + + if "Both" in option: + pp.extra_images.append(pp.image.transpose(Image.Transpose.FLIP_TOP_BOTTOM).transpose(Image.Transpose.FLIP_LEFT_RIGHT)) diff --git a/scripts/postprocessing_focal_crop.py b/scripts/postprocessing_focal_crop.py new file mode 100644 index 000000000..d3baf2987 --- /dev/null +++ b/scripts/postprocessing_focal_crop.py @@ -0,0 +1,54 @@ + +from modules import scripts_postprocessing, ui_components, errors +import gradio as gr + +from modules.textual_inversion import autocrop + + +class ScriptPostprocessingFocalCrop(scripts_postprocessing.ScriptPostprocessing): + name = "Auto focal point crop" + order = 4000 + + def ui(self): + with ui_components.InputAccordion(False, label="Auto focal point crop") as enable: + face_weight = gr.Slider(label='Focal point face weight', value=0.9, minimum=0.0, maximum=1.0, step=0.05, elem_id="postprocess_focal_crop_face_weight") + entropy_weight = gr.Slider(label='Focal point entropy weight', value=0.15, minimum=0.0, maximum=1.0, step=0.05, elem_id="postprocess_focal_crop_entropy_weight") + edges_weight = gr.Slider(label='Focal point edges weight', value=0.5, minimum=0.0, maximum=1.0, step=0.05, elem_id="postprocess_focal_crop_edges_weight") + debug = gr.Checkbox(label='Create debug image', elem_id="train_process_focal_crop_debug") + + return { + "enable": enable, + "face_weight": face_weight, + "entropy_weight": entropy_weight, + "edges_weight": edges_weight, + "debug": debug, + } + + def process(self, pp: scripts_postprocessing.PostprocessedImage, enable, face_weight, entropy_weight, edges_weight, debug): + if not enable: + return + + if not pp.shared.target_width or not pp.shared.target_height: + return + + dnn_model_path = None + try: + dnn_model_path = autocrop.download_and_cache_models() + except Exception: + errors.report("Unable to load face detection model for auto crop selection. Falling back to lower quality haar method.", exc_info=True) + + autocrop_settings = autocrop.Settings( + crop_width=pp.shared.target_width, + crop_height=pp.shared.target_height, + face_points_weight=face_weight, + entropy_points_weight=entropy_weight, + corner_points_weight=edges_weight, + annotate_image=debug, + dnn_model_path=dnn_model_path, + ) + + result, *others = autocrop.crop_image(pp.image, autocrop_settings) + + pp.image = result + pp.extra_images = [pp.create_copy(x, nametags=["focal-crop-debug"], disable_processing=True) for x in others] + diff --git a/scripts/postprocessing_gfpgan.py b/scripts/postprocessing_gfpgan.py index d854f3f77..6e7566055 100644 --- a/scripts/postprocessing_gfpgan.py +++ b/scripts/postprocessing_gfpgan.py @@ -1,26 +1,25 @@ from PIL import Image import numpy as np -from modules import scripts_postprocessing, gfpgan_model +from modules import scripts_postprocessing, gfpgan_model, ui_components import gradio as gr -from modules.ui_components import FormRow - class ScriptPostprocessingGfpGan(scripts_postprocessing.ScriptPostprocessing): name = "GFPGAN" order = 2000 def ui(self): - with FormRow(): - gfpgan_visibility = gr.Slider(minimum=0.0, maximum=1.0, step=0.001, label="GFPGAN visibility", value=0, elem_id="extras_gfpgan_visibility") + with ui_components.InputAccordion(False, label="GFPGAN") as enable: + gfpgan_visibility = gr.Slider(minimum=0.0, maximum=1.0, step=0.001, label="Visibility", value=1.0, elem_id="extras_gfpgan_visibility") return { + "enable": enable, "gfpgan_visibility": gfpgan_visibility, } - def process(self, pp: scripts_postprocessing.PostprocessedImage, gfpgan_visibility): - if gfpgan_visibility == 0: + def process(self, pp: scripts_postprocessing.PostprocessedImage, enable, gfpgan_visibility): + if gfpgan_visibility == 0 or not enable: return restored_img = gfpgan_model.gfpgan_fix_faces(np.array(pp.image, dtype=np.uint8)) diff --git a/scripts/postprocessing_split_oversized.py b/scripts/postprocessing_split_oversized.py new file mode 100644 index 000000000..c4a03160f --- /dev/null +++ b/scripts/postprocessing_split_oversized.py @@ -0,0 +1,71 @@ +import math + +from modules import scripts_postprocessing, ui_components +import gradio as gr + + +def split_pic(image, inverse_xy, width, height, overlap_ratio): + if inverse_xy: + from_w, from_h = image.height, image.width + to_w, to_h = height, width + else: + from_w, from_h = image.width, image.height + to_w, to_h = width, height + h = from_h * to_w // from_w + if inverse_xy: + image = image.resize((h, to_w)) + else: + image = image.resize((to_w, h)) + + split_count = math.ceil((h - to_h * overlap_ratio) / (to_h * (1.0 - overlap_ratio))) + y_step = (h - to_h) / (split_count - 1) + for i in range(split_count): + y = int(y_step * i) + if inverse_xy: + splitted = image.crop((y, 0, y + to_h, to_w)) + else: + splitted = image.crop((0, y, to_w, y + to_h)) + yield splitted + + +class ScriptPostprocessingSplitOversized(scripts_postprocessing.ScriptPostprocessing): + name = "Split oversized images" + order = 4000 + + def ui(self): + with ui_components.InputAccordion(False, label="Split oversized images") as enable: + with gr.Row(): + split_threshold = gr.Slider(label='Threshold', value=0.5, minimum=0.0, maximum=1.0, step=0.05, elem_id="postprocess_split_threshold") + overlap_ratio = gr.Slider(label='Overlap ratio', value=0.2, minimum=0.0, maximum=0.9, step=0.05, elem_id="postprocess_overlap_ratio") + + return { + "enable": enable, + "split_threshold": split_threshold, + "overlap_ratio": overlap_ratio, + } + + def process(self, pp: scripts_postprocessing.PostprocessedImage, enable, split_threshold, overlap_ratio): + if not enable: + return + + width = pp.shared.target_width + height = pp.shared.target_height + + if not width or not height: + return + + if pp.image.height > pp.image.width: + ratio = (pp.image.width * height) / (pp.image.height * width) + inverse_xy = False + else: + ratio = (pp.image.height * width) / (pp.image.width * height) + inverse_xy = True + + if ratio >= 1.0 and ratio > split_threshold: + return + + result, *others = split_pic(pp.image, inverse_xy, width, height, overlap_ratio) + + pp.image = result + pp.extra_images = [pp.create_copy(x) for x in others] + diff --git a/scripts/postprocessing_upscale.py b/scripts/postprocessing_upscale.py index eb42a29e5..ed709688d 100644 --- a/scripts/postprocessing_upscale.py +++ b/scripts/postprocessing_upscale.py @@ -81,6 +81,14 @@ class ScriptPostprocessingUpscale(scripts_postprocessing.ScriptPostprocessing): return image + def process_firstpass(self, pp: scripts_postprocessing.PostprocessedImage, upscale_mode=1, upscale_by=2.0, upscale_to_width=None, upscale_to_height=None, upscale_crop=False, upscaler_1_name=None, upscaler_2_name=None, upscaler_2_visibility=0.0): + if upscale_mode == 1: + pp.shared.target_width = upscale_to_width + pp.shared.target_height = upscale_to_height + else: + pp.shared.target_width = int(pp.image.width * upscale_by) + pp.shared.target_height = int(pp.image.height * upscale_by) + def process(self, pp: scripts_postprocessing.PostprocessedImage, upscale_mode=1, upscale_by=2.0, upscale_to_width=None, upscale_to_height=None, upscale_crop=False, upscaler_1_name=None, upscaler_2_name=None, upscaler_2_visibility=0.0): if upscaler_1_name == "None": upscaler_1_name = None @@ -126,6 +134,10 @@ class ScriptPostprocessingUpscaleSimple(ScriptPostprocessingUpscale): "upscaler_name": upscaler_name, } + def process_firstpass(self, pp: scripts_postprocessing.PostprocessedImage, upscale_by=2.0, upscaler_name=None): + pp.shared.target_width = int(pp.image.width * upscale_by) + pp.shared.target_height = int(pp.image.height * upscale_by) + def process(self, pp: scripts_postprocessing.PostprocessedImage, upscale_by=2.0, upscaler_name=None): if upscaler_name is None or upscaler_name == "None": return diff --git a/scripts/processing_autosized_crop.py b/scripts/processing_autosized_crop.py new file mode 100644 index 000000000..c09802264 --- /dev/null +++ b/scripts/processing_autosized_crop.py @@ -0,0 +1,64 @@ +from PIL import Image + +from modules import scripts_postprocessing, ui_components +import gradio as gr + + +def center_crop(image: Image, w: int, h: int): + iw, ih = image.size + if ih / h < iw / w: + sw = w * ih / h + box = (iw - sw) / 2, 0, iw - (iw - sw) / 2, ih + else: + sh = h * iw / w + box = 0, (ih - sh) / 2, iw, ih - (ih - sh) / 2 + return image.resize((w, h), Image.Resampling.LANCZOS, box) + + +def multicrop_pic(image: Image, mindim, maxdim, minarea, maxarea, objective, threshold): + iw, ih = image.size + err = lambda w, h: 1 - (lambda x: x if x < 1 else 1 / x)(iw / ih / (w / h)) + wh = max(((w, h) for w in range(mindim, maxdim + 1, 64) for h in range(mindim, maxdim + 1, 64) + if minarea <= w * h <= maxarea and err(w, h) <= threshold), + key=lambda wh: (wh[0] * wh[1], -err(*wh))[::1 if objective == 'Maximize area' else -1], + default=None + ) + return wh and center_crop(image, *wh) + + +class ScriptPostprocessingAutosizedCrop(scripts_postprocessing.ScriptPostprocessing): + name = "Auto-sized crop" + order = 4000 + + def ui(self): + with ui_components.InputAccordion(False, label="Auto-sized crop") as enable: + gr.Markdown('Each image is center-cropped with an automatically chosen width and height.') + with gr.Row(): + mindim = gr.Slider(minimum=64, maximum=2048, step=8, label="Dimension lower bound", value=384, elem_id="postprocess_multicrop_mindim") + maxdim = gr.Slider(minimum=64, maximum=2048, step=8, label="Dimension upper bound", value=768, elem_id="postprocess_multicrop_maxdim") + with gr.Row(): + minarea = gr.Slider(minimum=64 * 64, maximum=2048 * 2048, step=1, label="Area lower bound", value=64 * 64, elem_id="postprocess_multicrop_minarea") + maxarea = gr.Slider(minimum=64 * 64, maximum=2048 * 2048, step=1, label="Area upper bound", value=640 * 640, elem_id="postprocess_multicrop_maxarea") + with gr.Row(): + objective = gr.Radio(["Maximize area", "Minimize error"], value="Maximize area", label="Resizing objective", elem_id="postprocess_multicrop_objective") + threshold = gr.Slider(minimum=0, maximum=1, step=0.01, label="Error threshold", value=0.1, elem_id="postprocess_multicrop_threshold") + + return { + "enable": enable, + "mindim": mindim, + "maxdim": maxdim, + "minarea": minarea, + "maxarea": maxarea, + "objective": objective, + "threshold": threshold, + } + + def process(self, pp: scripts_postprocessing.PostprocessedImage, enable, mindim, maxdim, minarea, maxarea, objective, threshold): + if not enable: + return + + cropped = multicrop_pic(pp.image, mindim, maxdim, minarea, maxarea, objective, threshold) + if cropped is not None: + pp.image = cropped + else: + print(f"skipped {pp.image.width}x{pp.image.height} image (can't find suitable size within error threshold)") From a5f61aa8c5933d8e5a0e0aa841138eeaccd86d62 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sat, 2 Dec 2023 18:03:34 +0300 Subject: [PATCH 180/342] potential fix for #14172 --- modules/sd_hijack.py | 12 ++++-------- 1 file changed, 4 insertions(+), 8 deletions(-) diff --git a/modules/sd_hijack.py b/modules/sd_hijack.py index 3d340fc9b..14fe62c73 100644 --- a/modules/sd_hijack.py +++ b/modules/sd_hijack.py @@ -38,6 +38,10 @@ ldm.models.diffusion.ddpm.print = shared.ldm_print optimizers = [] current_optimizer: sd_hijack_optimizations.SdOptimization = None +ldm_original_forward = patches.patch(__file__, ldm.modules.diffusionmodules.openaimodel.UNetModel, "forward", sd_unet.UNetModel_forward) +sgm_original_forward = patches.patch(__file__, sgm.modules.diffusionmodules.openaimodel.UNetModel, "forward", sd_unet.UNetModel_forward) + + def list_optimizers(): new_optimizers = script_callbacks.list_optimizers_callback() @@ -255,9 +259,6 @@ class StableDiffusionModelHijack: import modules.models.diffusion.ddpm_edit - ldm_original_forward = patches.patch(__file__, ldm.modules.diffusionmodules.openaimodel.UNetModel, "forward", sd_unet.UNetModel_forward) - sgm_original_forward = patches.patch(__file__, sgm.modules.diffusionmodules.openaimodel.UNetModel, "forward", sd_unet.UNetModel_forward) - if isinstance(m, ldm.models.diffusion.ddpm.LatentDiffusion): sd_unet.original_forward = ldm_original_forward elif isinstance(m, modules.models.diffusion.ddpm_edit.LatentDiffusion): @@ -303,11 +304,6 @@ class StableDiffusionModelHijack: self.layers = None self.clip = None - patches.undo(__file__, ldm.modules.diffusionmodules.openaimodel.UNetModel, "forward") - patches.undo(__file__, sgm.modules.diffusionmodules.openaimodel.UNetModel, "forward") - - sd_unet.original_forward = None - def apply_circular(self, enable): if self.circular_enabled == enable: From ac02216e540cd581f9169c6c791e55721e3117b0 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sat, 2 Dec 2023 19:35:47 +0300 Subject: [PATCH 181/342] alternate implementation for unet forward replacement that does not depend on hijack being applied --- modules/sd_hijack.py | 7 +++++-- modules/sd_unet.py | 14 ++++++++------ 2 files changed, 13 insertions(+), 8 deletions(-) diff --git a/modules/sd_hijack.py b/modules/sd_hijack.py index 14fe62c73..e139d9964 100644 --- a/modules/sd_hijack.py +++ b/modules/sd_hijack.py @@ -38,8 +38,11 @@ ldm.models.diffusion.ddpm.print = shared.ldm_print optimizers = [] current_optimizer: sd_hijack_optimizations.SdOptimization = None -ldm_original_forward = patches.patch(__file__, ldm.modules.diffusionmodules.openaimodel.UNetModel, "forward", sd_unet.UNetModel_forward) -sgm_original_forward = patches.patch(__file__, sgm.modules.diffusionmodules.openaimodel.UNetModel, "forward", sd_unet.UNetModel_forward) +ldm_patched_forward = sd_unet.create_unet_forward(ldm.modules.diffusionmodules.openaimodel.UNetModel.forward) +ldm_original_forward = patches.patch(__file__, ldm.modules.diffusionmodules.openaimodel.UNetModel, "forward", ldm_patched_forward) + +sgm_patched_forward = sd_unet.create_unet_forward(sgm.modules.diffusionmodules.openaimodel.UNetModel.forward) +sgm_original_forward = patches.patch(__file__, sgm.modules.diffusionmodules.openaimodel.UNetModel, "forward", sgm_patched_forward) def list_optimizers(): diff --git a/modules/sd_unet.py b/modules/sd_unet.py index 6a7bc9e26..a771849c8 100644 --- a/modules/sd_unet.py +++ b/modules/sd_unet.py @@ -5,8 +5,7 @@ from modules import script_callbacks, shared, devices unet_options = [] current_unet_option = None current_unet = None -original_forward = None - +original_forward = None # not used, only left temporarily for compatibility def list_unets(): new_unets = script_callbacks.list_unets_callback() @@ -84,9 +83,12 @@ class SdUnet(torch.nn.Module): pass -def UNetModel_forward(self, x, timesteps=None, context=None, *args, **kwargs): - if current_unet is not None: - return current_unet.forward(x, timesteps, context, *args, **kwargs) +def create_unet_forward(original_forward): + def UNetModel_forward(self, x, timesteps=None, context=None, *args, **kwargs): + if current_unet is not None: + return current_unet.forward(x, timesteps, context, *args, **kwargs) - return original_forward(self, x, timesteps, context, *args, **kwargs) + return original_forward(self, x, timesteps, context, *args, **kwargs) + + return UNetModel_forward From 309a606c2fa645b6b8623f96ea56117e685a47fb Mon Sep 17 00:00:00 2001 From: drhead <1313496+drhead@users.noreply.github.com> Date: Sat, 2 Dec 2023 13:07:45 -0500 Subject: [PATCH 182/342] ensure that original alpha bar always exists --- modules/processing.py | 18 ++++++++++-------- 1 file changed, 10 insertions(+), 8 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index d73c8bfc1..bfa59038b 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -882,15 +882,17 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: alphas_bar[-1] = 4.8973451890853435e-08 return alphas_bar - if hasattr(p.sd_model, 'alphas_cumprod') and hasattr(p.sd_model, 'alphas_cumprod_original'): - p.sd_model.alphas_cumprod = p.sd_model.alphas_cumprod_original.to(shared.device) + if hasattr(p.sd_model, 'alphas_cumprod') and not hasattr(p.sd_model, 'alphas_cumprod_original'): + p.sd_model.alphas_cumprod_original = p.sd_model.alphas_cumprod + + p.sd_model.alphas_cumprod = p.sd_model.alphas_cumprod_original.to(shared.device) - if opts.use_downcasted_alpha_bar: - p.extra_generation_params['Downcast alphas_cumprod'] = opts.use_downcasted_alpha_bar - p.sd_model.alphas_cumprod = p.sd_model.alphas_cumprod.half().to(shared.device) - if opts.sd_noise_schedule == "Zero Terminal SNR": - p.extra_generation_params['Noise Schedule'] = opts.sd_noise_schedule - p.sd_model.alphas_cumprod = rescale_zero_terminal_snr_abar(p.sd_model.alphas_cumprod).to(shared.device) + if opts.use_downcasted_alpha_bar: + p.extra_generation_params['Downcast alphas_cumprod'] = opts.use_downcasted_alpha_bar + p.sd_model.alphas_cumprod = p.sd_model.alphas_cumprod.half().to(shared.device) + if opts.sd_noise_schedule == "Zero Terminal SNR": + p.extra_generation_params['Noise Schedule'] = opts.sd_noise_schedule + p.sd_model.alphas_cumprod = rescale_zero_terminal_snr_abar(p.sd_model.alphas_cumprod).to(shared.device) with devices.without_autocast() if devices.unet_needs_upcast else devices.autocast(): samples_ddim = p.sample(conditioning=p.c, unconditional_conditioning=p.uc, seeds=p.seeds, subseeds=p.subseeds, subseed_strength=p.subseed_strength, prompts=p.prompts) From 81c4ddf6ebebe6f18338de3b0391da1d8521a525 Mon Sep 17 00:00:00 2001 From: drhead <1313496+drhead@users.noreply.github.com> Date: Sat, 2 Dec 2023 13:11:00 -0500 Subject: [PATCH 183/342] fix linting --- modules/processing.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/processing.py b/modules/processing.py index bfa59038b..eeccea743 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -884,7 +884,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: if hasattr(p.sd_model, 'alphas_cumprod') and not hasattr(p.sd_model, 'alphas_cumprod_original'): p.sd_model.alphas_cumprod_original = p.sd_model.alphas_cumprod - + p.sd_model.alphas_cumprod = p.sd_model.alphas_cumprod_original.to(shared.device) if opts.use_downcasted_alpha_bar: From 83e8c322762c545fd589c060811379582926060f Mon Sep 17 00:00:00 2001 From: catboxanon <122327233+catboxanon@users.noreply.github.com> Date: Sat, 2 Dec 2023 13:30:53 -0500 Subject: [PATCH 184/342] Fix `save_samples` being checked early when saving masked composite --- modules/processing.py | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index 5ab6dddef..4f265801c 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -938,14 +938,14 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: if opts.enable_pnginfo: image.info["parameters"] = text output_images.append(image) - if save_samples and hasattr(p, 'mask_for_overlay') and p.mask_for_overlay and any([opts.save_mask, opts.save_mask_composite, opts.return_mask, opts.return_mask_composite]): + if hasattr(p, 'mask_for_overlay') and p.mask_for_overlay and any([opts.save_mask, opts.save_mask_composite, opts.return_mask, opts.return_mask_composite]): image_mask = p.mask_for_overlay.convert('RGB') image_mask_composite = Image.composite(image.convert('RGBA').convert('RGBa'), Image.new('RGBa', image.size), images.resize_image(2, p.mask_for_overlay, image.width, image.height).convert('L')).convert('RGBA') - if opts.save_mask: + if save_samples and opts.save_mask: images.save_image(image_mask, p.outpath_samples, "", p.seeds[i], p.prompts[i], opts.samples_format, info=infotext(i), p=p, suffix="-mask") - if opts.save_mask_composite: + if save_samples and opts.save_mask_composite: images.save_image(image_mask_composite, p.outpath_samples, "", p.seeds[i], p.prompts[i], opts.samples_format, info=infotext(i), p=p, suffix="-mask-composite") if opts.return_mask: From 4a43334376d9e116f7a1446f042f9af9c0484fc6 Mon Sep 17 00:00:00 2001 From: drhead Date: Sat, 2 Dec 2023 14:05:42 -0500 Subject: [PATCH 185/342] Revert 309a606c --- modules/processing.py | 18 ++++++++---------- 1 file changed, 8 insertions(+), 10 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index eeccea743..d73c8bfc1 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -882,17 +882,15 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: alphas_bar[-1] = 4.8973451890853435e-08 return alphas_bar - if hasattr(p.sd_model, 'alphas_cumprod') and not hasattr(p.sd_model, 'alphas_cumprod_original'): - p.sd_model.alphas_cumprod_original = p.sd_model.alphas_cumprod + if hasattr(p.sd_model, 'alphas_cumprod') and hasattr(p.sd_model, 'alphas_cumprod_original'): + p.sd_model.alphas_cumprod = p.sd_model.alphas_cumprod_original.to(shared.device) - p.sd_model.alphas_cumprod = p.sd_model.alphas_cumprod_original.to(shared.device) - - if opts.use_downcasted_alpha_bar: - p.extra_generation_params['Downcast alphas_cumprod'] = opts.use_downcasted_alpha_bar - p.sd_model.alphas_cumprod = p.sd_model.alphas_cumprod.half().to(shared.device) - if opts.sd_noise_schedule == "Zero Terminal SNR": - p.extra_generation_params['Noise Schedule'] = opts.sd_noise_schedule - p.sd_model.alphas_cumprod = rescale_zero_terminal_snr_abar(p.sd_model.alphas_cumprod).to(shared.device) + if opts.use_downcasted_alpha_bar: + p.extra_generation_params['Downcast alphas_cumprod'] = opts.use_downcasted_alpha_bar + p.sd_model.alphas_cumprod = p.sd_model.alphas_cumprod.half().to(shared.device) + if opts.sd_noise_schedule == "Zero Terminal SNR": + p.extra_generation_params['Noise Schedule'] = opts.sd_noise_schedule + p.sd_model.alphas_cumprod = rescale_zero_terminal_snr_abar(p.sd_model.alphas_cumprod).to(shared.device) with devices.without_autocast() if devices.unet_needs_upcast else devices.autocast(): samples_ddim = p.sample(conditioning=p.c, unconditional_conditioning=p.uc, seeds=p.seeds, subseeds=p.subseeds, subseed_strength=p.subseed_strength, prompts=p.prompts) From dc1adeecdd02f3fb910481e808a6d60a77100fea Mon Sep 17 00:00:00 2001 From: drhead Date: Sat, 2 Dec 2023 14:06:56 -0500 Subject: [PATCH 186/342] Create alphas_cumprod_original on full precision path --- modules/sd_models.py | 1 + 1 file changed, 1 insertion(+) diff --git a/modules/sd_models.py b/modules/sd_models.py index de80a493a..976c7d5be 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -374,6 +374,7 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer if shared.cmd_opts.no_half: model.float() + model.alphas_cumprod_original = alphas_cumprod devices.dtype_unet = torch.float32 timer.record("apply float()") else: From 78acdcf677a96894651ff0d7d8287f2a994f3781 Mon Sep 17 00:00:00 2001 From: drhead Date: Sat, 2 Dec 2023 14:09:18 -0500 Subject: [PATCH 187/342] fix variable --- modules/sd_models.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/sd_models.py b/modules/sd_models.py index 976c7d5be..5a19a00a5 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -374,7 +374,7 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer if shared.cmd_opts.no_half: model.float() - model.alphas_cumprod_original = alphas_cumprod + model.alphas_cumprod_original = model.alphas_cumprod devices.dtype_unet = torch.float32 timer.record("apply float()") else: From 9528d66c9479d02c83b8db6107f6b0cb741612dc Mon Sep 17 00:00:00 2001 From: catboxanon <122327233+catboxanon@users.noreply.github.com> Date: Sat, 2 Dec 2023 14:56:26 -0500 Subject: [PATCH 188/342] Re-add setting lost as part of e294e46 --- modules/shared_options.py | 1 + 1 file changed, 1 insertion(+) diff --git a/modules/shared_options.py b/modules/shared_options.py index 859dee404..e5de0d018 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -255,6 +255,7 @@ options_templates.update(options_section(('ui_prompt_editing', "Prompt editing", "keyedit_precision_attention": OptionInfo(0.1, "Precision for (attention:1.1) when editing the prompt with Ctrl+up/down", gr.Slider, {"minimum": 0.01, "maximum": 0.2, "step": 0.001}), "keyedit_precision_extra": OptionInfo(0.05, "Precision for when editing the prompt with Ctrl+up/down", gr.Slider, {"minimum": 0.01, "maximum": 0.2, "step": 0.001}), "keyedit_delimiters": OptionInfo(r".,\/!?%^*;:{}=`~() ", "Word delimiters when editing the prompt with Ctrl+up/down"), + "keyedit_delimiters_whitespace": OptionInfo(["Tab", "Carriage Return", "Line Feed"], "Ctrl+up/down whitespace delimiters", gr.CheckboxGroup, lambda: {"choices": ["Tab", "Carriage Return", "Line Feed"]}), "disable_token_counters": OptionInfo(False, "Disable prompt token counters").needs_reload_ui(), })) From 609dea36ea919aa7db42fd4233c416a45c74578b Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Sat, 2 Dec 2023 18:56:49 -0700 Subject: [PATCH 189/342] Added utility functions related to processing masks. --- modules/images.py | 191 ++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 191 insertions(+) diff --git a/modules/images.py b/modules/images.py index eb6447338..b5a0cead6 100644 --- a/modules/images.py +++ b/modules/images.py @@ -776,3 +776,194 @@ def flatten(img, bgcolor): img = background return img.convert('RGB') + + +def weighted_histogram_filter(img, kernel, kernel_center, percentile_min=0.0, percentile_max=1.0, min_width=1.0): + """ + Generalization convolution filter capable of applying + weighted mean, median, maximum, and minimum filters + parametrically using an arbitrary kernel. + + Args: + img (nparray): + The image, a 2-D array of floats, to which the filter is being applied. + kernel (nparray): + The kernel, a 2-D array of floats. + kernel_center (nparray): + The kernel center coordinate, a 1-D array with two elements. + percentile_min (float): + The lower bound of the histogram window used by the filter, + from 0 to 1. + percentile_max (float): + The upper bound of the histogram window used by the filter, + from 0 to 1. + min_width (float): + The minimum size of the histogram window bounds, in weight units. + Must be greater than 0. + + Returns: + (nparray): A filtered copy of the input image "img", a 2-D array of floats. + """ + + # Converts an index tuple into a vector. + def vec(x): + return np.array(x) + + kernel_min = -kernel_center + kernel_max = vec(kernel.shape) - kernel_center + + def weighted_histogram_filter_single(idx): + idx = vec(idx) + min_index = np.maximum(0, idx + kernel_min) + max_index = np.minimum(vec(img.shape), idx + kernel_max) + window_shape = max_index - min_index + + class WeightedElement: + """ + An element of the histogram, its weight + and bounds. + """ + def __init__(self, value, weight): + self.value: float = value + self.weight: float = weight + self.window_min: float = 0.0 + self.window_max: float = 1.0 + + # Collect the values in the image as WeightedElements, + # weighted by their corresponding kernel values. + values = [] + for window_tup in np.ndindex(tuple(window_shape)): + window_index = vec(window_tup) + image_index = window_index + min_index + centered_kernel_index = image_index - idx + kernel_index = centered_kernel_index + kernel_center + element = WeightedElement(img[tuple(image_index)], kernel[tuple(kernel_index)]) + values.append(element) + + def sort_key(x: WeightedElement): + return x.value + + values.sort(key=sort_key) + + # Calculate the height of the stack (sum) + # and each sample's range they occupy in the stack + sum = 0 + for i in range(len(values)): + values[i].window_min = sum + sum += values[i].weight + values[i].window_max = sum + + # Calculate what range of this stack ("window") + # we want to get the weighted average across. + window_min = sum * percentile_min + window_max = sum * percentile_max + window_width = window_max - window_min + + # Ensure the window is within the stack and at least a certain size. + if window_width < min_width: + window_center = (window_min + window_max) / 2 + window_min = window_center - min_width / 2 + window_max = window_center + min_width / 2 + + if window_max > sum: + window_max = sum + window_min = sum - min_width + + if window_min < 0: + window_min = 0 + window_max = min_width + + value = 0 + value_weight = 0 + + # Get the weighted average of all the samples + # that overlap with the window, weighted + # by the size of their overlap. + for i in range(len(values)): + if window_min >= values[i].window_max: + continue + if window_max <= values[i].window_min: + break + + s = max(window_min, values[i].window_min) + e = min(window_max, values[i].window_max) + w = e - s + + value += values[i].value * w + value_weight += w + + return value / value_weight if value_weight != 0 else 0 + + img_out = img.copy() + + # Apply the kernel operation over each pixel. + for index in np.ndindex(img.shape): + img_out[index] = weighted_histogram_filter_single(index) + + return img_out + +def smoothstep(x): + """ + The smoothstep function, input should be clamped to 0-1 range. + Turns a diagonal line (f(x) = x) into a sigmoid-like curve. + """ + return x * x * (3 - 2 * x) + +def smootherstep(x): + """ + The smootherstep function, input should be clamped to 0-1 range. + Turns a diagonal line (f(x) = x) into a sigmoid-like curve. + """ + return x * x * x * (x * (6 * x - 15) + 10) + + +def get_gaussian_kernel(stddev_radius=1.0, max_radius=2): + """ + Creates a Gaussian kernel with thresholded edges. + + Args: + stddev_radius (float): + Standard deviation of the gaussian kernel, in pixels. + max_radius (int): + The size of the filter kernel. The number of pixels is (max_radius*2+1) ** 2. + The kernel is thresholded so that any values one pixel beyond this radius + is weighted at 0. + + Returns: + (nparray, nparray): A kernel array (shape: (N, N)), its center coordinate (shape: (2)) + """ + # Evaluates a 0-1 normalized gaussian function for a given square distance from the mean. + def gaussian(sqr_mag): + return math.exp(-sqr_mag / (stddev_radius * stddev_radius)) + + # Helper function for converting a tuple to an array. + def vec(x): + return np.array(x) + + """ + Since a gaussian is unbounded, we need to limit ourselves + to a finite range. + We taper the ends off at the end of that range so they equal zero + while preserving the maximum value of 1 at the mean. + """ + zero_radius = max_radius + 1.0 + gauss_zero = gaussian(zero_radius * zero_radius) + gauss_kernel_scale = 1 / (1 - gauss_zero) + + def gaussian_kernel_func(coordinate): + x = coordinate[0] ** 2.0 + coordinate[1] ** 2.0 + x = gaussian(x) + x -= gauss_zero + x /= gauss_kernel_scale + x = max(0.0, x) + return x + + size = max_radius * 2 + 1 + kernel_center = max_radius + kernel = np.zeros((size, size)) + + for index in np.ndindex(kernel.shape): + kernel[index] = gaussian_kernel_func(vec(index) - kernel_center) + + return kernel, kernel_center + From 73ab982d1b7394574d1cf2e0a151bc457eeed769 Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Sat, 2 Dec 2023 21:07:02 -0700 Subject: [PATCH 190/342] Blend masks are now produced afterward, based on an estimate of the visual difference between the original and modified latent images. This should remove ghosting and clipping artifacts from masks, while preserving the details of largely unchanged content. --- modules/processing.py | 119 ++++++++++++++++++++++++++++++++---------- 1 file changed, 90 insertions(+), 29 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index 92fdebadd..ad716e11f 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -9,7 +9,7 @@ from dataclasses import dataclass, field import torch import numpy as np -from PIL import Image, ImageOps +from PIL import Image, ImageOps, ImageFilter import random import cv2 from skimage import exposure @@ -62,6 +62,16 @@ def apply_color_correction(correction, original_image): return image.convert('RGB') +def uncrop(image, dest_size, paste_loc): + x, y, w, h = paste_loc + base_image = Image.new('RGBA', dest_size) + image = images.resize_image(1, image, w, h) + base_image.paste(image, (x, y)) + image = base_image + + return image + + def apply_overlay(image, paste_loc, index, overlays): if overlays is None or index >= len(overlays): return image @@ -69,11 +79,7 @@ def apply_overlay(image, paste_loc, index, overlays): overlay = overlays[index] if paste_loc is not None: - x, y, w, h = paste_loc - base_image = Image.new('RGBA', (overlay.width, overlay.height)) - image = images.resize_image(1, image, w, h) - base_image.paste(image, (x, y)) - image = base_image + image = uncrop(image, (overlay.width, overlay.height), paste_loc) image = image.convert('RGBA') image.alpha_composite(overlay) @@ -140,6 +146,7 @@ class StableDiffusionProcessing: do_not_save_grid: bool = False extra_generation_params: dict[str, Any] = None overlay_images: list = None + masks_for_overlay: list = None eta: float = None do_not_reload_embeddings: bool = False denoising_strength: float = 0 @@ -865,11 +872,66 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: if getattr(samples_ddim, 'already_decoded', False): x_samples_ddim = samples_ddim + # todo: generate masks the old fashioned way else: if opts.sd_vae_decode_method != 'Full': p.extra_generation_params['VAE Decoder'] = opts.sd_vae_decode_method - x_samples_ddim = decode_latent_batch(p.sd_model, samples_ddim, target_device=devices.cpu, check_for_nans=True) + # Generate the mask(s) based on similarity between the original and denoised latent vectors + if getattr(p, "image_mask", None) is not None: + # latent_mask = p.nmask[0].float().cpu() + + # convert the original mask into a form we use to scale distances for thresholding + # mask_scalar = 1-(torch.clamp(latent_mask, min=0, max=1) ** (p.mask_blend_scale / 2)) + # mask_scalar = mask_scalar / (1.00001-mask_scalar) + # mask_scalar = mask_scalar.numpy() + + latent_orig = p.init_latent + latent_proc = samples_ddim + latent_distance = torch.norm(latent_proc - latent_orig, p=2, dim=1) + + kernel, kernel_center = images.get_gaussian_kernel(stddev_radius=1.5, max_radius=2) + + for i, (distance_map, overlay_image) in enumerate(zip(latent_distance, p.overlay_images)): + converted_mask = distance_map.float().cpu().numpy() + converted_mask = images.weighted_histogram_filter(converted_mask, kernel, kernel_center, + percentile_min=0.9, percentile_max=1, min_width=1) + converted_mask = images.weighted_histogram_filter(converted_mask, kernel, kernel_center, + percentile_min=0.25, percentile_max=0.75, min_width=1) + + # The distance at which opacity of original decreases to 50% + # half_weighted_distance = 1 # * mask_scalar + # converted_mask = converted_mask / half_weighted_distance + + converted_mask = 1 / (1 + converted_mask ** 2) + converted_mask = images.smootherstep(converted_mask) + converted_mask = 1 - converted_mask + converted_mask = 255. * converted_mask + converted_mask = converted_mask.astype(np.uint8) + converted_mask = Image.fromarray(converted_mask) + converted_mask = images.resize_image(2, converted_mask, p.width, p.height) + converted_mask = create_binary_mask(converted_mask) + + # Remove aliasing artifacts using a gaussian blur. + converted_mask = converted_mask.filter(ImageFilter.GaussianBlur(radius=4)) + + # Expand the mask to fit the whole image if needed. + if p.paste_to is not None: + converted_mask = uncrop(converted_mask, + (overlay_image.width, overlay_image.height), + p.paste_to) + + p.masks_for_overlay[i] = converted_mask + + image_masked = Image.new('RGBa', (overlay_image.width, overlay_image.height)) + image_masked.paste(overlay_image.convert("RGBA").convert("RGBa"), + mask=ImageOps.invert(converted_mask.convert('L'))) + + p.overlay_images[i] = image_masked.convert('RGBA') + + x_samples_ddim = decode_latent_batch(p.sd_model, samples_ddim, + target_device=devices.cpu, + check_for_nans=True) x_samples_ddim = torch.stack(x_samples_ddim).float() x_samples_ddim = torch.clamp((x_samples_ddim + 1.0) / 2.0, min=0.0, max=1.0) @@ -892,7 +954,9 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: x_samples_ddim = batch_params.images def infotext(index=0, use_main_prompt=False): - return create_infotext(p, p.prompts, p.seeds, p.subseeds, use_main_prompt=use_main_prompt, index=index, all_negative_prompts=p.negative_prompts) + return create_infotext(p, p.prompts, p.seeds, p.subseeds, + use_main_prompt=use_main_prompt, index=index, + all_negative_prompts=p.negative_prompts) save_samples = p.save_samples() @@ -923,19 +987,27 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: images.save_image(image_without_cc, p.outpath_samples, "", p.seeds[i], p.prompts[i], opts.samples_format, info=infotext(i), p=p, suffix="-before-color-correction") image = apply_color_correction(p.color_corrections[i], image) + # If the intention is to show the output from the model + # that is being composited over the original image, + # we need to keep the original image around + # and use it in the composite step. + original_denoised_image = image.copy() image = apply_overlay(image, p.paste_to, i, p.overlay_images) if save_samples: - images.save_image(image, p.outpath_samples, "", p.seeds[i], p.prompts[i], opts.samples_format, info=infotext(i), p=p) + images.save_image(image, p.outpath_samples, "", p.seeds[i], + p.prompts[i], opts.samples_format, info=infotext(i), p=p) text = infotext(i) infotexts.append(text) if opts.enable_pnginfo: image.info["parameters"] = text output_images.append(image) - if save_samples and hasattr(p, 'mask_for_overlay') and p.mask_for_overlay and any([opts.save_mask, opts.save_mask_composite, opts.return_mask, opts.return_mask_composite]): - image_mask = p.mask_for_overlay.convert('RGB') - image_mask_composite = Image.composite(image.convert('RGBA').convert('RGBa'), Image.new('RGBa', image.size), images.resize_image(2, p.mask_for_overlay, image.width, image.height).convert('L')).convert('RGBA') + if save_samples and hasattr(p, 'masks_for_overlay') and p.masks_for_overlay and any([opts.save_mask, opts.save_mask_composite, opts.return_mask, opts.return_mask_composite]): + image_mask = p.masks_for_overlay[i].convert('RGB') + image_mask_composite = Image.composite( + original_denoised_image.convert('RGBA').convert('RGBa'), Image.new('RGBa', image.size), + images.resize_image(2, p.masks_for_overlay[i], image.width, image.height).convert('L')).convert('RGBA') if opts.save_mask: images.save_image(image_mask, p.outpath_samples, "", p.seeds[i], p.prompts[i], opts.samples_format, info=infotext(i), p=p, suffix="-mask") @@ -1364,7 +1436,6 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): nmask: torch.Tensor = field(default=None, init=False) image_conditioning: torch.Tensor = field(default=None, init=False) init_img_hash: str = field(default=None, init=False) - mask_for_overlay: Image = field(default=None, init=False) init_latent: torch.Tensor = field(default=None, init=False) def __post_init__(self): @@ -1415,12 +1486,6 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): image_mask = Image.fromarray(np_mask) if self.inpaint_full_res: - np_mask = np.array(image_mask).astype(np.float32) - np_mask /= 255 - np_mask = 1-pow(1-np_mask, 100) - np_mask *= 255 - np_mask = np.clip(np_mask, 0, 255).astype(np.uint8) - self.mask_for_overlay = Image.fromarray(np_mask) mask = image_mask.convert('L') crop_region = masking.get_crop_region(np.array(mask), self.inpaint_full_res_padding) crop_region = masking.expand_crop_region(crop_region, self.width, self.height, mask.width, mask.height) @@ -1431,13 +1496,8 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): self.paste_to = (x1, y1, x2-x1, y2-y1) else: image_mask = images.resize_image(self.resize_mode, image_mask, self.width, self.height) - np_mask = np.array(image_mask).astype(np.float32) - np_mask /= 255 - np_mask = 1-pow(1-np_mask, 100) - np_mask *= 255 - np_mask = np.clip(np_mask, 0, 255).astype(np.uint8) - self.mask_for_overlay = Image.fromarray(np_mask) + self.masks_for_overlay = [] self.overlay_images = [] latent_mask = self.latent_mask if self.latent_mask is not None else image_mask @@ -1459,10 +1519,8 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): image = images.resize_image(self.resize_mode, image, self.width, self.height) if image_mask is not None: - image_masked = Image.new('RGBa', (image.width, image.height)) - image_masked.paste(image.convert("RGBA").convert("RGBa"), mask=ImageOps.invert(self.mask_for_overlay.convert('L'))) - - self.overlay_images.append(image_masked.convert('RGBA')) + self.overlay_images.append(image) + self.masks_for_overlay.append(image_mask) # crop_region is not None if we are doing inpaint full res if crop_region is not None: @@ -1486,6 +1544,9 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): if self.overlay_images is not None: self.overlay_images = self.overlay_images * self.batch_size + if self.masks_for_overlay is not None: + self.masks_for_overlay = self.masks_for_overlay * self.batch_size + if self.color_corrections is not None and len(self.color_corrections) == 1: self.color_corrections = self.color_corrections * self.batch_size From bb04d400c95df01d191ef6c1a43e66b95425fa33 Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Sat, 2 Dec 2023 21:08:26 -0700 Subject: [PATCH 191/342] Rewrote latent_blend() to use in-place operations and to aggressively "del" references with the intention of minimizing allocations and easing garbage collection. --- modules/sd_samplers_cfg_denoiser.py | 39 ++++++++++++++++++++--------- 1 file changed, 27 insertions(+), 12 deletions(-) diff --git a/modules/sd_samplers_cfg_denoiser.py b/modules/sd_samplers_cfg_denoiser.py index ceb612d79..efbe7a403 100644 --- a/modules/sd_samplers_cfg_denoiser.py +++ b/modules/sd_samplers_cfg_denoiser.py @@ -102,29 +102,44 @@ class CFGDenoiser(torch.nn.Module): The "detail_preservation" factor biases the magnitude interpolation towards the larger of the two magnitudes. """ - # Record the original latent vector magnitudes. - # We bring them to a power so that larger magnitudes are favored over smaller ones. - # 64-bit operations are used here to allow large exponents. - a_magnitude = torch.norm(a, p=2, dim=1).to(torch.float64) ** self.inpaint_detail_preservation - b_magnitude = torch.norm(b, p=2, dim=1).to(torch.float64) ** self.inpaint_detail_preservation + # NOTE: We use inplace operations wherever possible. one_minus_t = 1 - t - # Interpolate the powered magnitudes, then un-power them (bring them back to a power of 1). - interp_magnitude = (a_magnitude * one_minus_t + b_magnitude * t) ** (1 / self.inpaint_detail_preservation) - # Linearly interpolate the image vectors. - image_interp = a * one_minus_t + b * t + a_scaled = a * one_minus_t + b_scaled = b * t + image_interp = a_scaled + image_interp.add_(b_scaled) + result_type = image_interp.dtype + del a_scaled, b_scaled # Calculate the magnitude of the interpolated vectors. (We will remove this magnitude.) # 64-bit operations are used here to allow large exponents. - image_interp_magnitude = torch.norm(image_interp, p=2, dim=1).to(torch.float64) + 0.0001 + current_magnitude = torch.norm(image_interp, p=2, dim=1).to(torch.float64).add_(0.00001) + + # Interpolate the powered magnitudes, then un-power them (bring them back to a power of 1). + a_magnitude = torch.norm(a, p=2, dim=1).to(torch.float64).pow_(self.inpaint_detail_preservation) * one_minus_t + b_magnitude = torch.norm(b, p=2, dim=1).to(torch.float64).pow_(self.inpaint_detail_preservation) * t + desired_magnitude = a_magnitude + desired_magnitude.add_(b_magnitude).pow_(1 / self.inpaint_detail_preservation) + del a_magnitude, b_magnitude, one_minus_t # Change the linearly interpolated image vectors' magnitudes to the value we want. # This is the last 64-bit operation. - image_interp *= (interp_magnitude / image_interp_magnitude).to(image_interp.dtype) + image_interp_scaling_factor = desired_magnitude + image_interp_scaling_factor.div_(current_magnitude) + image_interp_scaled = image_interp + image_interp_scaled.mul_(image_interp_scaling_factor) + del current_magnitude + del desired_magnitude + del image_interp + del image_interp_scaling_factor - return image_interp + image_interp_scaled = image_interp_scaled.to(result_type) + del result_type + + return image_interp_scaled def get_modified_nmask(nmask, _sigma): """ From d3fdc4af61b7560eede52290e1ede48185680089 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Sun, 3 Dec 2023 18:22:00 +0900 Subject: [PATCH 192/342] rework mask and mask_composite logic --- modules/processing.py | 27 +++++++++++++-------------- 1 file changed, 13 insertions(+), 14 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index 4f265801c..6f01c95f5 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -938,21 +938,20 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: if opts.enable_pnginfo: image.info["parameters"] = text output_images.append(image) - if hasattr(p, 'mask_for_overlay') and p.mask_for_overlay and any([opts.save_mask, opts.save_mask_composite, opts.return_mask, opts.return_mask_composite]): - image_mask = p.mask_for_overlay.convert('RGB') - image_mask_composite = Image.composite(image.convert('RGBA').convert('RGBa'), Image.new('RGBa', image.size), images.resize_image(2, p.mask_for_overlay, image.width, image.height).convert('L')).convert('RGBA') + if hasattr(p, 'mask_for_overlay') and p.mask_for_overlay: + if opts.return_mask or opts.save_mask: + image_mask = p.mask_for_overlay.convert('RGB') + if save_samples and opts.save_mask: + images.save_image(image_mask, p.outpath_samples, "", p.seeds[i], p.prompts[i], opts.samples_format, info=infotext(i), p=p, suffix="-mask") + if opts.return_mask: + output_images.append(image_mask) - if save_samples and opts.save_mask: - images.save_image(image_mask, p.outpath_samples, "", p.seeds[i], p.prompts[i], opts.samples_format, info=infotext(i), p=p, suffix="-mask") - - if save_samples and opts.save_mask_composite: - images.save_image(image_mask_composite, p.outpath_samples, "", p.seeds[i], p.prompts[i], opts.samples_format, info=infotext(i), p=p, suffix="-mask-composite") - - if opts.return_mask: - output_images.append(image_mask) - - if opts.return_mask_composite: - output_images.append(image_mask_composite) + if opts.return_mask_composite or opts.save_mask_composite: + image_mask_composite = Image.composite(image.convert('RGBA').convert('RGBa'), Image.new('RGBa', image.size), images.resize_image(2, p.mask_for_overlay, image.width, image.height).convert('L')).convert('RGBA') + if save_samples and opts.save_mask_composite: + images.save_image(image_mask_composite, p.outpath_samples, "", p.seeds[i], p.prompts[i], opts.samples_format, info=infotext(i), p=p, suffix="-mask-composite") + if opts.return_mask_composite: + output_images.append(image_mask_composite) del x_samples_ddim From d92ce145bba714c5b257b9853aa22681233651b8 Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Sun, 3 Dec 2023 16:50:20 +0200 Subject: [PATCH 193/342] Add import_hook hack to work around basicsr incompatibility Fixes #13985 --- modules/import_hook.py | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/modules/import_hook.py b/modules/import_hook.py index 28c67dfa8..eba9a3729 100644 --- a/modules/import_hook.py +++ b/modules/import_hook.py @@ -3,3 +3,14 @@ import sys # this will break any attempt to import xformers which will prevent stability diffusion repo from trying to use it if "--xformers" not in "".join(sys.argv): sys.modules["xformers"] = None + +# Hack to fix a changed import in torchvision 0.17+, which otherwise breaks +# basicsr; see https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/13985 +try: + import torchvision.transforms.functional_tensor # noqa: F401 +except ImportError: + try: + import torchvision.transforms.functional as functional + sys.modules["torchvision.transforms.functional_tensor"] = functional + except ImportError: + pass # shrug... From 28a2b5b4aab43424733039c31d910e8b8dd507cd Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Sun, 3 Dec 2023 14:20:20 -0700 Subject: [PATCH 194/342] Fixed a math mistake. --- modules/images.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/images.py b/modules/images.py index 6648097e8..949534986 100644 --- a/modules/images.py +++ b/modules/images.py @@ -969,7 +969,7 @@ def get_gaussian_kernel(stddev_radius=1.0, max_radius=2): x = coordinate[0] ** 2.0 + coordinate[1] ** 2.0 x = gaussian(x) x -= gauss_zero - x /= gauss_kernel_scale + x *= gauss_kernel_scale x = max(0.0, x) return x From 552f8bc832cd21ee0338e08b6a701687d0d79fad Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Sun, 3 Dec 2023 14:49:41 -0700 Subject: [PATCH 195/342] "Uncrop" the original denoised image for the composite step, fixing a "ValueError: Images do not match" *shudder* --- modules/processing.py | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/modules/processing.py b/modules/processing.py index 66aaab831..cd7216f83 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -994,6 +994,10 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: # we need to keep the original image around # and use it in the composite step. original_denoised_image = image.copy() + + if p.paste_to is not None: + original_denoised_image = uncrop(original_denoised_image, (p.overlay_images[i].width, p.overlay_images[i].height), p.paste_to) + image = apply_overlay(image, p.paste_to, i, p.overlay_images) if save_samples: From 639ccf254bd4d072f33333abb1ada3d08aaab470 Mon Sep 17 00:00:00 2001 From: illtellyoulater <3078931+illtellyoulater@users.noreply.github.com> Date: Mon, 4 Dec 2023 02:35:35 +0000 Subject: [PATCH 196/342] Update launch_utils.py to fix wrong dep. checks and reinstalls Fixes failing dependency checks for extensions having a different package name and import name (for example ffmpeg-python / ffmpeg), which currently is causing the unneeded reinstall of packages at runtime. In fact with current code, the same string is used when installing a package and when checking for its presence, as you can see in the following example: > launch_utils.run_pip("install ffmpeg-python", "required package") [ Installing required package: "ffmpeg-python" ... ] [ Installed ] > launch_utils.is_installed("ffmpeg-python") False ... which would actually return true with: > launch_utils.is_installed("ffmpeg") True --- modules/launch_utils.py | 14 ++++++++++---- 1 file changed, 10 insertions(+), 4 deletions(-) diff --git a/modules/launch_utils.py b/modules/launch_utils.py index 6e54d0636..6664c5e04 100644 --- a/modules/launch_utils.py +++ b/modules/launch_utils.py @@ -6,6 +6,7 @@ import os import shutil import sys import importlib.util +import importlib.metadata import platform import json from functools import lru_cache @@ -119,11 +120,16 @@ def run(command, desc=None, errdesc=None, custom_env=None, live: bool = default_ def is_installed(package): try: - spec = importlib.util.find_spec(package) - except ModuleNotFoundError: - return False + dist = importlib.metadata.distribution(package) + except importlib.metadata.PackageNotFoundError: + try: + spec = importlib.util.find_spec(package) + except ModuleNotFoundError: + return False - return spec is not None + return spec is not None + + return dist is not None def repo_dir(name): From 06725af40b94a146c56e693a47cbec6d0af55396 Mon Sep 17 00:00:00 2001 From: missionfloyd Date: Sun, 3 Dec 2023 21:26:12 -0700 Subject: [PATCH 197/342] Lint --- modules/launch_utils.py | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/modules/launch_utils.py b/modules/launch_utils.py index 6664c5e04..e71edd01d 100644 --- a/modules/launch_utils.py +++ b/modules/launch_utils.py @@ -120,12 +120,12 @@ def run(command, desc=None, errdesc=None, custom_env=None, live: bool = default_ def is_installed(package): try: - dist = importlib.metadata.distribution(package) + dist = importlib.metadata.distribution(package) except importlib.metadata.PackageNotFoundError: - try: + try: spec = importlib.util.find_spec(package) except ModuleNotFoundError: - return False + return False return spec is not None From 9e1f3feb12a7cfe4fd426dd3df5431c805746ecc Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Mon, 4 Dec 2023 09:15:19 +0300 Subject: [PATCH 198/342] make webui not crash when running with --disable-all-extensions option --- modules/models/diffusion/ddpm_edit.py | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/modules/models/diffusion/ddpm_edit.py b/modules/models/diffusion/ddpm_edit.py index b892d5fc7..6db340da4 100644 --- a/modules/models/diffusion/ddpm_edit.py +++ b/modules/models/diffusion/ddpm_edit.py @@ -24,10 +24,15 @@ from pytorch_lightning.utilities.distributed import rank_zero_only from ldm.util import log_txt_as_img, exists, default, ismap, isimage, mean_flat, count_params, instantiate_from_config from ldm.modules.ema import LitEma from ldm.modules.distributions.distributions import normal_kl, DiagonalGaussianDistribution -from ldm.models.autoencoder import VQModelInterface, IdentityFirstStage, AutoencoderKL +from ldm.models.autoencoder import IdentityFirstStage, AutoencoderKL from ldm.modules.diffusionmodules.util import make_beta_schedule, extract_into_tensor, noise_like from ldm.models.diffusion.ddim import DDIMSampler +try: + from ldm.models.autoencoder import VQModelInterface +except Exception: + class VQModelInterface: + pass __conditioning_keys__ = {'concat': 'c_concat', 'crossattn': 'c_crossattn', From 48fae7ccdc2fe2d2ba8e8cfcb17b56028734e570 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Mon, 4 Dec 2023 09:35:52 +0300 Subject: [PATCH 199/342] update changelog --- CHANGELOG.md | 162 +++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 162 insertions(+) diff --git a/CHANGELOG.md b/CHANGELOG.md index 2c72359fc..67429bbff 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -1,3 +1,165 @@ +## 1.7.0 + +### Features: +* settings tab rework: add search field, add categories, split UI settings page into many +* add altdiffusion-m18 support ([#13364](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13364)) +* support inference with LyCORIS GLora networks ([#13610](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13610)) +* add lora-embedding bundle system ([#13568](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13568)) +* option to move prompt from top row into generation parameters +* add support for SSD-1B ([#13865](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13865)) +* support inference with OFT networks ([#13692](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13692)) +* script metadata and DAG sorting mechanism ([#13944](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13944)) +* support HyperTile optimization ([#13948](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13948)) +* add support for SD 2.1 Turbo ([#14170](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14170)) +* remove Train->Preprocessing tab and put all its functionality into Extras tab +* initial IPEX support for Intel Arc GPU ([#14171](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14171)) + +### Minor: +* allow reading model hash from images in img2img batch mode ([#12767](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12767)) +* add option to align with sgm repo's sampling implementation ([#12818](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12818)) +* extra field for lora metadata viewer: `ss_output_name` ([#12838](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12838)) +* add action in settings page to calculate all SD checkpoint hashes ([#12909](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12909)) +* add button to copy prompt to style editor ([#12975](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12975)) +* add --skip-load-model-at-start option ([#13253](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13253)) +* write infotext to gif images +* read infotext from gif images ([#13068](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13068)) +* allow configuring the initial state of InputAccordion in ui-config.json ([#13189](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13189)) +* allow editing whitespace delimiters for ctrl+up/ctrl+down prompt editing ([#13444](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13444)) +* prevent accidentally closing popup dialogs ([#13480](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13480)) +* added option to play notification sound or not ([#13631](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13631)) +* show the preview image in the full screen image viewer if available ([#13459](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13459)) +* support for webui.settings.bat ([#13638](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13638)) +* add an option to not print stack traces on ctrl+c +* start/restart generation by Ctrl (Alt) + Enter ([#13644](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13644)) +* update prompts_from_file script to allow concatenating entries with the general prompt ([#13733](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13733)) +* added a visible checkbox to input accordion +* added an option to hide all txt2img/img2img parameters in an accordion ([#13826](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13826)) +* added 'Path' sorting option for Extra network cards ([#13968](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13968)) +* enable prompt hotkeys in style editor ([#13931](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13931)) +* option to show batch img2img results in UI ([#14009](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14009)) +* infotext updates: add option to disregard certain infotext fields, add option to not include VAE in infotext, add explanation to infotext settings page, move some options to infotext settings page +* add FP32 fallback support on sd_vae_approx ([#14046](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14046)) +* support XYZ scripts / split hires path from unet ([#14126](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14126)) +* allow use of mutiple styles csv files ([#14125](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14125)) + +### Extensions and API: +* update gradio to 3.41.2 +* support installed extensions list api ([#12774](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12774)) +* update pnginfo API to return dict with parsed values +* add noisy latent to `ExtraNoiseParams` for callback ([#12856](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12856)) +* show extension datetime in UTC ([#12864](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12864), [#12865](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12865), [#13281](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13281)) +* add an option to choose how to combine hires fix and refiner +* include program version in info response. ([#13135](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13135)) +* sd_unet support for SDXL +* patch DDPM.register_betas so that users can put given_betas in model yaml ([#13276](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13276)) +* xyz_grid: add prepare ([#13266](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13266)) +* allow multiple localization files with same language in extensions ([#13077](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13077)) +* add onEdit function for js and rework token-counter.js to use it +* fix the key error exception when processing override_settings keys ([#13567](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13567)) +* ability for extensions to return custom data via api in response.images ([#13463](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13463)) +* call state.jobnext() before postproces*() ([#13762](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13762)) +* add option to set notification sound volume ([#13884](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13884)) +* update Ruff to 0.1.6 ([#14059](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14059)) +* add Block component creation callback ([#14119](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14119)) +* catch uncaught exception with ui creation scripts ([#14120](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14120)) +* use extension name for determining an extension is installed in the index ([#14063](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14063)) +* update is_installed() from launch_utils.py to fix reinstalling already installed packages ([#14192](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14192)) + +### Bug Fixes: +* fix pix2pix producing bad results +* fix defaults settings page breaking when any of main UI tabs are hidden +* fix error that causes some extra networks to be disabled if both and are present in the prompt +* fix for Reload UI function: if you reload UI on one tab, other opened tabs will no longer stop working +* prevent duplicate resize handler ([#12795](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12795)) +* small typo: vae resolve bug ([#12797](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12797)) +* hide broken image crop tool ([#12792](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12792)) +* don't show hidden samplers in dropdown for XYZ script ([#12780](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12780)) +* fix style editing dialog breaking if it's opened in both img2img and txt2img tabs +* hide --gradio-auth and --api-auth values from /internal/sysinfo report +* add missing infotext for RNG in options ([#12819](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12819)) +* fix notification not playing when built-in webui tab is inactive ([#12834](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12834)) +* honor `--skip-install` for extension installers ([#12832](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12832)) +* don't print blank stdout in extension installers ([#12833](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12833), [#12855](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12855)) +* get progressbar to display correctly in extensions tab +* keep order in list of checkpoints when loading model that doesn't have a checksum +* fix inpainting models in txt2img creating black pictures +* fix generation params regex ([#12876](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12876)) +* fix batch img2img output dir with script ([#12926](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12926)) +* fix #13080 - Hypernetwork/TI preview generation ([#13084](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13084)) +* fix bug with sigma min/max overrides. ([#12995](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12995)) +* more accurate check for enabling cuDNN benchmark on 16XX cards ([#12924](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12924)) +* don't use multicond parser for negative prompt counter ([#13118](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13118)) +* fix data-sort-name containing spaces ([#13412](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13412)) +* update card on correct tab when editing metadata ([#13411](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13411)) +* fix viewing/editing metadata when filename contains an apostrophe ([#13395](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13395)) +* fix: --sd_model in "Prompts from file or textbox" script is not working ([#13302](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13302)) +* better Support for Portable Git ([#13231](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13231)) +* fix issues when webui_dir is not work_dir ([#13210](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13210)) +* fix: lora-bias-backup don't reset cache ([#13178](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13178)) +* account for customizable extra network separators whyen removing extra network text from the prompt ([#12877](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12877)) +* re fix batch img2img output dir with script ([#13170](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13170)) +* fix `--ckpt-dir` path separator and option use `short name` for checkpoint dropdown ([#13139](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13139)) +* consolidated allowed preview formats, Fix extra network `.gif` not woking as preview ([#13121](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13121)) +* fix venv_dir=- environment variable not working as expected on linux ([#13469](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13469)) +* repair unload sd checkpoint button +* edit-attention fixes ([#13533](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13533)) +* fix bug when using --gfpgan-models-path ([#13718](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13718)) +* properly apply sort order for extra network cards when selected from dropdown +* fixes generation restart not working for some users when 'Ctrl+Enter' is pressed ([#13962](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13962)) +* thread safe extra network list_items ([#13014](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13014)) +* fix not able to exit metadata popup when pop up is too big ([#14156](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14156)) +* fix auto focal point crop for opencv >= 4.8 ([#14121](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14121)) +* make 'use-cpu all' actually apply to 'all' ([#14131](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14131)) +* extras tab batch: actually use original filename +* make webui not crash when running with --disable-all-extensions option + +### Other: +* non-local condition ([#12814](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12814)) +* fix minor typos ([#12827](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12827)) +* remove xformers Python version check ([#12842](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12842)) +* style: file-metadata word-break ([#12837](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12837)) +* revert SGM noise multiplier change for img2img because it breaks hires fix +* do not change quicksettings dropdown option when value returned is `None` ([#12854](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12854)) +* [RC 1.6.0 - zoom is partly hidden] Update style.css ([#12839](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12839)) +* chore: change extension time format ([#12851](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12851)) +* WEBUI.SH - Use torch 2.1.0 release candidate for Navi 3 ([#12929](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12929)) +* add Fallback at images.read_info_from_image if exif data was invalid ([#13028](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13028)) +* update cmd arg description ([#12986](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12986)) +* fix: update shared.opts.data when add_option ([#12957](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12957), [#13213](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13213)) +* restore missing tooltips ([#12976](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12976)) +* use default dropdown padding on mobile ([#12880](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12880)) +* put enable console prompts option into settings from commandline args ([#13119](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13119)) +* fix some deprecated types ([#12846](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12846)) +* bump to torchsde==0.2.6 ([#13418](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13418)) +* update dragdrop.js ([#13372](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13372)) +* use orderdict as lru cache:opt/bug ([#13313](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13313)) +* XYZ if not include sub grids do not save sub grid ([#13282](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13282)) +* initialize state.time_start befroe state.job_count ([#13229](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13229)) +* fix fieldname regex ([#13458](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13458)) +* change denoising_strength default to None. ([#13466](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13466)) +* fix regression ([#13475](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13475)) +* fix IndexError ([#13630](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13630)) +* fix: checkpoints_loaded:{checkpoint:state_dict}, model.load_state_dict issue in dict value empty ([#13535](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13535)) +* update bug_report.yml ([#12991](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/12991)) +* requirements_versions httpx==0.24.1 ([#13839](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13839)) +* fix parenthesis auto selection ([#13829](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13829)) +* fix #13796 ([#13797](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13797)) +* corrected a typo in `modules/cmd_args.py` ([#13855](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13855)) +* feat: fix randn found element of type float at pos 2 ([#14004](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14004)) +* adds tqdm handler to logging_config.py for progress bar integration ([#13996](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13996)) +* hotfix: call shared.state.end() after postprocessing done ([#13977](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13977)) +* fix dependency address patch 1 ([#13929](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13929)) +* save sysinfo as .json ([#14035](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14035)) +* move exception_records related methods to errors.py ([#14084](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14084)) +* compatibility ([#13936](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13936)) +* json.dump(ensure_ascii=False) ([#14108](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14108)) +* dir buttons start with / so only the correct dir will be shown and no… ([#13957](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/13957)) +* alternate implementation for unet forward replacement that does not depend on hijack being applied +* re-add `keyedit_delimiters_whitespace` setting lost as part of commit e294e46 ([#14178](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14178)) +* fix `save_samples` being checked early when saving masked composite ([#14177](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14177)) +* slight optimization for mask and mask_composite ([#14181](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14181)) +* add import_hook hack to work around basicsr/torchvision incompatibility ([#14186](https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/14186)) + ## 1.6.1 ### Bug Fixes: From aaacf4823241450d88315af9d465d6815119fe0d Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Mon, 4 Dec 2023 01:27:22 -0700 Subject: [PATCH 200/342] Organized the settings and UI of soft inpainting to allow for toggling the feature, and centralizes default values to reduce the amount of copy-pasta. --- modules/img2img.py | 14 +-- modules/processing.py | 5 +- modules/sd_samplers_cfg_denoiser.py | 35 +++++--- modules/sd_samplers_common.py | 4 +- modules/soft_inpainting.py | 133 ++++++++++++++++++++++++++++ modules/ui.py | 17 ++-- scripts/outpainting_mk_2.py | 15 ++-- scripts/poor_mans_outpainting.py | 15 ++-- test/test_img2img.py | 8 +- 9 files changed, 197 insertions(+), 49 deletions(-) create mode 100644 modules/soft_inpainting.py diff --git a/modules/img2img.py b/modules/img2img.py index 596f741c1..3aa8a9cef 100644 --- a/modules/img2img.py +++ b/modules/img2img.py @@ -15,6 +15,7 @@ import modules.shared as shared import modules.processing as processing from modules.ui import plaintext_to_html import modules.scripts +import modules.soft_inpainting as si def process_batch(p, input_dir, output_dir, inpaint_mask_dir, args, to_scale=False, scale_by=1.0, use_png_info=False, png_info_props=None, png_info_dir=None): @@ -162,6 +163,7 @@ def img2img(id_task: str, sampler_name: str, mask_blur: int, mask_alpha: float, + mask_blend_enabled: bool, mask_blend_power: float, mask_blend_scale: float, inpaint_detail_preservation: float, @@ -227,6 +229,9 @@ def img2img(id_task: str, assert 0. <= denoising_strength <= 1., 'can only work with strength in [0.0, 1.0]' + soft_inpainting = si.SoftInpaintingSettings(mask_blend_power, mask_blend_scale, inpaint_detail_preservation) \ + if mask_blend_enabled else None + p = StableDiffusionProcessingImg2Img( sd_model=shared.sd_model, outpath_samples=opts.outdir_samples or opts.outdir_img2img_samples, @@ -244,9 +249,7 @@ def img2img(id_task: str, init_images=[image], mask=mask, mask_blur=mask_blur, - mask_blend_power=mask_blend_power, - mask_blend_scale=mask_blend_scale, - inpaint_detail_preservation=inpaint_detail_preservation, + soft_inpainting=soft_inpainting, inpainting_fill=inpainting_fill, resize_mode=resize_mode, denoising_strength=denoising_strength, @@ -267,9 +270,8 @@ def img2img(id_task: str, if mask: p.extra_generation_params["Mask blur"] = mask_blur - p.extra_generation_params["Mask blending bias"] = mask_blend_power - p.extra_generation_params["Mask blending preservation"] = mask_blend_scale - p.extra_generation_params["Mask blending contrast boost"] = inpaint_detail_preservation + if soft_inpainting is not None: + soft_inpainting.add_generation_params(p.extra_generation_params) with closing(p): if is_batch: diff --git a/modules/processing.py b/modules/processing.py index cd7216f83..b209c84a3 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -30,6 +30,7 @@ import modules.sd_models as sd_models import modules.sd_vae as sd_vae from ldm.data.util import AddMiDaS from ldm.models.diffusion.ddpm import LatentDepth2ImageDiffusion +import modules.soft_inpainting as si from einops import repeat, rearrange from blendmodes.blend import blendLayers, BlendType @@ -1425,9 +1426,7 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): mask_blur_x: int = 4 mask_blur_y: int = 4 mask_blur: int = None - mask_blend_power: float = 1 - mask_blend_scale: float = 0.5 - inpaint_detail_preservation: float = 4 + soft_inpainting: si.SoftInpaintingParameters = si.default inpainting_fill: int = 0 inpaint_full_res: bool = True inpaint_full_res_padding: int = 0 diff --git a/modules/sd_samplers_cfg_denoiser.py b/modules/sd_samplers_cfg_denoiser.py index efbe7a403..0ee0b7dde 100644 --- a/modules/sd_samplers_cfg_denoiser.py +++ b/modules/sd_samplers_cfg_denoiser.py @@ -6,6 +6,7 @@ import modules.shared as shared from modules.script_callbacks import CFGDenoiserParams, cfg_denoiser_callback from modules.script_callbacks import CFGDenoisedParams, cfg_denoised_callback from modules.script_callbacks import AfterCFGCallbackParams, cfg_after_cfg_callback +import modules.soft_inpainting as si def catenate_conds(conds): @@ -43,9 +44,7 @@ class CFGDenoiser(torch.nn.Module): self.model_wrap = None self.mask = None self.nmask = None - self.mask_blend_power = 1 - self.mask_blend_scale = 0.5 - self.inpaint_detail_preservation = 4 + self.soft_inpainting: si.SoftInpaintingParameters = None self.init_latent = None self.steps = None """number of steps as specified by user in UI""" @@ -95,7 +94,8 @@ class CFGDenoiser(torch.nn.Module): self.sampler.sampler_extra_args['uncond'] = uc def forward(self, x, sigma, uncond, cond, cond_scale, s_min_uncond, image_cond): - def latent_blend(a, b, t): + def latent_blend(a, b, t, one_minus_t=None): + """ Interpolates two latent image representations according to the parameter t, where the interpolated vectors' magnitudes are also interpolated separately. @@ -104,7 +104,11 @@ class CFGDenoiser(torch.nn.Module): """ # NOTE: We use inplace operations wherever possible. - one_minus_t = 1 - t + if one_minus_t is None: + one_minus_t = 1 - t + + if self.soft_inpainting is None: + return a * one_minus_t + b * t # Linearly interpolate the image vectors. a_scaled = a * one_minus_t @@ -119,10 +123,10 @@ class CFGDenoiser(torch.nn.Module): current_magnitude = torch.norm(image_interp, p=2, dim=1).to(torch.float64).add_(0.00001) # Interpolate the powered magnitudes, then un-power them (bring them back to a power of 1). - a_magnitude = torch.norm(a, p=2, dim=1).to(torch.float64).pow_(self.inpaint_detail_preservation) * one_minus_t - b_magnitude = torch.norm(b, p=2, dim=1).to(torch.float64).pow_(self.inpaint_detail_preservation) * t + a_magnitude = torch.norm(a, p=2, dim=1).to(torch.float64).pow_(self.soft_inpainting.inpaint_detail_preservation) * one_minus_t + b_magnitude = torch.norm(b, p=2, dim=1).to(torch.float64).pow_(self.soft_inpainting.inpaint_detail_preservation) * t desired_magnitude = a_magnitude - desired_magnitude.add_(b_magnitude).pow_(1 / self.inpaint_detail_preservation) + desired_magnitude.add_(b_magnitude).pow_(1 / self.soft_inpainting.inpaint_detail_preservation) del a_magnitude, b_magnitude, one_minus_t # Change the linearly interpolated image vectors' magnitudes to the value we want. @@ -156,7 +160,10 @@ class CFGDenoiser(torch.nn.Module): NOTE: "mask" is not used """ - return torch.pow(nmask, (_sigma ** self.mask_blend_power) * self.mask_blend_scale) + if self.soft_inpainting is None: + return nmask + + return torch.pow(nmask, (_sigma ** self.soft_inpainting.mask_blend_power) * self.soft_inpainting.mask_blend_scale) if state.interrupted or state.skipped: raise sd_samplers_common.InterruptedException @@ -176,7 +183,10 @@ class CFGDenoiser(torch.nn.Module): # Blend in the original latents (before) if self.mask_before_denoising and self.mask is not None: - x = latent_blend(self.init_latent, x, get_modified_nmask(self.nmask, sigma)) + if self.soft_inpainting is None: + x = latent_blend(self.init_latent, x, self.nmask, self.mask) + else: + x = latent_blend(self.init_latent, x, get_modified_nmask(self.nmask, sigma)) batch_size = len(conds_list) repeats = [len(conds_list[i]) for i in range(batch_size)] @@ -279,7 +289,10 @@ class CFGDenoiser(torch.nn.Module): # Blend in the original latents (after) if not self.mask_before_denoising and self.mask is not None: - denoised = latent_blend(self.init_latent, denoised, get_modified_nmask(self.nmask, sigma)) + if self.soft_inpainting is None: + denoised = latent_blend(self.init_latent, denoised, self.nmask, self.mask) + else: + denoised = latent_blend(self.init_latent, denoised, get_modified_nmask(self.nmask, sigma)) self.sampler.last_latent = self.get_pred_x0(torch.cat([x_in[i:i + 1] for i in denoised_image_indexes]), torch.cat([x_out[i:i + 1] for i in denoised_image_indexes]), sigma) diff --git a/modules/sd_samplers_common.py b/modules/sd_samplers_common.py index ecd8ab0a0..9682bee3d 100644 --- a/modules/sd_samplers_common.py +++ b/modules/sd_samplers_common.py @@ -277,9 +277,7 @@ class Sampler: self.model_wrap_cfg.p = p self.model_wrap_cfg.mask = p.mask if hasattr(p, 'mask') else None self.model_wrap_cfg.nmask = p.nmask if hasattr(p, 'nmask') else None - self.model_wrap_cfg.mask_blend_power = p.mask_blend_power if hasattr(p, 'mask_blend_power') else None - self.model_wrap_cfg.mask_blend_scale = p.mask_blend_scale if hasattr(p, 'mask_blend_scale') else None - self.model_wrap_cfg.inpaint_detail_preservation = p.inpaint_detail_preservation if hasattr(p, 'inpaint_detail_preservation') else None + self.model_wrap_cfg.soft_inpainting = p.soft_inpainting if hasattr(p, 'soft_inpainting') else None self.model_wrap_cfg.step = 0 self.model_wrap_cfg.image_cfg_scale = getattr(p, 'image_cfg_scale', None) self.eta = p.eta if p.eta is not None else getattr(opts, self.eta_option_field, 0.0) diff --git a/modules/soft_inpainting.py b/modules/soft_inpainting.py new file mode 100644 index 000000000..259c36ec8 --- /dev/null +++ b/modules/soft_inpainting.py @@ -0,0 +1,133 @@ +class SoftInpaintingSettings: + def __init__(self, mask_blend_power, mask_blend_scale, inpaint_detail_preservation): + self.mask_blend_power = mask_blend_power + self.mask_blend_scale = mask_blend_scale + self.inpaint_detail_preservation = inpaint_detail_preservation + + def get_paste_fields(self): + return [ + (self.mask_blend_power, gen_param_labels.mask_blend_power), + (self.mask_blend_scale, gen_param_labels.mask_blend_scale), + (self.inpaint_detail_preservation, gen_param_labels.inpaint_detail_preservation), + ] + + def add_generation_params(self, dest): + dest[enabled_gen_param_label] = True + dest[gen_param_labels.mask_blend_power] = self.mask_blend_power + dest[gen_param_labels.mask_blend_scale] = self.mask_blend_scale + dest[gen_param_labels.inpaint_detail_preservation] = self.inpaint_detail_preservation + + +enabled_ui_label = "Soft inpainting" +enabled_gen_param_label = "Soft inpainting enabled" +enabled_el_id = "soft_inpainting_enabled" + +default = SoftInpaintingSettings(1, 0.5, 4) +ui_labels = SoftInpaintingSettings("Schedule bias", "Preservation strength", "Transition contrast boost") + +ui_info = SoftInpaintingSettings( + mask_blend_power="Shifts when preservation of original content occurs during denoising.", + # "Below 1: Stronger preservation near the end (with low sigma)\n" + # "1: Balanced (proportional to sigma)\n" + # "Above 1: Stronger preservation in the beginning (with high sigma)", + mask_blend_scale="How strongly partially masked content should be preserved.", + # "Low values: Favors generated content.\n" + # "High values: Favors original content.", + inpaint_detail_preservation="Amplifies the contrast that may be lost in partially masked regions.") + +gen_param_labels = SoftInpaintingSettings("Soft inpainting schedule bias", "Soft inpainting preservation strength", "Soft inpainting transition contrast boost") +el_ids = SoftInpaintingSettings("mask_blend_power", "mask_blend_scale", "inpaint_detail_preservation") + + +def gradio_ui(): + import gradio as gr + from modules.ui_components import InputAccordion + """ + with InputAccordion(False, label="Refiner", elem_id=self.elem_id("enable")) as enable_refiner: + with gr.Row(): + refiner_checkpoint = gr.Dropdown(label='Checkpoint', elem_id=self.elem_id("checkpoint"), choices=sd_models.checkpoint_tiles(), value='', tooltip="switch to another model in the middle of generation") + create_refresh_button(refiner_checkpoint, sd_models.list_models, lambda: {"choices": sd_models.checkpoint_tiles()}, self.elem_id("checkpoint_refresh")) + + refiner_switch_at = gr.Slider(value=0.8, label="Switch at", minimum=0.01, maximum=1.0, step=0.01, elem_id=self.elem_id("switch_at"), tooltip="fraction of sampling steps when the switch to refiner model should happen; 1=never, 0.5=switch in the middle of generation") + + """ + with InputAccordion(False, label=enabled_ui_label, elem_id=enabled_el_id) as soft_inpainting_enabled: + with gr.Group(): + gr.Markdown( + """ + Soft inpainting allows you to **seamlessly blend original content with inpainted content** according to the mask opacity. + **High _Mask blur_** values are recommended! + """) + + result = SoftInpaintingSettings( + gr.Slider(label=ui_labels.mask_blend_power, + info=ui_info.mask_blend_power, + minimum=0, + maximum=8, + step=0.1, + value=default.mask_blend_power, + elem_id=el_ids.mask_blend_power), + gr.Slider(label=ui_labels.mask_blend_scale, + info=ui_info.mask_blend_scale, + minimum=0, + maximum=8, + step=0.05, + value=default.mask_blend_scale, + elem_id=el_ids.mask_blend_scale), + gr.Slider(label=ui_labels.inpaint_detail_preservation, + info=ui_info.inpaint_detail_preservation, + minimum=1, + maximum=32, + step=0.5, + value=default.inpaint_detail_preservation, + elem_id=el_ids.inpaint_detail_preservation)) + + with gr.Accordion("Help", open=False): + gr.Markdown( + f""" + ### {ui_labels.mask_blend_power} + + The blending strength of original content is scaled proportionally with the decreasing noise level values at each step (sigmas). + This ensures that the influence of the denoiser and original content preservation is roughly balanced at each step. + This balance can be shifted using this parameter, controlling whether earlier or later steps have stronger preservation. + + - **Below 1**: Stronger preservation near the end (with low sigma) + - **1**: Balanced (proportional to sigma) + - **Above 1**: Stronger preservation in the beginning (with high sigma) + """) + gr.Markdown( + f""" + ### {ui_labels.mask_blend_scale} + + Skews whether partially masked image regions should be more likely to preserve the original content or favor inpainted content. + This may need to be adjusted depending on the {ui_labels.mask_blend_power}, CFG Scale, prompt and Denoising strength. + + - **Low values**: Favors generated content. + - **High values**: Favors original content. + """) + gr.Markdown( + f""" + ### {ui_labels.inpaint_detail_preservation} + + This parameter controls how the original latent vectors and denoised latent vectors are interpolated. + With higher values, the magnitude of the resulting blended vector will be closer to the maximum of the two interpolated vectors. + This can prevent the loss of contrast that occurs with linear interpolation. + + - **Low values**: Softer blending, details may fade. + - **High values**: Stronger contrast, may over-saturate colors. + """) + + return ( + [ + soft_inpainting_enabled, + result.mask_blend_power, + result.mask_blend_scale, + result.inpaint_detail_preservation + ], + [ + (soft_inpainting_enabled, enabled_gen_param_label), + (result.mask_blend_power, gen_param_labels.mask_blend_power), + (result.mask_blend_scale, gen_param_labels.mask_blend_scale), + (result.inpaint_detail_preservation, gen_param_labels.inpaint_detail_preservation) + ] + ) diff --git a/modules/ui.py b/modules/ui.py index b13ed66cb..0e4fb17aa 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -29,6 +29,7 @@ import modules.shared as shared from modules import prompt_parser from modules.sd_hijack import model_hijack from modules.generation_parameters_copypaste import image_from_url_text +import modules.soft_inpainting as si create_setting_component = ui_settings.create_setting_component @@ -678,9 +679,16 @@ def create_ui(): with FormRow(): mask_blur = gr.Slider(label='Mask blur', minimum=0, maximum=64, step=1, value=4, elem_id="img2img_mask_blur") mask_alpha = gr.Slider(label="Mask transparency", visible=False, elem_id="img2img_mask_alpha") + + with FormRow(): + soft_inpainting = si.gradio_ui() + + + """ mask_blend_power = gr.Slider(label='Blending bias', minimum=0, maximum=8, step=0.1, value=1, elem_id="img2img_mask_blend_power") mask_blend_scale = gr.Slider(label='Blending preservation', minimum=0, maximum=8, step=0.05, value=0.5, elem_id="img2img_mask_blend_scale") inpaint_detail_preservation = gr.Slider(label='Blending contrast boost', minimum=1, maximum=32, step=0.5, value=4, elem_id="img2img_mask_blend_offset") + """ with FormRow(): inpainting_mask_invert = gr.Radio(label='Mask mode', choices=['Inpaint masked', 'Inpaint not masked'], value='Inpaint masked', type="index", elem_id="img2img_mask_mode") @@ -736,9 +744,7 @@ def create_ui(): sampler_name, mask_blur, mask_alpha, - mask_blend_power, - mask_blend_scale, - inpaint_detail_preservation, + *(soft_inpainting[0]), inpainting_fill, batch_count, batch_size, @@ -837,11 +843,10 @@ def create_ui(): (toprow.ui_styles.dropdown, lambda d: d["Styles array"] if isinstance(d.get("Styles array"), list) else gr.update()), (denoising_strength, "Denoising strength"), (mask_blur, "Mask blur"), - (mask_blend_power, "Mask blending bias"), - (mask_blend_scale, "Mask blending preservation"), - (inpaint_detail_preservation, "Mask blending contrast boost"), + *(soft_inpainting[1]), *scripts.scripts_img2img.infotext_fields ] + parameters_copypaste.add_paste_fields("img2img", init_img, img2img_paste_fields, override_settings) parameters_copypaste.add_paste_fields("inpaint", init_img_with_mask, img2img_paste_fields, override_settings) parameters_copypaste.register_paste_params_button(parameters_copypaste.ParamBinding( diff --git a/scripts/outpainting_mk_2.py b/scripts/outpainting_mk_2.py index bd9cb61bf..f78886883 100644 --- a/scripts/outpainting_mk_2.py +++ b/scripts/outpainting_mk_2.py @@ -10,6 +10,7 @@ from PIL import Image, ImageDraw from modules import images from modules.processing import Processed, process_images from modules.shared import opts, state +import modules.soft_inpainting as si # this function is taken from https://github.com/parlance-zz/g-diffuser-bot @@ -133,16 +134,14 @@ class Script(scripts.Script): pixels = gr.Slider(label="Pixels to expand", minimum=8, maximum=256, step=8, value=128, elem_id=self.elem_id("pixels")) mask_blur = gr.Slider(label='Mask blur', minimum=0, maximum=64, step=1, value=8, elem_id=self.elem_id("mask_blur")) - mask_blend_power = gr.Slider(label='Blending bias', minimum=0, maximum=8, step=0.1, value=1, elem_id=self.elem_id("mask_blend_power")) - mask_blend_scale = gr.Slider(label='Blending preservation', minimum=0, maximum=8, step=0.05, value=0.5, elem_id=self.elem_id("mask_blend_scale")) - inpaint_detail_preservation = gr.Slider(label='Blending contrast boost', minimum=1, maximum=32, step=0.5, value=4, elem_id=self.elem_id("inpaint_detail_preservation")) + soft_inpainting = si.gradio_ui()[0] direction = gr.CheckboxGroup(label="Outpainting direction", choices=['left', 'right', 'up', 'down'], value=['left', 'right', 'up', 'down'], elem_id=self.elem_id("direction")) noise_q = gr.Slider(label="Fall-off exponent (lower=higher detail)", minimum=0.0, maximum=4.0, step=0.01, value=1.0, elem_id=self.elem_id("noise_q")) color_variation = gr.Slider(label="Color variation", minimum=0.0, maximum=1.0, step=0.01, value=0.05, elem_id=self.elem_id("color_variation")) - return [info, pixels, mask_blur, mask_blend_power, mask_blend_scale, inpaint_detail_preservation, direction, noise_q, color_variation] + return [info, pixels, mask_blur, *soft_inpainting, direction, noise_q, color_variation] - def run(self, p, _, pixels, mask_blur, mask_blend_power, mask_blend_scale, inpaint_detail_preservation, direction, noise_q, color_variation): + def run(self, p, _, pixels, mask_blur, mask_blend_enabled, mask_blend_power, mask_blend_scale, inpaint_detail_preservation, direction, noise_q, color_variation): initial_seed_and_info = [None, None] process_width = p.width @@ -170,9 +169,9 @@ class Script(scripts.Script): p.mask_blur_x = mask_blur_x*4 p.mask_blur_y = mask_blur_y*4 - p.mask_blend_power = mask_blend_power - p.mask_blend_scale = mask_blend_scale - p.inpaint_detail_preservation = inpaint_detail_preservation + + p.soft_inpainting = si.SoftInpaintingSettings(mask_blend_power, mask_blend_scale, inpaint_detail_preservation) \ + if mask_blend_enabled else None init_img = p.init_images[0] target_w = math.ceil((init_img.width + left + right) / 64) * 64 diff --git a/scripts/poor_mans_outpainting.py b/scripts/poor_mans_outpainting.py index 5388f5db4..11f7f74a8 100644 --- a/scripts/poor_mans_outpainting.py +++ b/scripts/poor_mans_outpainting.py @@ -7,6 +7,7 @@ from PIL import Image, ImageDraw from modules import images, devices from modules.processing import Processed, process_images from modules.shared import opts, state +import modules.soft_inpainting as si class Script(scripts.Script): @@ -22,23 +23,19 @@ class Script(scripts.Script): pixels = gr.Slider(label="Pixels to expand", minimum=8, maximum=256, step=8, value=128, elem_id=self.elem_id("pixels")) mask_blur = gr.Slider(label='Mask blur', minimum=0, maximum=64, step=1, value=4, elem_id=self.elem_id("mask_blur")) - mask_blend_power = gr.Slider(label='Blending bias', minimum=0, maximum=8, step=0.1, value=1, elem_id=self.elem_id("mask_blend_power")) - mask_blend_scale = gr.Slider(label='Blending preservation', minimum=0, maximum=8, step=0.05, value=0.5, elem_id=self.elem_id("mask_blend_scale")) - inpaint_detail_preservation = gr.Slider(label='Blending contrast boost', minimum=1, maximum=32, step=0.5, value=4, elem_id=self.elem_id("inpaint_detail_preservation")) + soft_inpainting = si.gradio_ui()[0] inpainting_fill = gr.Radio(label='Masked content', choices=['fill', 'original', 'latent noise', 'latent nothing'], value='fill', type="index", elem_id=self.elem_id("inpainting_fill")) direction = gr.CheckboxGroup(label="Outpainting direction", choices=['left', 'right', 'up', 'down'], value=['left', 'right', 'up', 'down'], elem_id=self.elem_id("direction")) - return [pixels, mask_blur, mask_blend_power, mask_blend_scale, inpaint_detail_preservation, inpainting_fill, direction] + return [pixels, mask_blur, *soft_inpainting, inpainting_fill, direction] - def run(self, p, pixels, mask_blur, mask_blend_power, mask_blend_scale, inpaint_detail_preservation, inpainting_fill, direction): + def run(self, p, pixels, mask_blur, mask_blend_enabled, mask_blend_power, mask_blend_scale, inpaint_detail_preservation, inpainting_fill, direction): initial_seed = None initial_info = None p.mask_blur = mask_blur * 2 - p.mask_blend_power = mask_blend_power - p.mask_blend_scale = mask_blend_scale - p.inpaint_detail_preservation = inpaint_detail_preservation - + p.soft_inpainting = si.SoftInpaintingSettings(mask_blend_power, mask_blend_scale, inpaint_detail_preservation) \ + if mask_blend_enabled else None p.inpainting_fill = inpainting_fill p.inpaint_full_res = False diff --git a/test/test_img2img.py b/test/test_img2img.py index 5cda2dbae..87bd85091 100644 --- a/test/test_img2img.py +++ b/test/test_img2img.py @@ -1,6 +1,7 @@ import pytest import requests +import modules.soft_inpainting as si @pytest.fixture() @@ -24,9 +25,10 @@ def simple_img2img_request(img2img_basic_image_base64): "inpainting_mask_invert": False, "mask": None, "mask_blur": 4, - "mask_blend_power": 1, - "mask_blend_scale": 0.5, - "inpaint_detail_preservation": 4, + "mask_blend_enabled": True, + "mask_blend_power": si.default.mask_blend_power, + "mask_blend_scale": si.default.mask_blend_scale, + "inpaint_detail_preservation": si.default.inpaint_detail_preservation, "n_iter": 1, "negative_prompt": "", "override_settings": {}, From 259d33c3c8e27557cb9bab9b3a1dd7fc7450d16c Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Mon, 4 Dec 2023 01:57:21 -0700 Subject: [PATCH 201/342] Enables the original functionality to be toggled on and off. --- modules/processing.py | 91 +++++++++++++++++++++++++++++++------------ 1 file changed, 66 insertions(+), 25 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index b209c84a3..b40b1a40d 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -88,9 +88,12 @@ def apply_overlay(image, paste_loc, index, overlays): return image -def create_binary_mask(image): +def create_binary_mask(image, round=True): if image.mode == 'RGBA' and image.getextrema()[-1] != (255, 255): - image = image.split()[-1].convert("L") + if round: + image = image.split()[-1].convert("L").point(lambda x: 255 if x > 128 else 0) + else: + image = image.split()[-1].convert("L") else: image = image.convert('L') return image @@ -316,7 +319,7 @@ class StableDiffusionProcessing: c_adm = torch.cat((c_adm, noise_level_emb), 1) return c_adm - def inpainting_image_conditioning(self, source_image, latent_image, image_mask=None): + def inpainting_image_conditioning(self, source_image, latent_image, image_mask=None, round_image_mask=True): self.is_using_inpainting_conditioning = True # Handle the different mask inputs @@ -327,6 +330,11 @@ class StableDiffusionProcessing: conditioning_mask = np.array(image_mask.convert("L")) conditioning_mask = conditioning_mask.astype(np.float32) / 255.0 conditioning_mask = torch.from_numpy(conditioning_mask[None, None]) + + if round_image_mask: + # Caller is requesting a discretized mask as input, so we round to either 1.0 or 0.0 + conditioning_mask = torch.round(conditioning_mask) + else: conditioning_mask = source_image.new_ones(1, 1, *source_image.shape[-2:]) @@ -350,7 +358,7 @@ class StableDiffusionProcessing: return image_conditioning - def img2img_image_conditioning(self, source_image, latent_image, image_mask=None): + def img2img_image_conditioning(self, source_image, latent_image, image_mask=None, round_image_mask=True): source_image = devices.cond_cast_float(source_image) # HACK: Using introspection as the Depth2Image model doesn't appear to uniquely @@ -362,7 +370,10 @@ class StableDiffusionProcessing: return self.edit_image_conditioning(source_image) if self.sampler.conditioning_key in {'hybrid', 'concat'}: - return self.inpainting_image_conditioning(source_image, latent_image, image_mask=image_mask) + return self.inpainting_image_conditioning(source_image, + latent_image, + image_mask=image_mask, + round_image_mask=round_image_mask) if self.sampler.conditioning_key == "crossattn-adm": return self.unclip_image_conditioning(source_image) @@ -878,8 +889,9 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: else: if opts.sd_vae_decode_method != 'Full': p.extra_generation_params['VAE Decoder'] = opts.sd_vae_decode_method + # Generate the mask(s) based on similarity between the original and denoised latent vectors - if getattr(p, "image_mask", None) is not None: + if getattr(p, "image_mask", None) is not None and getattr(p, "soft_inpainting", None) is not None: # latent_mask = p.nmask[0].float().cpu() # convert the original mask into a form we use to scale distances for thresholding @@ -911,7 +923,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: converted_mask = converted_mask.astype(np.uint8) converted_mask = Image.fromarray(converted_mask) converted_mask = images.resize_image(2, converted_mask, p.width, p.height) - converted_mask = create_binary_mask(converted_mask) + converted_mask = create_binary_mask(converted_mask, round=False) # Remove aliasing artifacts using a gaussian blur. converted_mask = converted_mask.filter(ImageFilter.GaussianBlur(radius=4)) @@ -1010,23 +1022,33 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: if opts.enable_pnginfo: image.info["parameters"] = text output_images.append(image) - if save_samples and hasattr(p, 'masks_for_overlay') and p.masks_for_overlay and any([opts.save_mask, opts.save_mask_composite, opts.return_mask, opts.return_mask_composite]): - image_mask = p.masks_for_overlay[i].convert('RGB') - image_mask_composite = Image.composite( - original_denoised_image.convert('RGBA').convert('RGBa'), Image.new('RGBa', image.size), - images.resize_image(2, p.masks_for_overlay[i], image.width, image.height).convert('L')).convert('RGBA') + if save_samples and any([opts.save_mask, opts.save_mask_composite, opts.return_mask, opts.return_mask_composite]): + if hasattr(p, 'masks_for_overlay') and p.masks_for_overlay: + image_mask = p.masks_for_overlay[i].convert('RGB') + image_mask_composite = Image.composite( + original_denoised_image.convert('RGBA').convert('RGBa'), Image.new('RGBa', image.size), + images.resize_image(2, p.masks_for_overlay[i], image.width, image.height).convert('L')).convert('RGBA') + elif hasattr(p, 'mask_for_overlay') and p.mask_for_overlay: + image_mask = p.mask_for_overlay.convert('RGB') + image_mask_composite = Image.composite( + original_denoised_image.convert('RGBA').convert('RGBa'), Image.new('RGBa', image.size), + images.resize_image(2, p.mask_for_overlay, image.width, image.height).convert('L')).convert('RGBA') + else: + image_mask = None + image_mask_composite = None - if opts.save_mask: - images.save_image(image_mask, p.outpath_samples, "", p.seeds[i], p.prompts[i], opts.samples_format, info=infotext(i), p=p, suffix="-mask") + if image_mask is not None and image_mask_composite is not None: + if opts.save_mask: + images.save_image(image_mask, p.outpath_samples, "", p.seeds[i], p.prompts[i], opts.samples_format, info=infotext(i), p=p, suffix="-mask") - if opts.save_mask_composite: - images.save_image(image_mask_composite, p.outpath_samples, "", p.seeds[i], p.prompts[i], opts.samples_format, info=infotext(i), p=p, suffix="-mask-composite") + if opts.save_mask_composite: + images.save_image(image_mask_composite, p.outpath_samples, "", p.seeds[i], p.prompts[i], opts.samples_format, info=infotext(i), p=p, suffix="-mask-composite") - if opts.return_mask: - output_images.append(image_mask) + if opts.return_mask: + output_images.append(image_mask) - if opts.return_mask_composite: - output_images.append(image_mask_composite) + if opts.return_mask_composite: + output_images.append(image_mask_composite) del x_samples_ddim @@ -1439,6 +1461,7 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): nmask: torch.Tensor = field(default=None, init=False) image_conditioning: torch.Tensor = field(default=None, init=False) init_img_hash: str = field(default=None, init=False) + mask_for_overlay: Image = field(default=None, init=False) init_latent: torch.Tensor = field(default=None, init=False) def __post_init__(self): @@ -1471,7 +1494,7 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): if image_mask is not None: # image_mask is passed in as RGBA by Gradio to support alpha masks, # but we still want to support binary masks. - image_mask = create_binary_mask(image_mask) + image_mask = create_binary_mask(image_mask, round=(self.soft_inpainting is None)) if self.inpainting_mask_invert: image_mask = ImageOps.invert(image_mask) @@ -1489,6 +1512,7 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): image_mask = Image.fromarray(np_mask) if self.inpaint_full_res: + self.mask_for_overlay = image_mask if self.soft_inpainting is None else None mask = image_mask.convert('L') crop_region = masking.get_crop_region(np.array(mask), self.inpaint_full_res_padding) crop_region = masking.expand_crop_region(crop_region, self.width, self.height, mask.width, mask.height) @@ -1500,7 +1524,12 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): else: image_mask = images.resize_image(self.resize_mode, image_mask, self.width, self.height) - self.masks_for_overlay = [] + if self.soft_inpainting is None: + np_mask = np.array(image_mask) + np_mask = np.clip((np_mask.astype(np.float32)) * 2, 0, 255).astype(np.uint8) + self.mask_for_overlay = Image.fromarray(np_mask) + + self.masks_for_overlay = [] if self.soft_inpainting is not None else None self.overlay_images = [] latent_mask = self.latent_mask if self.latent_mask is not None else image_mask @@ -1522,8 +1551,15 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): image = images.resize_image(self.resize_mode, image, self.width, self.height) if image_mask is not None: - self.overlay_images.append(image) - self.masks_for_overlay.append(image_mask) + if self.soft_inpainting is not None: + # We apply the masks AFTER to adjust mask based on changed content. + self.overlay_images.append(image) + self.masks_for_overlay.append(image_mask) + else: + image_masked = Image.new('RGBa', (image.width, image.height)) + image_masked.paste(image.convert("RGBA").convert("RGBa"), + mask=ImageOps.invert(self.mask_for_overlay.convert('L'))) + self.overlay_images.append(image_masked.convert('RGBA')) # crop_region is not None if we are doing inpaint full res if crop_region is not None: @@ -1576,6 +1612,8 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): latmask = init_mask.convert('RGB').resize((self.init_latent.shape[3], self.init_latent.shape[2])) latmask = np.moveaxis(np.array(latmask, dtype=np.float32), 2, 0) / 255 latmask = latmask[0] + if self.soft_inpainting is None: + latmask = np.around(latmask) latmask = np.tile(latmask[None], (4, 1, 1)) self.mask = torch.asarray(1.0 - latmask).to(shared.device).type(self.sd_model.dtype) @@ -1587,7 +1625,10 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): elif self.inpainting_fill == 3: self.init_latent = self.init_latent * self.mask - self.image_conditioning = self.img2img_image_conditioning(image * 2 - 1, self.init_latent, image_mask) + self.image_conditioning = self.img2img_image_conditioning(image * 2 - 1, + self.init_latent, + image_mask, + self.soft_inpainting is None) def sample(self, conditioning, unconditional_conditioning, seeds, subseeds, subseed_strength, prompts): x = self.rng.next() From 15322e1b1a9e31edcc2f7d72a32d02365058737d Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Mon, 4 Dec 2023 12:36:41 +0300 Subject: [PATCH 202/342] repair old handler for postprocessing API --- modules/postprocessing.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/postprocessing.py b/modules/postprocessing.py index 3c85a74c1..d166f859b 100644 --- a/modules/postprocessing.py +++ b/modules/postprocessing.py @@ -153,4 +153,4 @@ def run_extras(extras_mode, resize_mode, image, image_folder, input_dir, output_ }, }) - return run_postprocessing(extras_mode, image, image_folder, input_dir, output_dir, show_extras_results, *args, save_output=save_output) + return run_postprocessing("", extras_mode, image, image_folder, input_dir, output_dir, show_extras_results, *args, save_output=save_output) From 24dae9bc4cc03a30236957d9c35d37aed79f6f5d Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Mon, 4 Dec 2023 12:36:41 +0300 Subject: [PATCH 203/342] repair old handler for postprocessing API --- modules/postprocessing.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/postprocessing.py b/modules/postprocessing.py index 3c85a74c1..d166f859b 100644 --- a/modules/postprocessing.py +++ b/modules/postprocessing.py @@ -153,4 +153,4 @@ def run_extras(extras_mode, resize_mode, image, image_folder, input_dir, output_ }, }) - return run_postprocessing(extras_mode, image, image_folder, input_dir, output_dir, show_extras_results, *args, save_output=save_output) + return run_postprocessing("", extras_mode, image, image_folder, input_dir, output_dir, show_extras_results, *args, save_output=save_output) From 883d6a2b34a2817304d23c2481a6f9fc56687a53 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Mon, 4 Dec 2023 13:11:00 +0300 Subject: [PATCH 204/342] repair old handler for postprocessing API in a way that doesn't break interface --- modules/postprocessing.py | 8 ++++++-- modules/ui_postprocessing.py | 2 +- 2 files changed, 7 insertions(+), 3 deletions(-) diff --git a/modules/postprocessing.py b/modules/postprocessing.py index d166f859b..0c59fad48 100644 --- a/modules/postprocessing.py +++ b/modules/postprocessing.py @@ -6,7 +6,7 @@ from modules import shared, images, devices, scripts, scripts_postprocessing, ui from modules.shared import opts -def run_postprocessing(id_task, extras_mode, image, image_folder, input_dir, output_dir, show_extras_results, *args, save_output: bool = True): +def run_postprocessing(extras_mode, image, image_folder, input_dir, output_dir, show_extras_results, *args, save_output: bool = True): devices.torch_gc() shared.state.begin(job="extras") @@ -128,6 +128,10 @@ def run_postprocessing(id_task, extras_mode, image, image_folder, input_dir, out return outputs, ui_common.plaintext_to_html(infotext), '' +def run_postprocessing_webui(id_task, *args, **kwargs): + return run_postprocessing(*args, **kwargs) + + def run_extras(extras_mode, resize_mode, image, image_folder, input_dir, output_dir, show_extras_results, gfpgan_visibility, codeformer_visibility, codeformer_weight, upscaling_resize, upscaling_resize_w, upscaling_resize_h, upscaling_crop, extras_upscaler_1, extras_upscaler_2, extras_upscaler_2_visibility, upscale_first: bool, save_output: bool = True): """old handler for API""" @@ -153,4 +157,4 @@ def run_extras(extras_mode, resize_mode, image, image_folder, input_dir, output_ }, }) - return run_postprocessing("", extras_mode, image, image_folder, input_dir, output_dir, show_extras_results, *args, save_output=save_output) + return run_postprocessing(extras_mode, image, image_folder, input_dir, output_dir, show_extras_results, *args, save_output=save_output) diff --git a/modules/ui_postprocessing.py b/modules/ui_postprocessing.py index fbad0800a..13d888e48 100644 --- a/modules/ui_postprocessing.py +++ b/modules/ui_postprocessing.py @@ -35,7 +35,7 @@ def create_ui(): tab_batch_dir.select(fn=lambda: 2, inputs=[], outputs=[tab_index]) submit.click( - fn=call_queue.wrap_gradio_gpu_call(postprocessing.run_postprocessing, extra_outputs=[None, '']), + fn=call_queue.wrap_gradio_gpu_call(postprocessing.run_postprocessing_webui, extra_outputs=[None, '']), _js="submit_extras", inputs=[ dummy_component, From 81105ee0135f1c475920bf44d3a04fc181aed29e Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Mon, 4 Dec 2023 13:11:00 +0300 Subject: [PATCH 205/342] repair old handler for postprocessing API in a way that doesn't break interface --- modules/postprocessing.py | 8 ++++++-- modules/ui_postprocessing.py | 2 +- 2 files changed, 7 insertions(+), 3 deletions(-) diff --git a/modules/postprocessing.py b/modules/postprocessing.py index d166f859b..0c59fad48 100644 --- a/modules/postprocessing.py +++ b/modules/postprocessing.py @@ -6,7 +6,7 @@ from modules import shared, images, devices, scripts, scripts_postprocessing, ui from modules.shared import opts -def run_postprocessing(id_task, extras_mode, image, image_folder, input_dir, output_dir, show_extras_results, *args, save_output: bool = True): +def run_postprocessing(extras_mode, image, image_folder, input_dir, output_dir, show_extras_results, *args, save_output: bool = True): devices.torch_gc() shared.state.begin(job="extras") @@ -128,6 +128,10 @@ def run_postprocessing(id_task, extras_mode, image, image_folder, input_dir, out return outputs, ui_common.plaintext_to_html(infotext), '' +def run_postprocessing_webui(id_task, *args, **kwargs): + return run_postprocessing(*args, **kwargs) + + def run_extras(extras_mode, resize_mode, image, image_folder, input_dir, output_dir, show_extras_results, gfpgan_visibility, codeformer_visibility, codeformer_weight, upscaling_resize, upscaling_resize_w, upscaling_resize_h, upscaling_crop, extras_upscaler_1, extras_upscaler_2, extras_upscaler_2_visibility, upscale_first: bool, save_output: bool = True): """old handler for API""" @@ -153,4 +157,4 @@ def run_extras(extras_mode, resize_mode, image, image_folder, input_dir, output_ }, }) - return run_postprocessing("", extras_mode, image, image_folder, input_dir, output_dir, show_extras_results, *args, save_output=save_output) + return run_postprocessing(extras_mode, image, image_folder, input_dir, output_dir, show_extras_results, *args, save_output=save_output) diff --git a/modules/ui_postprocessing.py b/modules/ui_postprocessing.py index fbad0800a..13d888e48 100644 --- a/modules/ui_postprocessing.py +++ b/modules/ui_postprocessing.py @@ -35,7 +35,7 @@ def create_ui(): tab_batch_dir.select(fn=lambda: 2, inputs=[], outputs=[tab_index]) submit.click( - fn=call_queue.wrap_gradio_gpu_call(postprocessing.run_postprocessing, extra_outputs=[None, '']), + fn=call_queue.wrap_gradio_gpu_call(postprocessing.run_postprocessing_webui, extra_outputs=[None, '']), _js="submit_extras", inputs=[ dummy_component, From 22e23dbf29b0bbc807daa57318c31145f8dd0774 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Mon, 4 Dec 2023 15:56:03 +0300 Subject: [PATCH 206/342] add hypertile infotext --- .../hypertile/scripts/hypertile_script.py | 53 +++++++++++++++---- 1 file changed, 42 insertions(+), 11 deletions(-) diff --git a/extensions-builtin/hypertile/scripts/hypertile_script.py b/extensions-builtin/hypertile/scripts/hypertile_script.py index d3ab60915..395d584b6 100644 --- a/extensions-builtin/hypertile/scripts/hypertile_script.py +++ b/extensions-builtin/hypertile/scripts/hypertile_script.py @@ -17,11 +17,42 @@ class ScriptHypertile(scripts.Script): configure_hypertile(p.width, p.height, enable_unet=shared.opts.hypertile_enable_unet) + self.add_infotext(p) + def before_hr(self, p, *args): + + enable = shared.opts.hypertile_enable_unet_secondpass or shared.opts.hypertile_enable_unet + # exclusive hypertile seed for the second pass - if not shared.opts.hypertile_enable_unet: + if enable: hypertile.set_hypertile_seed(p.all_seeds[0]) - configure_hypertile(p.hr_upscale_to_x, p.hr_upscale_to_y, enable_unet=shared.opts.hypertile_enable_unet_secondpass) + + configure_hypertile(p.hr_upscale_to_x, p.hr_upscale_to_y, enable_unet=enable) + + if enable and not shared.opts.hypertile_enable_unet: + p.extra_generation_params["Hypertile U-Net second pass"] = True + + self.add_infotext(p, add_unet_params=True) + + def add_infotext(self, p, add_unet_params=False): + def option(name): + value = getattr(shared.opts, name) + default_value = shared.opts.get_default(name) + return None if value == default_value else value + + if shared.opts.hypertile_enable_unet: + p.extra_generation_params["Hypertile U-Net"] = True + + if shared.opts.hypertile_enable_unet or add_unet_params: + p.extra_generation_params["Hypertile U-Net max depth"] = option('hypertile_max_depth_unet') + p.extra_generation_params["Hypertile U-Net max tile size"] = option('hypertile_max_tile_unet') + p.extra_generation_params["Hypertile U-Net swap size"] = option('hypertile_swap_size_unet') + + if shared.opts.hypertile_enable_vae: + p.extra_generation_params["Hypertile VAE"] = True + p.extra_generation_params["Hypertile VAE max depth"] = option('hypertile_max_depth_vae') + p.extra_generation_params["Hypertile VAE max tile size"] = option('hypertile_max_tile_vae') + p.extra_generation_params["Hypertile VAE swap size"] = option('hypertile_swap_size_vae') def configure_hypertile(width, height, enable_unet=True): @@ -57,16 +88,16 @@ def on_ui_settings(): benefit. """), - "hypertile_enable_unet": shared.OptionInfo(False, "Enable Hypertile U-Net").info("noticeable change in details of the generated picture; if enabled, overrides the setting below"), - "hypertile_enable_unet_secondpass": shared.OptionInfo(False, "Enable Hypertile U-Net for hires fix second pass"), - "hypertile_max_depth_unet": shared.OptionInfo(3, "Hypertile U-Net max depth", gr.Slider, {"minimum": 0, "maximum": 3, "step": 1}), - "hypertile_max_tile_unet": shared.OptionInfo(256, "Hypertile U-net max tile size", gr.Slider, {"minimum": 0, "maximum": 512, "step": 16}), - "hypertile_swap_size_unet": shared.OptionInfo(3, "Hypertile U-net swap size", gr.Slider, {"minimum": 0, "maximum": 64, "step": 1}), + "hypertile_enable_unet": shared.OptionInfo(False, "Enable Hypertile U-Net", infotext="Hypertile U-Net").info("enables hypertile for all modes, including hires fix second pass; noticeable change in details of the generated picture"), + "hypertile_enable_unet_secondpass": shared.OptionInfo(False, "Enable Hypertile U-Net for hires fix second pass", infotext="Hypertile U-Net second pass").info("enables hypertile just for hires fix second pass - regardless of whether the above setting is enabled"), + "hypertile_max_depth_unet": shared.OptionInfo(3, "Hypertile U-Net max depth", gr.Slider, {"minimum": 0, "maximum": 3, "step": 1}, infotext="Hypertile U-Net max depth").info("larger = more neural network layers affected; minor effect on performance"), + "hypertile_max_tile_unet": shared.OptionInfo(256, "Hypertile U-Net max tile size", gr.Slider, {"minimum": 0, "maximum": 512, "step": 16}, infotext="Hypertile U-Net max tile size").info("larger = worse performance"), + "hypertile_swap_size_unet": shared.OptionInfo(3, "Hypertile U-Net swap size", gr.Slider, {"minimum": 0, "maximum": 64, "step": 1}, infotext="Hypertile U-Net swap size"), - "hypertile_enable_vae": shared.OptionInfo(False, "Enable Hypertile VAE").info("minimal change in the generated picture"), - "hypertile_max_depth_vae": shared.OptionInfo(3, "Hypertile VAE max depth", gr.Slider, {"minimum": 0, "maximum": 3, "step": 1}), - "hypertile_max_tile_vae": shared.OptionInfo(128, "Hypertile VAE max tile size", gr.Slider, {"minimum": 0, "maximum": 512, "step": 16}), - "hypertile_swap_size_vae": shared.OptionInfo(3, "Hypertile VAE swap size ", gr.Slider, {"minimum": 0, "maximum": 64, "step": 1}), + "hypertile_enable_vae": shared.OptionInfo(False, "Enable Hypertile VAE", infotext="Hypertile VAE").info("minimal change in the generated picture"), + "hypertile_max_depth_vae": shared.OptionInfo(3, "Hypertile VAE max depth", gr.Slider, {"minimum": 0, "maximum": 3, "step": 1}, infotext="Hypertile VAE max depth"), + "hypertile_max_tile_vae": shared.OptionInfo(128, "Hypertile VAE max tile size", gr.Slider, {"minimum": 0, "maximum": 512, "step": 16}, infotext="Hypertile VAE max tile size"), + "hypertile_swap_size_vae": shared.OptionInfo(3, "Hypertile VAE swap size ", gr.Slider, {"minimum": 0, "maximum": 64, "step": 1}, infotext="Hypertile VAE swap size"), } for name, opt in options.items(): From 368d66c9ccca0270cc64d6a64d22bfa562f28361 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Mon, 4 Dec 2023 15:56:03 +0300 Subject: [PATCH 207/342] add hypertile infotext --- .../hypertile/scripts/hypertile_script.py | 53 +++++++++++++++---- 1 file changed, 42 insertions(+), 11 deletions(-) diff --git a/extensions-builtin/hypertile/scripts/hypertile_script.py b/extensions-builtin/hypertile/scripts/hypertile_script.py index d3ab60915..395d584b6 100644 --- a/extensions-builtin/hypertile/scripts/hypertile_script.py +++ b/extensions-builtin/hypertile/scripts/hypertile_script.py @@ -17,11 +17,42 @@ class ScriptHypertile(scripts.Script): configure_hypertile(p.width, p.height, enable_unet=shared.opts.hypertile_enable_unet) + self.add_infotext(p) + def before_hr(self, p, *args): + + enable = shared.opts.hypertile_enable_unet_secondpass or shared.opts.hypertile_enable_unet + # exclusive hypertile seed for the second pass - if not shared.opts.hypertile_enable_unet: + if enable: hypertile.set_hypertile_seed(p.all_seeds[0]) - configure_hypertile(p.hr_upscale_to_x, p.hr_upscale_to_y, enable_unet=shared.opts.hypertile_enable_unet_secondpass) + + configure_hypertile(p.hr_upscale_to_x, p.hr_upscale_to_y, enable_unet=enable) + + if enable and not shared.opts.hypertile_enable_unet: + p.extra_generation_params["Hypertile U-Net second pass"] = True + + self.add_infotext(p, add_unet_params=True) + + def add_infotext(self, p, add_unet_params=False): + def option(name): + value = getattr(shared.opts, name) + default_value = shared.opts.get_default(name) + return None if value == default_value else value + + if shared.opts.hypertile_enable_unet: + p.extra_generation_params["Hypertile U-Net"] = True + + if shared.opts.hypertile_enable_unet or add_unet_params: + p.extra_generation_params["Hypertile U-Net max depth"] = option('hypertile_max_depth_unet') + p.extra_generation_params["Hypertile U-Net max tile size"] = option('hypertile_max_tile_unet') + p.extra_generation_params["Hypertile U-Net swap size"] = option('hypertile_swap_size_unet') + + if shared.opts.hypertile_enable_vae: + p.extra_generation_params["Hypertile VAE"] = True + p.extra_generation_params["Hypertile VAE max depth"] = option('hypertile_max_depth_vae') + p.extra_generation_params["Hypertile VAE max tile size"] = option('hypertile_max_tile_vae') + p.extra_generation_params["Hypertile VAE swap size"] = option('hypertile_swap_size_vae') def configure_hypertile(width, height, enable_unet=True): @@ -57,16 +88,16 @@ def on_ui_settings(): benefit. """), - "hypertile_enable_unet": shared.OptionInfo(False, "Enable Hypertile U-Net").info("noticeable change in details of the generated picture; if enabled, overrides the setting below"), - "hypertile_enable_unet_secondpass": shared.OptionInfo(False, "Enable Hypertile U-Net for hires fix second pass"), - "hypertile_max_depth_unet": shared.OptionInfo(3, "Hypertile U-Net max depth", gr.Slider, {"minimum": 0, "maximum": 3, "step": 1}), - "hypertile_max_tile_unet": shared.OptionInfo(256, "Hypertile U-net max tile size", gr.Slider, {"minimum": 0, "maximum": 512, "step": 16}), - "hypertile_swap_size_unet": shared.OptionInfo(3, "Hypertile U-net swap size", gr.Slider, {"minimum": 0, "maximum": 64, "step": 1}), + "hypertile_enable_unet": shared.OptionInfo(False, "Enable Hypertile U-Net", infotext="Hypertile U-Net").info("enables hypertile for all modes, including hires fix second pass; noticeable change in details of the generated picture"), + "hypertile_enable_unet_secondpass": shared.OptionInfo(False, "Enable Hypertile U-Net for hires fix second pass", infotext="Hypertile U-Net second pass").info("enables hypertile just for hires fix second pass - regardless of whether the above setting is enabled"), + "hypertile_max_depth_unet": shared.OptionInfo(3, "Hypertile U-Net max depth", gr.Slider, {"minimum": 0, "maximum": 3, "step": 1}, infotext="Hypertile U-Net max depth").info("larger = more neural network layers affected; minor effect on performance"), + "hypertile_max_tile_unet": shared.OptionInfo(256, "Hypertile U-Net max tile size", gr.Slider, {"minimum": 0, "maximum": 512, "step": 16}, infotext="Hypertile U-Net max tile size").info("larger = worse performance"), + "hypertile_swap_size_unet": shared.OptionInfo(3, "Hypertile U-Net swap size", gr.Slider, {"minimum": 0, "maximum": 64, "step": 1}, infotext="Hypertile U-Net swap size"), - "hypertile_enable_vae": shared.OptionInfo(False, "Enable Hypertile VAE").info("minimal change in the generated picture"), - "hypertile_max_depth_vae": shared.OptionInfo(3, "Hypertile VAE max depth", gr.Slider, {"minimum": 0, "maximum": 3, "step": 1}), - "hypertile_max_tile_vae": shared.OptionInfo(128, "Hypertile VAE max tile size", gr.Slider, {"minimum": 0, "maximum": 512, "step": 16}), - "hypertile_swap_size_vae": shared.OptionInfo(3, "Hypertile VAE swap size ", gr.Slider, {"minimum": 0, "maximum": 64, "step": 1}), + "hypertile_enable_vae": shared.OptionInfo(False, "Enable Hypertile VAE", infotext="Hypertile VAE").info("minimal change in the generated picture"), + "hypertile_max_depth_vae": shared.OptionInfo(3, "Hypertile VAE max depth", gr.Slider, {"minimum": 0, "maximum": 3, "step": 1}, infotext="Hypertile VAE max depth"), + "hypertile_max_tile_vae": shared.OptionInfo(128, "Hypertile VAE max tile size", gr.Slider, {"minimum": 0, "maximum": 512, "step": 16}, infotext="Hypertile VAE max tile size"), + "hypertile_swap_size_vae": shared.OptionInfo(3, "Hypertile VAE swap size ", gr.Slider, {"minimum": 0, "maximum": 64, "step": 1}, infotext="Hypertile VAE swap size"), } for name, opt in options.items(): From 854f8c318c2610c76259056ab02739176aa849e8 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Tue, 5 Dec 2023 04:40:12 +0900 Subject: [PATCH 208/342] remove clean_text() --- modules/styles.py | 23 +++-------------------- 1 file changed, 3 insertions(+), 20 deletions(-) diff --git a/modules/styles.py b/modules/styles.py index 4d218cd7e..7fb6c2e11 100644 --- a/modules/styles.py +++ b/modules/styles.py @@ -2,7 +2,6 @@ import csv import fnmatch import os import os.path -import re import typing import shutil @@ -14,22 +13,6 @@ class PromptStyle(typing.NamedTuple): path: str = None -def clean_text(text: str) -> str: - """ - Iterating through a list of regular expressions and replacement strings, we - clean up the prompt and style text to make it easier to match against each - other. - """ - re_list = [ - ("multiple commas", re.compile("(,+\s+)+,?"), ", "), - ("multiple spaces", re.compile("\s{2,}"), " "), - ] - for _, regex, replace in re_list: - text = regex.sub(replace, text) - - return text.strip(", ") - - def merge_prompts(style_prompt: str, prompt: str) -> str: if "{prompt}" in style_prompt: res = style_prompt.replace("{prompt}", prompt) @@ -44,7 +27,7 @@ def apply_styles_to_prompt(prompt, styles): for style in styles: prompt = merge_prompts(style, prompt) - return clean_text(prompt) + return prompt def unwrap_style_text_from_prompt(style_text, prompt): @@ -56,8 +39,8 @@ def unwrap_style_text_from_prompt(style_text, prompt): Note that the "cleaned" version of the style text is only used for matching purposes here. It isn't returned; the original style text is not modified. """ - stripped_prompt = clean_text(prompt) - stripped_style_text = clean_text(style_text) + stripped_prompt = prompt + stripped_style_text = style_text if "{prompt}" in stripped_style_text: # Work out whether the prompt is wrapped in the style text. If so, we # return True and the "inner" prompt text that isn't part of the style. From 976c1053efeb5054692ed3cfa294cf79196f3946 Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Mon, 4 Dec 2023 16:06:58 -0700 Subject: [PATCH 209/342] Cleaned up code, moved main code contributions into soft_inpainting.py --- modules/processing.py | 56 ++------- modules/sd_samplers_cfg_denoiser.py | 84 ++----------- modules/soft_inpainting.py | 175 +++++++++++++++++++++++++--- modules/ui.py | 7 -- 4 files changed, 173 insertions(+), 149 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index b40b1a40d..0b3603875 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -892,55 +892,13 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: # Generate the mask(s) based on similarity between the original and denoised latent vectors if getattr(p, "image_mask", None) is not None and getattr(p, "soft_inpainting", None) is not None: - # latent_mask = p.nmask[0].float().cpu() - - # convert the original mask into a form we use to scale distances for thresholding - # mask_scalar = 1-(torch.clamp(latent_mask, min=0, max=1) ** (p.mask_blend_scale / 2)) - # mask_scalar = mask_scalar / (1.00001-mask_scalar) - # mask_scalar = mask_scalar.numpy() - - latent_orig = p.init_latent - latent_proc = samples_ddim - latent_distance = torch.norm(latent_proc - latent_orig, p=2, dim=1) - - kernel, kernel_center = images.get_gaussian_kernel(stddev_radius=1.5, max_radius=2) - - for i, (distance_map, overlay_image) in enumerate(zip(latent_distance, p.overlay_images)): - converted_mask = distance_map.float().cpu().numpy() - converted_mask = images.weighted_histogram_filter(converted_mask, kernel, kernel_center, - percentile_min=0.9, percentile_max=1, min_width=1) - converted_mask = images.weighted_histogram_filter(converted_mask, kernel, kernel_center, - percentile_min=0.25, percentile_max=0.75, min_width=1) - - # The distance at which opacity of original decreases to 50% - # half_weighted_distance = 1 # * mask_scalar - # converted_mask = converted_mask / half_weighted_distance - - converted_mask = 1 / (1 + converted_mask ** 2) - converted_mask = images.smootherstep(converted_mask) - converted_mask = 1 - converted_mask - converted_mask = 255. * converted_mask - converted_mask = converted_mask.astype(np.uint8) - converted_mask = Image.fromarray(converted_mask) - converted_mask = images.resize_image(2, converted_mask, p.width, p.height) - converted_mask = create_binary_mask(converted_mask, round=False) - - # Remove aliasing artifacts using a gaussian blur. - converted_mask = converted_mask.filter(ImageFilter.GaussianBlur(radius=4)) - - # Expand the mask to fit the whole image if needed. - if p.paste_to is not None: - converted_mask = uncrop(converted_mask, - (overlay_image.width, overlay_image.height), - p.paste_to) - - p.masks_for_overlay[i] = converted_mask - - image_masked = Image.new('RGBa', (overlay_image.width, overlay_image.height)) - image_masked.paste(overlay_image.convert("RGBA").convert("RGBa"), - mask=ImageOps.invert(converted_mask.convert('L'))) - - p.overlay_images[i] = image_masked.convert('RGBA') + si.generate_adaptive_masks(latent_orig=p.init_latent, + latent_processed=samples_ddim, + overlay_images=p.overlay_images, + masks_for_overlay=p.masks_for_overlay, + width=p.width, + height=p.height, + paste_to=p.paste_to) x_samples_ddim = decode_latent_batch(p.sd_model, samples_ddim, target_device=devices.cpu, diff --git a/modules/sd_samplers_cfg_denoiser.py b/modules/sd_samplers_cfg_denoiser.py index 0ee0b7dde..a700e6922 100644 --- a/modules/sd_samplers_cfg_denoiser.py +++ b/modules/sd_samplers_cfg_denoiser.py @@ -94,76 +94,6 @@ class CFGDenoiser(torch.nn.Module): self.sampler.sampler_extra_args['uncond'] = uc def forward(self, x, sigma, uncond, cond, cond_scale, s_min_uncond, image_cond): - def latent_blend(a, b, t, one_minus_t=None): - - """ - Interpolates two latent image representations according to the parameter t, - where the interpolated vectors' magnitudes are also interpolated separately. - The "detail_preservation" factor biases the magnitude interpolation towards - the larger of the two magnitudes. - """ - # NOTE: We use inplace operations wherever possible. - - if one_minus_t is None: - one_minus_t = 1 - t - - if self.soft_inpainting is None: - return a * one_minus_t + b * t - - # Linearly interpolate the image vectors. - a_scaled = a * one_minus_t - b_scaled = b * t - image_interp = a_scaled - image_interp.add_(b_scaled) - result_type = image_interp.dtype - del a_scaled, b_scaled - - # Calculate the magnitude of the interpolated vectors. (We will remove this magnitude.) - # 64-bit operations are used here to allow large exponents. - current_magnitude = torch.norm(image_interp, p=2, dim=1).to(torch.float64).add_(0.00001) - - # Interpolate the powered magnitudes, then un-power them (bring them back to a power of 1). - a_magnitude = torch.norm(a, p=2, dim=1).to(torch.float64).pow_(self.soft_inpainting.inpaint_detail_preservation) * one_minus_t - b_magnitude = torch.norm(b, p=2, dim=1).to(torch.float64).pow_(self.soft_inpainting.inpaint_detail_preservation) * t - desired_magnitude = a_magnitude - desired_magnitude.add_(b_magnitude).pow_(1 / self.soft_inpainting.inpaint_detail_preservation) - del a_magnitude, b_magnitude, one_minus_t - - # Change the linearly interpolated image vectors' magnitudes to the value we want. - # This is the last 64-bit operation. - image_interp_scaling_factor = desired_magnitude - image_interp_scaling_factor.div_(current_magnitude) - image_interp_scaled = image_interp - image_interp_scaled.mul_(image_interp_scaling_factor) - del current_magnitude - del desired_magnitude - del image_interp - del image_interp_scaling_factor - - image_interp_scaled = image_interp_scaled.to(result_type) - del result_type - - return image_interp_scaled - - def get_modified_nmask(nmask, _sigma): - """ - Converts a negative mask representing the transparency of the original latent vectors being overlayed - to a mask that is scaled according to the denoising strength for this step. - - Where: - 0 = fully opaque, infinite density, fully masked - 1 = fully transparent, zero density, fully unmasked - - We bring this transparency to a power, as this allows one to simulate N number of blending operations - where N can be any positive real value. Using this one can control the balance of influence between - the denoiser and the original latents according to the sigma value. - - NOTE: "mask" is not used - """ - if self.soft_inpainting is None: - return nmask - - return torch.pow(nmask, (_sigma ** self.soft_inpainting.mask_blend_power) * self.soft_inpainting.mask_blend_scale) if state.interrupted or state.skipped: raise sd_samplers_common.InterruptedException @@ -184,9 +114,12 @@ class CFGDenoiser(torch.nn.Module): # Blend in the original latents (before) if self.mask_before_denoising and self.mask is not None: if self.soft_inpainting is None: - x = latent_blend(self.init_latent, x, self.nmask, self.mask) + x = self.init_latent * self.mask + self.nmask * x else: - x = latent_blend(self.init_latent, x, get_modified_nmask(self.nmask, sigma)) + x = si.latent_blend(self.soft_inpainting, + self.init_latent, + x, + si.get_modified_nmask(self.soft_inpainting, self.nmask, sigma)) batch_size = len(conds_list) repeats = [len(conds_list[i]) for i in range(batch_size)] @@ -290,9 +223,12 @@ class CFGDenoiser(torch.nn.Module): # Blend in the original latents (after) if not self.mask_before_denoising and self.mask is not None: if self.soft_inpainting is None: - denoised = latent_blend(self.init_latent, denoised, self.nmask, self.mask) + denoised = self.init_latent * self.mask + self.nmask * denoised else: - denoised = latent_blend(self.init_latent, denoised, get_modified_nmask(self.nmask, sigma)) + denoised = si.latent_blend(self.soft_inpainting, + self.init_latent, + denoised, + si.get_modified_nmask(self.soft_inpainting, self.nmask, sigma)) self.sampler.last_latent = self.get_pred_x0(torch.cat([x_in[i:i + 1] for i in denoised_image_indexes]), torch.cat([x_out[i:i + 1] for i in denoised_image_indexes]), sigma) diff --git a/modules/soft_inpainting.py b/modules/soft_inpainting.py index 259c36ec8..b81c8dd95 100644 --- a/modules/soft_inpainting.py +++ b/modules/soft_inpainting.py @@ -4,13 +4,6 @@ class SoftInpaintingSettings: self.mask_blend_scale = mask_blend_scale self.inpaint_detail_preservation = inpaint_detail_preservation - def get_paste_fields(self): - return [ - (self.mask_blend_power, gen_param_labels.mask_blend_power), - (self.mask_blend_scale, gen_param_labels.mask_blend_scale), - (self.inpaint_detail_preservation, gen_param_labels.inpaint_detail_preservation), - ] - def add_generation_params(self, dest): dest[enabled_gen_param_label] = True dest[gen_param_labels.mask_blend_power] = self.mask_blend_power @@ -18,25 +11,169 @@ class SoftInpaintingSettings: dest[gen_param_labels.inpaint_detail_preservation] = self.inpaint_detail_preservation +# ------------------- Methods ------------------- + + +def latent_blend(soft_inpainting, a, b, t): + """ + Interpolates two latent image representations according to the parameter t, + where the interpolated vectors' magnitudes are also interpolated separately. + The "detail_preservation" factor biases the magnitude interpolation towards + the larger of the two magnitudes. + """ + import torch + + # NOTE: We use inplace operations wherever possible. + + one_minus_t = 1 - t + + # Linearly interpolate the image vectors. + a_scaled = a * one_minus_t + b_scaled = b * t + image_interp = a_scaled + image_interp.add_(b_scaled) + result_type = image_interp.dtype + del a_scaled, b_scaled + + # Calculate the magnitude of the interpolated vectors. (We will remove this magnitude.) + # 64-bit operations are used here to allow large exponents. + current_magnitude = torch.norm(image_interp, p=2, dim=1).to(torch.float64).add_(0.00001) + + # Interpolate the powered magnitudes, then un-power them (bring them back to a power of 1). + a_magnitude = torch.norm(a, p=2, dim=1).to(torch.float64).pow_(soft_inpainting.inpaint_detail_preservation) * one_minus_t + b_magnitude = torch.norm(b, p=2, dim=1).to(torch.float64).pow_(soft_inpainting.inpaint_detail_preservation) * t + desired_magnitude = a_magnitude + desired_magnitude.add_(b_magnitude).pow_(1 / soft_inpainting.inpaint_detail_preservation) + del a_magnitude, b_magnitude, one_minus_t + + # Change the linearly interpolated image vectors' magnitudes to the value we want. + # This is the last 64-bit operation. + image_interp_scaling_factor = desired_magnitude + image_interp_scaling_factor.div_(current_magnitude) + image_interp_scaling_factor = image_interp_scaling_factor.to(result_type) + image_interp_scaled = image_interp + image_interp_scaled.mul_(image_interp_scaling_factor) + del current_magnitude + del desired_magnitude + del image_interp + del image_interp_scaling_factor + del result_type + + return image_interp_scaled + + +def get_modified_nmask(soft_inpainting, nmask, sigma): + """ + Converts a negative mask representing the transparency of the original latent vectors being overlayed + to a mask that is scaled according to the denoising strength for this step. + + Where: + 0 = fully opaque, infinite density, fully masked + 1 = fully transparent, zero density, fully unmasked + + We bring this transparency to a power, as this allows one to simulate N number of blending operations + where N can be any positive real value. Using this one can control the balance of influence between + the denoiser and the original latents according to the sigma value. + + NOTE: "mask" is not used + """ + import torch + return torch.pow(nmask, (sigma ** soft_inpainting.mask_blend_power) * soft_inpainting.mask_blend_scale) + + +def generate_adaptive_masks( + latent_orig, + latent_processed, + overlay_images, + masks_for_overlay, + width, height, + paste_to): + import torch + import numpy as np + import modules.processing as proc + import modules.images as images + from PIL import Image, ImageOps, ImageFilter + + # TODO: Bias the blending according to the latent mask, add adjustable parameter for bias control. + # latent_mask = p.nmask[0].float().cpu() + # convert the original mask into a form we use to scale distances for thresholding + # mask_scalar = 1-(torch.clamp(latent_mask, min=0, max=1) ** (p.mask_blend_scale / 2)) + # mask_scalar = mask_scalar / (1.00001-mask_scalar) + # mask_scalar = mask_scalar.numpy() + + latent_distance = torch.norm(latent_processed - latent_orig, p=2, dim=1) + + kernel, kernel_center = images.get_gaussian_kernel(stddev_radius=1.5, max_radius=2) + + for i, (distance_map, overlay_image) in enumerate(zip(latent_distance, overlay_images)): + converted_mask = distance_map.float().cpu().numpy() + converted_mask = images.weighted_histogram_filter(converted_mask, kernel, kernel_center, + percentile_min=0.9, percentile_max=1, min_width=1) + converted_mask = images.weighted_histogram_filter(converted_mask, kernel, kernel_center, + percentile_min=0.25, percentile_max=0.75, min_width=1) + + # The distance at which opacity of original decreases to 50% + # half_weighted_distance = 1 # * mask_scalar + # converted_mask = converted_mask / half_weighted_distance + + converted_mask = 1 / (1 + converted_mask ** 2) + converted_mask = images.smootherstep(converted_mask) + converted_mask = 1 - converted_mask + converted_mask = 255. * converted_mask + converted_mask = converted_mask.astype(np.uint8) + converted_mask = Image.fromarray(converted_mask) + converted_mask = images.resize_image(2, converted_mask, width, height) + converted_mask = proc.create_binary_mask(converted_mask, round=False) + + # Remove aliasing artifacts using a gaussian blur. + converted_mask = converted_mask.filter(ImageFilter.GaussianBlur(radius=4)) + + # Expand the mask to fit the whole image if needed. + if paste_to is not None: + converted_mask = proc. uncrop(converted_mask, + (overlay_image.width, overlay_image.height), + paste_to) + + masks_for_overlay[i] = converted_mask + + image_masked = Image.new('RGBa', (overlay_image.width, overlay_image.height)) + image_masked.paste(overlay_image.convert("RGBA").convert("RGBa"), + mask=ImageOps.invert(converted_mask.convert('L'))) + + overlay_images[i] = image_masked.convert('RGBA') + + +# ------------------- Constants ------------------- + + +default = SoftInpaintingSettings(1, 0.5, 4) + enabled_ui_label = "Soft inpainting" enabled_gen_param_label = "Soft inpainting enabled" enabled_el_id = "soft_inpainting_enabled" -default = SoftInpaintingSettings(1, 0.5, 4) -ui_labels = SoftInpaintingSettings("Schedule bias", "Preservation strength", "Transition contrast boost") +ui_labels = SoftInpaintingSettings( + "Schedule bias", + "Preservation strength", + "Transition contrast boost") ui_info = SoftInpaintingSettings( - mask_blend_power="Shifts when preservation of original content occurs during denoising.", - # "Below 1: Stronger preservation near the end (with low sigma)\n" - # "1: Balanced (proportional to sigma)\n" - # "Above 1: Stronger preservation in the beginning (with high sigma)", - mask_blend_scale="How strongly partially masked content should be preserved.", - # "Low values: Favors generated content.\n" - # "High values: Favors original content.", - inpaint_detail_preservation="Amplifies the contrast that may be lost in partially masked regions.") + "Shifts when preservation of original content occurs during denoising.", + "How strongly partially masked content should be preserved.", + "Amplifies the contrast that may be lost in partially masked regions.") -gen_param_labels = SoftInpaintingSettings("Soft inpainting schedule bias", "Soft inpainting preservation strength", "Soft inpainting transition contrast boost") -el_ids = SoftInpaintingSettings("mask_blend_power", "mask_blend_scale", "inpaint_detail_preservation") +gen_param_labels = SoftInpaintingSettings( + "Soft inpainting schedule bias", + "Soft inpainting preservation strength", + "Soft inpainting transition contrast boost") + +el_ids = SoftInpaintingSettings( + "mask_blend_power", + "mask_blend_scale", + "inpaint_detail_preservation") + + +# ------------------- UI ------------------- def gradio_ui(): diff --git a/modules/ui.py b/modules/ui.py index 0e4fb17aa..4f1265a3e 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -683,13 +683,6 @@ def create_ui(): with FormRow(): soft_inpainting = si.gradio_ui() - - """ - mask_blend_power = gr.Slider(label='Blending bias', minimum=0, maximum=8, step=0.1, value=1, elem_id="img2img_mask_blend_power") - mask_blend_scale = gr.Slider(label='Blending preservation', minimum=0, maximum=8, step=0.05, value=0.5, elem_id="img2img_mask_blend_scale") - inpaint_detail_preservation = gr.Slider(label='Blending contrast boost', minimum=1, maximum=32, step=0.5, value=4, elem_id="img2img_mask_blend_offset") - """ - with FormRow(): inpainting_mask_invert = gr.Radio(label='Mask mode', choices=['Inpaint masked', 'Inpaint not masked'], value='Inpaint masked', type="index", elem_id="img2img_mask_mode") From 1455159cf44cd8c21656818463f6095eae887540 Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Mon, 4 Dec 2023 16:43:57 -0700 Subject: [PATCH 210/342] Fixed issue with whitespace, removed commented out code that was meant to be used as a reference. --- modules/soft_inpainting.py | 20 ++++++-------------- 1 file changed, 6 insertions(+), 14 deletions(-) diff --git a/modules/soft_inpainting.py b/modules/soft_inpainting.py index b81c8dd95..56a877746 100644 --- a/modules/soft_inpainting.py +++ b/modules/soft_inpainting.py @@ -179,15 +179,7 @@ el_ids = SoftInpaintingSettings( def gradio_ui(): import gradio as gr from modules.ui_components import InputAccordion - """ - with InputAccordion(False, label="Refiner", elem_id=self.elem_id("enable")) as enable_refiner: - with gr.Row(): - refiner_checkpoint = gr.Dropdown(label='Checkpoint', elem_id=self.elem_id("checkpoint"), choices=sd_models.checkpoint_tiles(), value='', tooltip="switch to another model in the middle of generation") - create_refresh_button(refiner_checkpoint, sd_models.list_models, lambda: {"choices": sd_models.checkpoint_tiles()}, self.elem_id("checkpoint_refresh")) - refiner_switch_at = gr.Slider(value=0.8, label="Switch at", minimum=0.01, maximum=1.0, step=0.01, elem_id=self.elem_id("switch_at"), tooltip="fraction of sampling steps when the switch to refiner model should happen; 1=never, 0.5=switch in the middle of generation") - - """ with InputAccordion(False, label=enabled_ui_label, elem_id=enabled_el_id) as soft_inpainting_enabled: with gr.Group(): gr.Markdown( @@ -223,11 +215,11 @@ def gradio_ui(): gr.Markdown( f""" ### {ui_labels.mask_blend_power} - + The blending strength of original content is scaled proportionally with the decreasing noise level values at each step (sigmas). This ensures that the influence of the denoiser and original content preservation is roughly balanced at each step. This balance can be shifted using this parameter, controlling whether earlier or later steps have stronger preservation. - + - **Below 1**: Stronger preservation near the end (with low sigma) - **1**: Balanced (proportional to sigma) - **Above 1**: Stronger preservation in the beginning (with high sigma) @@ -235,21 +227,21 @@ def gradio_ui(): gr.Markdown( f""" ### {ui_labels.mask_blend_scale} - + Skews whether partially masked image regions should be more likely to preserve the original content or favor inpainted content. This may need to be adjusted depending on the {ui_labels.mask_blend_power}, CFG Scale, prompt and Denoising strength. - + - **Low values**: Favors generated content. - **High values**: Favors original content. """) gr.Markdown( f""" ### {ui_labels.inpaint_detail_preservation} - + This parameter controls how the original latent vectors and denoised latent vectors are interpolated. With higher values, the magnitude of the resulting blended vector will be closer to the maximum of the two interpolated vectors. This can prevent the loss of contrast that occurs with linear interpolation. - + - **Low values**: Softer blending, details may fade. - **High values**: Stronger contrast, may over-saturate colors. """) From 57f29bd61dc30f1a8c94ead9b780f4655f7d7d6d Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Mon, 4 Dec 2023 17:41:18 -0700 Subject: [PATCH 211/342] Re-introduce latent blending step from the vanilla inpainting procedure. --- modules/processing.py | 3 +++ 1 file changed, 3 insertions(+) diff --git a/modules/processing.py b/modules/processing.py index 0b3603875..c8dc4d934 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -1597,6 +1597,9 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): samples = self.sampler.sample_img2img(self, self.init_latent, x, conditioning, unconditional_conditioning, image_conditioning=self.image_conditioning) + if self.mask is not None and self.soft_inpainting is None: + samples = samples * self.nmask + self.init_latent * self.mask + del x devices.torch_gc() From 60c602232fd760fb548fb0b3d18b5297f8823c2a Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Mon, 4 Dec 2023 17:41:51 -0700 Subject: [PATCH 212/342] Restored original formatting. --- modules/processing.py | 36 +++++++++++------------------------- 1 file changed, 11 insertions(+), 25 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index c8dc4d934..90ae249a4 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -370,10 +370,7 @@ class StableDiffusionProcessing: return self.edit_image_conditioning(source_image) if self.sampler.conditioning_key in {'hybrid', 'concat'}: - return self.inpainting_image_conditioning(source_image, - latent_image, - image_mask=image_mask, - round_image_mask=round_image_mask) + return self.inpainting_image_conditioning(source_image, latent_image, image_mask=image_mask, round_image_mask=round_image_mask) if self.sampler.conditioning_key == "crossattn-adm": return self.unclip_image_conditioning(source_image) @@ -885,7 +882,8 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: if getattr(samples_ddim, 'already_decoded', False): x_samples_ddim = samples_ddim - # todo: generate masks the old fashioned way + # todo: generate adaptive masks based on pixel differences. + # if p.masks_for_overlay is used, it will already be populated with masks else: if opts.sd_vae_decode_method != 'Full': p.extra_generation_params['VAE Decoder'] = opts.sd_vae_decode_method @@ -900,9 +898,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: height=p.height, paste_to=p.paste_to) - x_samples_ddim = decode_latent_batch(p.sd_model, samples_ddim, - target_device=devices.cpu, - check_for_nans=True) + x_samples_ddim = decode_latent_batch(p.sd_model, samples_ddim, target_device=devices.cpu, check_for_nans=True) x_samples_ddim = torch.stack(x_samples_ddim).float() x_samples_ddim = torch.clamp((x_samples_ddim + 1.0) / 2.0, min=0.0, max=1.0) @@ -927,9 +923,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: x_samples_ddim = batch_params.images def infotext(index=0, use_main_prompt=False): - return create_infotext(p, p.prompts, p.seeds, p.subseeds, - use_main_prompt=use_main_prompt, index=index, - all_negative_prompts=p.negative_prompts) + return create_infotext(p, p.prompts, p.seeds, p.subseeds, use_main_prompt=use_main_prompt, index=index, all_negative_prompts=p.negative_prompts) save_samples = p.save_samples() @@ -972,8 +966,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: image = apply_overlay(image, p.paste_to, i, p.overlay_images) if save_samples: - images.save_image(image, p.outpath_samples, "", p.seeds[i], - p.prompts[i], opts.samples_format, info=infotext(i), p=p) + images.save_image(image, p.outpath_samples, "", p.seeds[i], p.prompts[i], opts.samples_format, info=infotext(i), p=p) text = infotext(i) infotexts.append(text) @@ -983,14 +976,10 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: if save_samples and any([opts.save_mask, opts.save_mask_composite, opts.return_mask, opts.return_mask_composite]): if hasattr(p, 'masks_for_overlay') and p.masks_for_overlay: image_mask = p.masks_for_overlay[i].convert('RGB') - image_mask_composite = Image.composite( - original_denoised_image.convert('RGBA').convert('RGBa'), Image.new('RGBa', image.size), - images.resize_image(2, p.masks_for_overlay[i], image.width, image.height).convert('L')).convert('RGBA') + image_mask_composite = Image.composite(original_denoised_image.convert('RGBA').convert('RGBa'), Image.new('RGBa', image.size), images.resize_image(2, p.masks_for_overlay[i], image.width, image.height).convert('L')).convert('RGBA') elif hasattr(p, 'mask_for_overlay') and p.mask_for_overlay: image_mask = p.mask_for_overlay.convert('RGB') - image_mask_composite = Image.composite( - original_denoised_image.convert('RGBA').convert('RGBa'), Image.new('RGBa', image.size), - images.resize_image(2, p.mask_for_overlay, image.width, image.height).convert('L')).convert('RGBA') + image_mask_composite = Image.composite(original_denoised_image.convert('RGBA').convert('RGBa'), Image.new('RGBa', image.size), images.resize_image(2, p.mask_for_overlay, image.width, image.height).convert('L')).convert('RGBA') else: image_mask = None image_mask_composite = None @@ -1515,8 +1504,8 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): self.masks_for_overlay.append(image_mask) else: image_masked = Image.new('RGBa', (image.width, image.height)) - image_masked.paste(image.convert("RGBA").convert("RGBa"), - mask=ImageOps.invert(self.mask_for_overlay.convert('L'))) + image_masked.paste(image.convert("RGBA").convert("RGBa"), mask=ImageOps.invert(self.mask_for_overlay.convert('L'))) + self.overlay_images.append(image_masked.convert('RGBA')) # crop_region is not None if we are doing inpaint full res @@ -1583,10 +1572,7 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): elif self.inpainting_fill == 3: self.init_latent = self.init_latent * self.mask - self.image_conditioning = self.img2img_image_conditioning(image * 2 - 1, - self.init_latent, - image_mask, - self.soft_inpainting is None) + self.image_conditioning = self.img2img_image_conditioning(image * 2 - 1, self.init_latent, image_mask, self.soft_inpainting is None) def sample(self, conditioning, unconditional_conditioning, seeds, subseeds, subseed_strength, prompts): x = self.rng.next() From b32a334e3da7b06d82441beaa08a673b4f55bca1 Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Mon, 4 Dec 2023 17:57:10 -0700 Subject: [PATCH 213/342] Applies a convert('RGBA') operation early to mimic previous behaviour. --- modules/processing.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/processing.py b/modules/processing.py index 90ae249a4..7fc282cfd 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -1500,7 +1500,7 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): if image_mask is not None: if self.soft_inpainting is not None: # We apply the masks AFTER to adjust mask based on changed content. - self.overlay_images.append(image) + self.overlay_images.append(image.convert('RGBA')) self.masks_for_overlay.append(image_mask) else: image_masked = Image.new('RGBa', (image.width, image.height)) From 6fc12428e3c5f903584ca7986e0c441f80fa2807 Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Mon, 4 Dec 2023 19:42:59 -0700 Subject: [PATCH 214/342] Fixed issue where batched inpainting (batch size > 1) wouldn't work because of mismatched tensor sizes. The 'already_decoded' decoded case should also be handled correctly (tested indirectly). --- modules/processing.py | 23 ++++++++----- modules/soft_inpainting.py | 66 ++++++++++++++++++++++++++++++++------ 2 files changed, 71 insertions(+), 18 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index 7fc282cfd..71bb056a2 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -883,20 +883,27 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: if getattr(samples_ddim, 'already_decoded', False): x_samples_ddim = samples_ddim # todo: generate adaptive masks based on pixel differences. - # if p.masks_for_overlay is used, it will already be populated with masks + if getattr(p, "image_mask", None) is not None and getattr(p, "soft_inpainting", None) is not None: + si.apply_masks(soft_inpainting=p.soft_inpainting, + nmask=p.nmask, + overlay_images=p.overlay_images, + masks_for_overlay=p.masks_for_overlay, + width=p.width, + height=p.height, + paste_to=p.paste_to) else: if opts.sd_vae_decode_method != 'Full': p.extra_generation_params['VAE Decoder'] = opts.sd_vae_decode_method # Generate the mask(s) based on similarity between the original and denoised latent vectors if getattr(p, "image_mask", None) is not None and getattr(p, "soft_inpainting", None) is not None: - si.generate_adaptive_masks(latent_orig=p.init_latent, - latent_processed=samples_ddim, - overlay_images=p.overlay_images, - masks_for_overlay=p.masks_for_overlay, - width=p.width, - height=p.height, - paste_to=p.paste_to) + si.apply_adaptive_masks(latent_orig=p.init_latent, + latent_processed=samples_ddim, + overlay_images=p.overlay_images, + masks_for_overlay=p.masks_for_overlay, + width=p.width, + height=p.height, + paste_to=p.paste_to) x_samples_ddim = decode_latent_batch(p.sd_model, samples_ddim, target_device=devices.cpu, check_for_nans=True) diff --git a/modules/soft_inpainting.py b/modules/soft_inpainting.py index 56a877746..b36ac8fa1 100644 --- a/modules/soft_inpainting.py +++ b/modules/soft_inpainting.py @@ -25,26 +25,32 @@ def latent_blend(soft_inpainting, a, b, t): # NOTE: We use inplace operations wherever possible. - one_minus_t = 1 - t + # [4][w][h] to [1][4][w][h] + t2 = t.unsqueeze(0) + # [4][w][h] to [1][1][w][h] - the [4] seem redundant. + t3 = t[0].unsqueeze(0).unsqueeze(0) + + one_minus_t2 = 1 - t2 + one_minus_t3 = 1 - t3 # Linearly interpolate the image vectors. - a_scaled = a * one_minus_t - b_scaled = b * t + a_scaled = a * one_minus_t2 + b_scaled = b * t2 image_interp = a_scaled image_interp.add_(b_scaled) result_type = image_interp.dtype - del a_scaled, b_scaled + del a_scaled, b_scaled, t2, one_minus_t2 # Calculate the magnitude of the interpolated vectors. (We will remove this magnitude.) # 64-bit operations are used here to allow large exponents. - current_magnitude = torch.norm(image_interp, p=2, dim=1).to(torch.float64).add_(0.00001) + current_magnitude = torch.norm(image_interp, p=2, dim=1, keepdim=True).to(torch.float64).add_(0.00001) # Interpolate the powered magnitudes, then un-power them (bring them back to a power of 1). - a_magnitude = torch.norm(a, p=2, dim=1).to(torch.float64).pow_(soft_inpainting.inpaint_detail_preservation) * one_minus_t - b_magnitude = torch.norm(b, p=2, dim=1).to(torch.float64).pow_(soft_inpainting.inpaint_detail_preservation) * t + a_magnitude = torch.norm(a, p=2, dim=1, keepdim=True).to(torch.float64).pow_(soft_inpainting.inpaint_detail_preservation) * one_minus_t3 + b_magnitude = torch.norm(b, p=2, dim=1, keepdim=True).to(torch.float64).pow_(soft_inpainting.inpaint_detail_preservation) * t3 desired_magnitude = a_magnitude desired_magnitude.add_(b_magnitude).pow_(1 / soft_inpainting.inpaint_detail_preservation) - del a_magnitude, b_magnitude, one_minus_t + del a_magnitude, b_magnitude, t3, one_minus_t3 # Change the linearly interpolated image vectors' magnitudes to the value we want. # This is the last 64-bit operation. @@ -78,10 +84,11 @@ def get_modified_nmask(soft_inpainting, nmask, sigma): NOTE: "mask" is not used """ import torch - return torch.pow(nmask, (sigma ** soft_inpainting.mask_blend_power) * soft_inpainting.mask_blend_scale) + # todo: Why is sigma 2D? Both values are the same. + return torch.pow(nmask, (sigma[0] ** soft_inpainting.mask_blend_power) * soft_inpainting.mask_blend_scale) -def generate_adaptive_masks( +def apply_adaptive_masks( latent_orig, latent_processed, overlay_images, @@ -142,6 +149,45 @@ def generate_adaptive_masks( overlay_images[i] = image_masked.convert('RGBA') +def apply_masks( + soft_inpainting, + nmask, + overlay_images, + masks_for_overlay, + width, height, + paste_to): + import torch + import numpy as np + import modules.processing as proc + import modules.images as images + from PIL import Image, ImageOps, ImageFilter + + converted_mask = nmask[0].float() + converted_mask = torch.clamp(converted_mask, min=0, max=1).pow_(soft_inpainting.mask_blend_scale / 2) + converted_mask = 255. * converted_mask + converted_mask = converted_mask.cpu().numpy().astype(np.uint8) + converted_mask = Image.fromarray(converted_mask) + converted_mask = images.resize_image(2, converted_mask, width, height) + converted_mask = proc.create_binary_mask(converted_mask, round=False) + + # Remove aliasing artifacts using a gaussian blur. + converted_mask = converted_mask.filter(ImageFilter.GaussianBlur(radius=4)) + + # Expand the mask to fit the whole image if needed. + if paste_to is not None: + converted_mask = proc.uncrop(converted_mask, + (width, height), + paste_to) + + for i, overlay_image in enumerate(overlay_images): + masks_for_overlay[i] = converted_mask + + image_masked = Image.new('RGBa', (overlay_image.width, overlay_image.height)) + image_masked.paste(overlay_image.convert("RGBA").convert("RGBa"), + mask=ImageOps.invert(converted_mask.convert('L'))) + + overlay_images[i] = image_masked.convert('RGBA') + # ------------------- Constants ------------------- From 49bbf1140731036875573bb7c44aa7e74623c856 Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Mon, 4 Dec 2023 19:47:40 -0700 Subject: [PATCH 215/342] Fixed unused import. --- modules/processing.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/processing.py b/modules/processing.py index 71bb056a2..e1823ac33 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -9,7 +9,7 @@ from dataclasses import dataclass, field import torch import numpy as np -from PIL import Image, ImageOps, ImageFilter +from PIL import Image, ImageOps import random import cv2 from skimage import exposure From 120a84bd2f01ec4489bd12bd68f319798ef30782 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Tue, 5 Dec 2023 07:15:39 +0300 Subject: [PATCH 216/342] Merge pull request #14203 from AUTOMATIC1111/remove-clean_text() remove clean_text() --- modules/styles.py | 23 +++-------------------- 1 file changed, 3 insertions(+), 20 deletions(-) diff --git a/modules/styles.py b/modules/styles.py index 4d218cd7e..7fb6c2e11 100644 --- a/modules/styles.py +++ b/modules/styles.py @@ -2,7 +2,6 @@ import csv import fnmatch import os import os.path -import re import typing import shutil @@ -14,22 +13,6 @@ class PromptStyle(typing.NamedTuple): path: str = None -def clean_text(text: str) -> str: - """ - Iterating through a list of regular expressions and replacement strings, we - clean up the prompt and style text to make it easier to match against each - other. - """ - re_list = [ - ("multiple commas", re.compile("(,+\s+)+,?"), ", "), - ("multiple spaces", re.compile("\s{2,}"), " "), - ] - for _, regex, replace in re_list: - text = regex.sub(replace, text) - - return text.strip(", ") - - def merge_prompts(style_prompt: str, prompt: str) -> str: if "{prompt}" in style_prompt: res = style_prompt.replace("{prompt}", prompt) @@ -44,7 +27,7 @@ def apply_styles_to_prompt(prompt, styles): for style in styles: prompt = merge_prompts(style, prompt) - return clean_text(prompt) + return prompt def unwrap_style_text_from_prompt(style_text, prompt): @@ -56,8 +39,8 @@ def unwrap_style_text_from_prompt(style_text, prompt): Note that the "cleaned" version of the style text is only used for matching purposes here. It isn't returned; the original style text is not modified. """ - stripped_prompt = clean_text(prompt) - stripped_style_text = clean_text(style_text) + stripped_prompt = prompt + stripped_style_text = style_text if "{prompt}" in stripped_style_text: # Work out whether the prompt is wrapped in the style text. If so, we # return True and the "inner" prompt text that isn't part of the style. From 895456c4a2e87f5fe3ee23b4482e68fce317a1ca Mon Sep 17 00:00:00 2001 From: Jabasukuriputo Wang Date: Tue, 5 Dec 2023 18:00:48 -0600 Subject: [PATCH 217/342] change state dict comparison to ref compare --- modules/sd_disable_initialization.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/sd_disable_initialization.py b/modules/sd_disable_initialization.py index 8863107ae..273a7edd8 100644 --- a/modules/sd_disable_initialization.py +++ b/modules/sd_disable_initialization.py @@ -215,7 +215,7 @@ class LoadStateDictOnMeta(ReplaceHelper): would be on the meta device. """ - if state_dict == sd: + if state_dict is sd: state_dict = {k: v.to(device="meta", dtype=v.dtype) for k, v in state_dict.items()} original(module, state_dict, strict=strict) From 672dc4efa8e0da38426b121e7c7216d0a8e465fd Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Wed, 6 Dec 2023 15:16:10 +0800 Subject: [PATCH 218/342] Fix forced reload --- modules/sd_models.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/sd_models.py b/modules/sd_models.py index dcf816b3a..d0046f88c 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -801,7 +801,7 @@ def reload_model_weights(sd_model=None, info=None, forced_reload=False): if check_fp8(sd_model) != devices.fp8: # load from state dict again to prevent extra numerical errors forced_reload = True - elif sd_model.sd_model_checkpoint == checkpoint_info.filename: + elif sd_model.sd_model_checkpoint == checkpoint_info.filename and not forced_reload: return sd_model sd_model = reuse_model_from_already_loaded(sd_model, checkpoint_info, timer) From 4d56383025f2cbd00dc6296161e31a896624ab75 Mon Sep 17 00:00:00 2001 From: "fuchen.ljl" Date: Wed, 6 Dec 2023 20:23:56 +0800 Subject: [PATCH 219/342] Long distance memory overflow issue Problem: The memory will slowly increase with the drawing until restarting. Observation: GC analysis shows that no occupation has occurred, so it is suspected to be a problem with the underlying allocator. Reason: Under Linux, glibc is used to allocate memory. glibc uses brk and mmap to allocate memory, and the memory allocated by brk cannot be released until the high-address memory is released. That is to say, if you apply for two pieces of memory A and B through brk, it is impossible to release A before B is released, and it is still occupied by the process. Check the suspected "memory leak" through TOP. So I replaced TCMalloc, but found that libtcmalloc_minimal could not find ptthread_Key_Create. After analysis, it was found that pthread was not entered during compilation. --- webui.sh | 30 +++++++++++++++++++++++------- 1 file changed, 23 insertions(+), 7 deletions(-) diff --git a/webui.sh b/webui.sh index 3d0f87eed..081624c4e 100755 --- a/webui.sh +++ b/webui.sh @@ -222,13 +222,29 @@ fi # Try using TCMalloc on Linux prepare_tcmalloc() { if [[ "${OSTYPE}" == "linux"* ]] && [[ -z "${NO_TCMALLOC}" ]] && [[ -z "${LD_PRELOAD}" ]]; then - TCMALLOC="$(PATH=/usr/sbin:$PATH ldconfig -p | grep -Po "libtcmalloc(_minimal|)\.so\.\d" | head -n 1)" - if [[ ! -z "${TCMALLOC}" ]]; then - echo "Using TCMalloc: ${TCMALLOC}" - export LD_PRELOAD="${TCMALLOC}" - else - printf "\e[1m\e[31mCannot locate TCMalloc (improves CPU memory usage)\e[0m\n" - fi + # Define Tcmalloc Libs arrays + TCMALLOC_LIBS=("libtcmalloc(_minimal|)\.so\.\d" "libtcmalloc\.so\.\d") + + # Traversal array + for lib in "${TCMALLOC_LIBS[@]}" + do + #Determine which type of tcmalloc library the library supports + TCMALLOC="$(PATH=/usr/sbin:$PATH ldconfig -p | grep -P $lib | head -n 1)" + TC_INFO=(${TCMALLOC//=>/}) + if [[ ! -z "${TC_INFO}" ]]; then + echo "Using TCMalloc: ${TC_INFO}" + #Determine if the library is linked to libptthread and resolve undefined symbol: ptthread_Key_Create + if ldd ${TC_INFO[2]} | grep -q 'libpthread'; then + echo "$TC_INFO is linked with libpthread,execute LD_PRELOAD=${TC_INFO}" + export LD_PRELOAD="${TC_INFO}" + break + else + echo "$TC_INFO is not linked with libpthreadand will trigger undefined symbol: ptthread_Key_Create error" + fi + else + printf "\e[1m\e[31mCannot locate TCMalloc (improves CPU memory usage)\e[0m\n" + fi + done fi } From 746783f7a47f38f728f221cc26fe04035d3ca66b Mon Sep 17 00:00:00 2001 From: Nuullll Date: Wed, 6 Dec 2023 20:55:42 +0800 Subject: [PATCH 220/342] [IPEX] Fix embedding Cast `torch.bmm` args into same `dtype`. Fixes the following error when using Text Inversion embedding (#14224): ``` RuntimeError: could not create a primitive descriptor for a matmul primitive ``` --- modules/xpu_specific.py | 3 +++ 1 file changed, 3 insertions(+) diff --git a/modules/xpu_specific.py b/modules/xpu_specific.py index d933c7903..ec1ad100a 100644 --- a/modules/xpu_specific.py +++ b/modules/xpu_specific.py @@ -48,3 +48,6 @@ if has_xpu: CondFunc('torch.nn.modules.conv.Conv2d.forward', lambda orig_func, self, input: orig_func(self, input.to(self.weight.data.dtype)), lambda orig_func, self, input: input.dtype != self.weight.data.dtype) + CondFunc('torch.bmm', + lambda orig_func, input, mat2, out=None: orig_func(input.to(mat2.dtype), mat2, out=out), + lambda orig_func, input, mat2, out=None: input.dtype != mat2.dtype) From 9d2cbf8e97832662e446145d3961c39e78919d3d Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Wed, 6 Dec 2023 23:06:32 +0900 Subject: [PATCH 221/342] add option: Live preview in full page image viewer make #13459 "show the preview image in the modal view if available" optional --- javascript/imageviewer.js | 2 +- modules/shared_options.py | 1 + 2 files changed, 2 insertions(+), 1 deletion(-) diff --git a/javascript/imageviewer.js b/javascript/imageviewer.js index e4dae91bc..625c5d148 100644 --- a/javascript/imageviewer.js +++ b/javascript/imageviewer.js @@ -34,7 +34,7 @@ function updateOnBackgroundChange() { if (modalImage && modalImage.offsetParent) { let currentButton = selected_gallery_button(); let preview = gradioApp().querySelectorAll('.livePreview > img'); - if (preview.length > 0) { + if (opts.js_live_preview_in_modal_lightbox && preview.length > 0) { // show preview image if available modalImage.src = preview[preview.length - 1].src; } else if (currentButton?.children?.length > 0 && modalImage.src != currentButton.children[0].src) { diff --git a/modules/shared_options.py b/modules/shared_options.py index e5de0d018..88cfddede 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -330,6 +330,7 @@ options_templates.update(options_section(('ui', "Live previews", "ui"), { "live_preview_content": OptionInfo("Prompt", "Live preview subject", gr.Radio, {"choices": ["Combined", "Prompt", "Negative prompt"]}), "live_preview_refresh_period": OptionInfo(1000, "Progressbar and preview update period").info("in milliseconds"), "live_preview_fast_interrupt": OptionInfo(False, "Return image with chosen live preview method on interrupt").info("makes interrupts faster"), + "js_live_preview_in_modal_lightbox": OptionInfo(True, "Show Live preview in full page image viewer"), })) options_templates.update(options_section(('sampler-params', "Sampler parameters", "sd"), { From e90d4334ad37024a802f4ef27069b625a6508f72 Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Wed, 6 Dec 2023 16:54:42 -0700 Subject: [PATCH 222/342] A custom blending function can be provided by p, replacing the use of soft_inpainting. --- modules/sd_samplers_cfg_denoiser.py | 34 ++++++++++++++--------------- modules/sd_samplers_common.py | 1 - 2 files changed, 17 insertions(+), 18 deletions(-) diff --git a/modules/sd_samplers_cfg_denoiser.py b/modules/sd_samplers_cfg_denoiser.py index a700e6922..f13e8dcc5 100644 --- a/modules/sd_samplers_cfg_denoiser.py +++ b/modules/sd_samplers_cfg_denoiser.py @@ -6,7 +6,6 @@ import modules.shared as shared from modules.script_callbacks import CFGDenoiserParams, cfg_denoiser_callback from modules.script_callbacks import CFGDenoisedParams, cfg_denoised_callback from modules.script_callbacks import AfterCFGCallbackParams, cfg_after_cfg_callback -import modules.soft_inpainting as si def catenate_conds(conds): @@ -44,7 +43,6 @@ class CFGDenoiser(torch.nn.Module): self.model_wrap = None self.mask = None self.nmask = None - self.soft_inpainting: si.SoftInpaintingParameters = None self.init_latent = None self.steps = None """number of steps as specified by user in UI""" @@ -94,7 +92,6 @@ class CFGDenoiser(torch.nn.Module): self.sampler.sampler_extra_args['uncond'] = uc def forward(self, x, sigma, uncond, cond, cond_scale, s_min_uncond, image_cond): - if state.interrupted or state.skipped: raise sd_samplers_common.InterruptedException @@ -111,15 +108,24 @@ class CFGDenoiser(torch.nn.Module): assert not is_edit_model or all(len(conds) == 1 for conds in conds_list), "AND is not supported for InstructPix2Pix checkpoint (unless using Image CFG scale = 1.0)" + # If we use masks, blending between the denoised and original latent images occurs here. + def apply_blend(latent): + if hasattr(self.p, "denoiser_masked_blend_function") and callable(self.p.denoiser_masked_blend_function): + return self.p.denoiser_masked_blend_function( + self, + # Using an argument dictionary so that arguments can be added without breaking extensions. + args= + { + "denoiser": self, + "current_latent": latent, + "sigma": sigma + }) + else: + return self.init_latent * self.mask + self.nmask * latent + # Blend in the original latents (before) if self.mask_before_denoising and self.mask is not None: - if self.soft_inpainting is None: - x = self.init_latent * self.mask + self.nmask * x - else: - x = si.latent_blend(self.soft_inpainting, - self.init_latent, - x, - si.get_modified_nmask(self.soft_inpainting, self.nmask, sigma)) + x = apply_blend(x) batch_size = len(conds_list) repeats = [len(conds_list[i]) for i in range(batch_size)] @@ -222,13 +228,7 @@ class CFGDenoiser(torch.nn.Module): # Blend in the original latents (after) if not self.mask_before_denoising and self.mask is not None: - if self.soft_inpainting is None: - denoised = self.init_latent * self.mask + self.nmask * denoised - else: - denoised = si.latent_blend(self.soft_inpainting, - self.init_latent, - denoised, - si.get_modified_nmask(self.soft_inpainting, self.nmask, sigma)) + denoised = apply_blend(denoised) self.sampler.last_latent = self.get_pred_x0(torch.cat([x_in[i:i + 1] for i in denoised_image_indexes]), torch.cat([x_out[i:i + 1] for i in denoised_image_indexes]), sigma) diff --git a/modules/sd_samplers_common.py b/modules/sd_samplers_common.py index 9682bee3d..58efcad23 100644 --- a/modules/sd_samplers_common.py +++ b/modules/sd_samplers_common.py @@ -277,7 +277,6 @@ class Sampler: self.model_wrap_cfg.p = p self.model_wrap_cfg.mask = p.mask if hasattr(p, 'mask') else None self.model_wrap_cfg.nmask = p.nmask if hasattr(p, 'nmask') else None - self.model_wrap_cfg.soft_inpainting = p.soft_inpainting if hasattr(p, 'soft_inpainting') else None self.model_wrap_cfg.step = 0 self.model_wrap_cfg.image_cfg_scale = getattr(p, 'image_cfg_scale', None) self.eta = p.eta if p.eta is not None else getattr(opts, self.eta_option_field, 0.0) From 4608f6236fc24d937f89500b2c9bf48484537cf9 Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Wed, 6 Dec 2023 18:11:17 -0700 Subject: [PATCH 223/342] Removed changes in some scripts since the arguments for soft painting are no longer passed through the same path as "mask_blur". --- modules/img2img.py | 50 +------------------------------- modules/ui.py | 7 ----- scripts/outpainting_mk_2.py | 9 ++---- scripts/poor_mans_outpainting.py | 8 ++--- test/test_img2img.py | 5 ---- 5 files changed, 5 insertions(+), 74 deletions(-) diff --git a/modules/img2img.py b/modules/img2img.py index 3aa8a9cef..c583290a0 100644 --- a/modules/img2img.py +++ b/modules/img2img.py @@ -15,7 +15,6 @@ import modules.shared as shared import modules.processing as processing from modules.ui import plaintext_to_html import modules.scripts -import modules.soft_inpainting as si def process_batch(p, input_dir, output_dir, inpaint_mask_dir, args, to_scale=False, scale_by=1.0, use_png_info=False, png_info_props=None, png_info_dir=None): @@ -147,48 +146,7 @@ def process_batch(p, input_dir, output_dir, inpaint_mask_dir, args, to_scale=Fal return batch_results -def img2img(id_task: str, - mode: int, - prompt: str, - negative_prompt: str, - prompt_styles, - init_img, - sketch, - init_img_with_mask, - inpaint_color_sketch, - inpaint_color_sketch_orig, - init_img_inpaint, - init_mask_inpaint, - steps: int, - sampler_name: str, - mask_blur: int, - mask_alpha: float, - mask_blend_enabled: bool, - mask_blend_power: float, - mask_blend_scale: float, - inpaint_detail_preservation: float, - inpainting_fill: int, - n_iter: int, - batch_size: int, - cfg_scale: float, - image_cfg_scale: float, - denoising_strength: float, - selected_scale_tab: int, - height: int, - width: int, - scale_by: float, - resize_mode: int, - inpaint_full_res: bool, - inpaint_full_res_padding: int, - inpainting_mask_invert: int, - img2img_batch_input_dir: str, - img2img_batch_output_dir: str, - img2img_batch_inpaint_mask_dir: str, - override_settings_texts, - img2img_batch_use_png_info: bool, - img2img_batch_png_info_props: list, - img2img_batch_png_info_dir: str, - request: gr.Request, *args): +def img2img(id_task: str, mode: int, prompt: str, negative_prompt: str, prompt_styles, init_img, sketch, init_img_with_mask, inpaint_color_sketch, inpaint_color_sketch_orig, init_img_inpaint, init_mask_inpaint, steps: int, sampler_name: str, mask_blur: int, mask_alpha: float, inpainting_fill: int, n_iter: int, batch_size: int, cfg_scale: float, image_cfg_scale: float, denoising_strength: float, selected_scale_tab: int, height: int, width: int, scale_by: float, resize_mode: int, inpaint_full_res: bool, inpaint_full_res_padding: int, inpainting_mask_invert: int, img2img_batch_input_dir: str, img2img_batch_output_dir: str, img2img_batch_inpaint_mask_dir: str, override_settings_texts, img2img_batch_use_png_info: bool, img2img_batch_png_info_props: list, img2img_batch_png_info_dir: str, request: gr.Request, *args): override_settings = create_override_settings_dict(override_settings_texts) is_batch = mode == 5 @@ -229,9 +187,6 @@ def img2img(id_task: str, assert 0. <= denoising_strength <= 1., 'can only work with strength in [0.0, 1.0]' - soft_inpainting = si.SoftInpaintingSettings(mask_blend_power, mask_blend_scale, inpaint_detail_preservation) \ - if mask_blend_enabled else None - p = StableDiffusionProcessingImg2Img( sd_model=shared.sd_model, outpath_samples=opts.outdir_samples or opts.outdir_img2img_samples, @@ -249,7 +204,6 @@ def img2img(id_task: str, init_images=[image], mask=mask, mask_blur=mask_blur, - soft_inpainting=soft_inpainting, inpainting_fill=inpainting_fill, resize_mode=resize_mode, denoising_strength=denoising_strength, @@ -270,8 +224,6 @@ def img2img(id_task: str, if mask: p.extra_generation_params["Mask blur"] = mask_blur - if soft_inpainting is not None: - soft_inpainting.add_generation_params(p.extra_generation_params) with closing(p): if is_batch: diff --git a/modules/ui.py b/modules/ui.py index bd2091e1f..d80486dd4 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -29,7 +29,6 @@ import modules.shared as shared from modules import prompt_parser from modules.sd_hijack import model_hijack from modules.generation_parameters_copypaste import image_from_url_text -import modules.soft_inpainting as si create_setting_component = ui_settings.create_setting_component @@ -680,9 +679,6 @@ def create_ui(): mask_blur = gr.Slider(label='Mask blur', minimum=0, maximum=64, step=1, value=4, elem_id="img2img_mask_blur") mask_alpha = gr.Slider(label="Mask transparency", visible=False, elem_id="img2img_mask_alpha") - with FormRow(): - soft_inpainting = si.gradio_ui() - with FormRow(): inpainting_mask_invert = gr.Radio(label='Mask mode', choices=['Inpaint masked', 'Inpaint not masked'], value='Inpaint masked', type="index", elem_id="img2img_mask_mode") @@ -737,7 +733,6 @@ def create_ui(): sampler_name, mask_blur, mask_alpha, - *(soft_inpainting[0]), inpainting_fill, batch_count, batch_size, @@ -836,10 +831,8 @@ def create_ui(): (toprow.ui_styles.dropdown, lambda d: d["Styles array"] if isinstance(d.get("Styles array"), list) else gr.update()), (denoising_strength, "Denoising strength"), (mask_blur, "Mask blur"), - *(soft_inpainting[1]), *scripts.scripts_img2img.infotext_fields ] - parameters_copypaste.add_paste_fields("img2img", init_img, img2img_paste_fields, override_settings) parameters_copypaste.add_paste_fields("inpaint", init_img_with_mask, img2img_paste_fields, override_settings) parameters_copypaste.register_paste_params_button(parameters_copypaste.ParamBinding( diff --git a/scripts/outpainting_mk_2.py b/scripts/outpainting_mk_2.py index f78886883..c98ab4809 100644 --- a/scripts/outpainting_mk_2.py +++ b/scripts/outpainting_mk_2.py @@ -10,7 +10,6 @@ from PIL import Image, ImageDraw from modules import images from modules.processing import Processed, process_images from modules.shared import opts, state -import modules.soft_inpainting as si # this function is taken from https://github.com/parlance-zz/g-diffuser-bot @@ -134,14 +133,13 @@ class Script(scripts.Script): pixels = gr.Slider(label="Pixels to expand", minimum=8, maximum=256, step=8, value=128, elem_id=self.elem_id("pixels")) mask_blur = gr.Slider(label='Mask blur', minimum=0, maximum=64, step=1, value=8, elem_id=self.elem_id("mask_blur")) - soft_inpainting = si.gradio_ui()[0] direction = gr.CheckboxGroup(label="Outpainting direction", choices=['left', 'right', 'up', 'down'], value=['left', 'right', 'up', 'down'], elem_id=self.elem_id("direction")) noise_q = gr.Slider(label="Fall-off exponent (lower=higher detail)", minimum=0.0, maximum=4.0, step=0.01, value=1.0, elem_id=self.elem_id("noise_q")) color_variation = gr.Slider(label="Color variation", minimum=0.0, maximum=1.0, step=0.01, value=0.05, elem_id=self.elem_id("color_variation")) - return [info, pixels, mask_blur, *soft_inpainting, direction, noise_q, color_variation] + return [info, pixels, mask_blur, direction, noise_q, color_variation] - def run(self, p, _, pixels, mask_blur, mask_blend_enabled, mask_blend_power, mask_blend_scale, inpaint_detail_preservation, direction, noise_q, color_variation): + def run(self, p, _, pixels, mask_blur, direction, noise_q, color_variation): initial_seed_and_info = [None, None] process_width = p.width @@ -170,9 +168,6 @@ class Script(scripts.Script): p.mask_blur_x = mask_blur_x*4 p.mask_blur_y = mask_blur_y*4 - p.soft_inpainting = si.SoftInpaintingSettings(mask_blend_power, mask_blend_scale, inpaint_detail_preservation) \ - if mask_blend_enabled else None - init_img = p.init_images[0] target_w = math.ceil((init_img.width + left + right) / 64) * 64 target_h = math.ceil((init_img.height + up + down) / 64) * 64 diff --git a/scripts/poor_mans_outpainting.py b/scripts/poor_mans_outpainting.py index 11f7f74a8..ea0632b68 100644 --- a/scripts/poor_mans_outpainting.py +++ b/scripts/poor_mans_outpainting.py @@ -7,7 +7,6 @@ from PIL import Image, ImageDraw from modules import images, devices from modules.processing import Processed, process_images from modules.shared import opts, state -import modules.soft_inpainting as si class Script(scripts.Script): @@ -23,19 +22,16 @@ class Script(scripts.Script): pixels = gr.Slider(label="Pixels to expand", minimum=8, maximum=256, step=8, value=128, elem_id=self.elem_id("pixels")) mask_blur = gr.Slider(label='Mask blur', minimum=0, maximum=64, step=1, value=4, elem_id=self.elem_id("mask_blur")) - soft_inpainting = si.gradio_ui()[0] inpainting_fill = gr.Radio(label='Masked content', choices=['fill', 'original', 'latent noise', 'latent nothing'], value='fill', type="index", elem_id=self.elem_id("inpainting_fill")) direction = gr.CheckboxGroup(label="Outpainting direction", choices=['left', 'right', 'up', 'down'], value=['left', 'right', 'up', 'down'], elem_id=self.elem_id("direction")) - return [pixels, mask_blur, *soft_inpainting, inpainting_fill, direction] + return [pixels, mask_blur, inpainting_fill, direction] - def run(self, p, pixels, mask_blur, mask_blend_enabled, mask_blend_power, mask_blend_scale, inpaint_detail_preservation, inpainting_fill, direction): + def run(self, p, pixels, mask_blur, inpainting_fill, direction): initial_seed = None initial_info = None p.mask_blur = mask_blur * 2 - p.soft_inpainting = si.SoftInpaintingSettings(mask_blend_power, mask_blend_scale, inpaint_detail_preservation) \ - if mask_blend_enabled else None p.inpainting_fill = inpainting_fill p.inpaint_full_res = False diff --git a/test/test_img2img.py b/test/test_img2img.py index 87bd85091..117d2d1eb 100644 --- a/test/test_img2img.py +++ b/test/test_img2img.py @@ -1,7 +1,6 @@ import pytest import requests -import modules.soft_inpainting as si @pytest.fixture() @@ -25,10 +24,6 @@ def simple_img2img_request(img2img_basic_image_base64): "inpainting_mask_invert": False, "mask": None, "mask_blur": 4, - "mask_blend_enabled": True, - "mask_blend_power": si.default.mask_blend_power, - "mask_blend_scale": si.default.mask_blend_scale, - "inpaint_detail_preservation": si.default.inpaint_detail_preservation, "n_iter": 1, "negative_prompt": "", "override_settings": {}, From ac4578912395627731f2cd8529f87a95df1f7644 Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Wed, 6 Dec 2023 21:16:27 -0700 Subject: [PATCH 224/342] Removed soft inpainting, added hooks for softpainting to work instead. --- modules/processing.py | 94 ++++++++++++----------------- modules/scripts.py | 70 +++++++++++++++++++++ modules/sd_samplers_cfg_denoiser.py | 23 +++---- 3 files changed, 118 insertions(+), 69 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index 7d46949fa..5a1a90afe 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -30,7 +30,6 @@ import modules.sd_models as sd_models import modules.sd_vae as sd_vae from ldm.data.util import AddMiDaS from ldm.models.diffusion.ddpm import LatentDepth2ImageDiffusion -import modules.soft_inpainting as si from einops import repeat, rearrange from blendmodes.blend import blendLayers, BlendType @@ -73,12 +72,10 @@ def uncrop(image, dest_size, paste_loc): return image -def apply_overlay(image, paste_loc, index, overlays): - if overlays is None or index >= len(overlays): +def apply_overlay(image, paste_loc, overlay): + if overlay is None: return image - overlay = overlays[index] - if paste_loc is not None: image = uncrop(image, (overlay.width, overlay.height), paste_loc) @@ -150,7 +147,6 @@ class StableDiffusionProcessing: do_not_save_grid: bool = False extra_generation_params: dict[str, Any] = None overlay_images: list = None - masks_for_overlay: list = None eta: float = None do_not_reload_embeddings: bool = False denoising_strength: float = None @@ -880,31 +876,17 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: with devices.without_autocast() if devices.unet_needs_upcast else devices.autocast(): samples_ddim = p.sample(conditioning=p.c, unconditional_conditioning=p.uc, seeds=p.seeds, subseeds=p.subseeds, subseed_strength=p.subseed_strength, prompts=p.prompts) + if p.scripts is not None: + ps = scripts.PostSampleArgs(samples_ddim) + p.scripts.post_sample(p, ps) + samples_ddim = pp.samples + if getattr(samples_ddim, 'already_decoded', False): x_samples_ddim = samples_ddim - # todo: generate adaptive masks based on pixel differences. - if getattr(p, "image_mask", None) is not None and getattr(p, "soft_inpainting", None) is not None: - si.apply_masks(soft_inpainting=p.soft_inpainting, - nmask=p.nmask, - overlay_images=p.overlay_images, - masks_for_overlay=p.masks_for_overlay, - width=p.width, - height=p.height, - paste_to=p.paste_to) else: if opts.sd_vae_decode_method != 'Full': p.extra_generation_params['VAE Decoder'] = opts.sd_vae_decode_method - # Generate the mask(s) based on similarity between the original and denoised latent vectors - if getattr(p, "image_mask", None) is not None and getattr(p, "soft_inpainting", None) is not None: - si.apply_adaptive_masks(latent_orig=p.init_latent, - latent_processed=samples_ddim, - overlay_images=p.overlay_images, - masks_for_overlay=p.masks_for_overlay, - width=p.width, - height=p.height, - paste_to=p.paste_to) - x_samples_ddim = decode_latent_batch(p.sd_model, samples_ddim, target_device=devices.cpu, check_for_nans=True) x_samples_ddim = torch.stack(x_samples_ddim).float() @@ -955,9 +937,18 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: pp = scripts.PostprocessImageArgs(image) p.scripts.postprocess_image(p, pp) image = pp.image + + mask_for_overlay = p.mask_for_overlay + overlay_image = p.overlay_images[i] if p.overlay_images is not None and i < len(p.overlay_images) else None + + if p.scripts is not None: + ppmo = scripts.PostProcessMaskOverlayArgs(i, mask_for_overlay, overlay_image) + p.scripts.postprocess_maskoverlay(p, ppmo) + mask_for_overlay, overlay_image = pp.mask_for_overlay, pp.overlay_image + if p.color_corrections is not None and i < len(p.color_corrections): if save_samples and opts.save_images_before_color_correction: - image_without_cc = apply_overlay(image, p.paste_to, i, p.overlay_images) + image_without_cc = apply_overlay(image, p.paste_to, overlay_image) images.save_image(image_without_cc, p.outpath_samples, "", p.seeds[i], p.prompts[i], opts.samples_format, info=infotext(i), p=p, suffix="-before-color-correction") image = apply_color_correction(p.color_corrections[i], image) @@ -968,9 +959,9 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: original_denoised_image = image.copy() if p.paste_to is not None: - original_denoised_image = uncrop(original_denoised_image, (p.overlay_images[i].width, p.overlay_images[i].height), p.paste_to) + original_denoised_image = uncrop(original_denoised_image, (p.overlay_image.width, p.overlay_image.height), p.paste_to) - image = apply_overlay(image, p.paste_to, i, p.overlay_images) + image = apply_overlay(image, p.paste_to, overlay_image) if save_samples: images.save_image(image, p.outpath_samples, "", p.seeds[i], p.prompts[i], opts.samples_format, info=infotext(i), p=p) @@ -981,13 +972,6 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: image.info["parameters"] = text output_images.append(image) - if hasattr(p, 'mask_for_overlay') and p.mask_for_overlay: - mask_for_overlay = p.mask_for_overlay - elif hasattr(p, 'masks_for_overlay') and p.masks_for_overlay and p.masks_for_overlay[i]: - mask_for_overlay = p.masks_for_overlay[i] - else: - mask_for_overlay = None - if mask_for_overlay is not None: if opts.return_mask or opts.save_mask: image_mask = mask_for_overlay.convert('RGB') @@ -1401,7 +1385,7 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): mask_blur_x: int = 4 mask_blur_y: int = 4 mask_blur: int = None - soft_inpainting: si.SoftInpaintingParameters = si.default + mask_round: bool = True inpainting_fill: int = 0 inpaint_full_res: bool = True inpaint_full_res_padding: int = 0 @@ -1447,7 +1431,7 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): if image_mask is not None: # image_mask is passed in as RGBA by Gradio to support alpha masks, # but we still want to support binary masks. - image_mask = create_binary_mask(image_mask, round=(self.soft_inpainting is None)) + image_mask = create_binary_mask(image_mask, round=self.mask_round) if self.inpainting_mask_invert: image_mask = ImageOps.invert(image_mask) @@ -1465,7 +1449,7 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): image_mask = Image.fromarray(np_mask) if self.inpaint_full_res: - self.mask_for_overlay = image_mask if self.soft_inpainting is None else None + self.mask_for_overlay = image_mask mask = image_mask.convert('L') crop_region = masking.get_crop_region(np.array(mask), self.inpaint_full_res_padding) crop_region = masking.expand_crop_region(crop_region, self.width, self.height, mask.width, mask.height) @@ -1476,13 +1460,10 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): self.paste_to = (x1, y1, x2-x1, y2-y1) else: image_mask = images.resize_image(self.resize_mode, image_mask, self.width, self.height) + np_mask = np.array(image_mask) + np_mask = np.clip((np_mask.astype(np.float32)) * 2, 0, 255).astype(np.uint8) + self.mask_for_overlay = Image.fromarray(np_mask) - if self.soft_inpainting is None: - np_mask = np.array(image_mask) - np_mask = np.clip((np_mask.astype(np.float32)) * 2, 0, 255).astype(np.uint8) - self.mask_for_overlay = Image.fromarray(np_mask) - - self.masks_for_overlay = [] if self.soft_inpainting is not None else None self.overlay_images = [] latent_mask = self.latent_mask if self.latent_mask is not None else image_mask @@ -1504,15 +1485,10 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): image = images.resize_image(self.resize_mode, image, self.width, self.height) if image_mask is not None: - if self.soft_inpainting is not None: - # We apply the masks AFTER to adjust mask based on changed content. - self.overlay_images.append(image.convert('RGBA')) - self.masks_for_overlay.append(image_mask) - else: - image_masked = Image.new('RGBa', (image.width, image.height)) - image_masked.paste(image.convert("RGBA").convert("RGBa"), mask=ImageOps.invert(self.mask_for_overlay.convert('L'))) + image_masked = Image.new('RGBa', (image.width, image.height)) + image_masked.paste(image.convert("RGBA").convert("RGBa"), mask=ImageOps.invert(self.mask_for_overlay.convert('L'))) - self.overlay_images.append(image_masked.convert('RGBA')) + self.overlay_images.append(image_masked.convert('RGBA')) # crop_region is not None if we are doing inpaint full res if crop_region is not None: @@ -1565,7 +1541,7 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): latmask = init_mask.convert('RGB').resize((self.init_latent.shape[3], self.init_latent.shape[2])) latmask = np.moveaxis(np.array(latmask, dtype=np.float32), 2, 0) / 255 latmask = latmask[0] - if self.soft_inpainting is None: + if self.mask_round: latmask = np.around(latmask) latmask = np.tile(latmask[None], (4, 1, 1)) @@ -1578,7 +1554,7 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): elif self.inpainting_fill == 3: self.init_latent = self.init_latent * self.mask - self.image_conditioning = self.img2img_image_conditioning(image * 2 - 1, self.init_latent, image_mask, self.soft_inpainting is None) + self.image_conditioning = self.img2img_image_conditioning(image * 2 - 1, self.init_latent, image_mask, self.mask_round) def sample(self, conditioning, unconditional_conditioning, seeds, subseeds, subseed_strength, prompts): x = self.rng.next() @@ -1589,8 +1565,14 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): samples = self.sampler.sample_img2img(self, self.init_latent, x, conditioning, unconditional_conditioning, image_conditioning=self.image_conditioning) - if self.mask is not None and self.soft_inpainting is None: - samples = samples * self.nmask + self.init_latent * self.mask + blended_samples = samples * self.nmask + self.init_latent * self.mask + + if self.scripts is not None: + mba = scripts.MaskBlendArgs(self, samples, self.nmask, self.init_latent, self.mask, blended_samples, sigma=None, is_final_blend=True) + self.scripts.on_mask_blend(self, mba) + blended_samples = mba.blended_latent + + samples = blended_samples del x devices.torch_gc() diff --git a/modules/scripts.py b/modules/scripts.py index 7f9454eb5..92a07c564 100644 --- a/modules/scripts.py +++ b/modules/scripts.py @@ -11,11 +11,31 @@ from modules import shared, paths, script_callbacks, extensions, script_loading, AlwaysVisible = object() +class MaskBlendArgs: + def __init__(self, current_latent, nmask, init_latent, mask, blended_samples, denoiser=None, sigma=None): + self.current_latent = current_latent + self.nmask = nmask + self.init_latent = init_latent + self.mask = mask + self.blended_samples = blended_samples + + self.denoiser = denoiser + self.is_final_blend = denoiser is None + self.sigma = sigma + +class PostSampleArgs: + def __init__(self, samples): + self.samples = samples class PostprocessImageArgs: def __init__(self, image): self.image = image +class PostProcessMaskOverlayArgs: + def __init__(self, index, mask_for_overlay, overlay_image): + self.index = index + self.mask_for_overlay = mask_for_overlay + self.overlay_image = overlay_image class PostprocessBatchListArgs: def __init__(self, images): @@ -206,6 +226,25 @@ class Script: pass + def on_mask_blend(self, p, mba: MaskBlendArgs, *args): + """ + Called in inpainting mode when the original content is blended with the inpainted content. + This is called at every step in the denoising process and once at the end. + If is_final_blend is true, this is called for the final blending stage. + Otherwise, denoiser and sigma are defined and may be used to inform the procedure. + """ + + pass + + def post_sample(self, p, ps: PostSampleArgs, *args): + """ + Called after the samples have been generated, + but before they have been decoded by the VAE, if applicable. + Check getattr(samples, 'already_decoded', False) to test if the images are decoded. + """ + + pass + def postprocess_image(self, p, pp: PostprocessImageArgs, *args): """ Called for every image after it has been generated. @@ -213,6 +252,13 @@ class Script: pass + def postprocess_maskoverlay(self, p, ppmo: PostProcessMaskOverlayArgs, *args): + """ + Called for every image after it has been generated. + """ + + pass + def postprocess(self, p, processed, *args): """ This function is called after processing ends for AlwaysVisible scripts. @@ -767,6 +813,22 @@ class ScriptRunner: except Exception: errors.report(f"Error running postprocess_batch_list: {script.filename}", exc_info=True) + def post_sample(self, p, ps: PostSampleArgs): + for script in self.alwayson_scripts: + try: + script_args = p.script_args[script.args_from:script.args_to] + script.post_sample(p, ps, *script_args) + except Exception: + errors.report(f"Error running post_sample: {script.filename}", exc_info=True) + + def on_mask_blend(self, p, mba: MaskBlendArgs): + for script in self.alwayson_scripts: + try: + script_args = p.script_args[script.args_from:script.args_to] + script.on_mask_blend(p, mba, *script_args) + except Exception: + errors.report(f"Error running post_sample: {script.filename}", exc_info=True) + def postprocess_image(self, p, pp: PostprocessImageArgs): for script in self.alwayson_scripts: try: @@ -775,6 +837,14 @@ class ScriptRunner: except Exception: errors.report(f"Error running postprocess_image: {script.filename}", exc_info=True) + def postprocess_maskoverlay(self, p, ppmo: PostProcessMaskOverlayArgs): + for script in self.alwayson_scripts: + try: + script_args = p.script_args[script.args_from:script.args_to] + script.postprocess_maskoverlay(p, ppmo, *script_args) + except Exception: + errors.report(f"Error running postprocess_image: {script.filename}", exc_info=True) + def before_component(self, component, **kwargs): for callback, script in self.on_before_component_elem_id.get(kwargs.get("elem_id"), []): try: diff --git a/modules/sd_samplers_cfg_denoiser.py b/modules/sd_samplers_cfg_denoiser.py index f13e8dcc5..eb9d5dafa 100644 --- a/modules/sd_samplers_cfg_denoiser.py +++ b/modules/sd_samplers_cfg_denoiser.py @@ -109,19 +109,16 @@ class CFGDenoiser(torch.nn.Module): assert not is_edit_model or all(len(conds) == 1 for conds in conds_list), "AND is not supported for InstructPix2Pix checkpoint (unless using Image CFG scale = 1.0)" # If we use masks, blending between the denoised and original latent images occurs here. - def apply_blend(latent): - if hasattr(self.p, "denoiser_masked_blend_function") and callable(self.p.denoiser_masked_blend_function): - return self.p.denoiser_masked_blend_function( - self, - # Using an argument dictionary so that arguments can be added without breaking extensions. - args= - { - "denoiser": self, - "current_latent": latent, - "sigma": sigma - }) - else: - return self.init_latent * self.mask + self.nmask * latent + def apply_blend(current_latent): + blended_latent = current_latent * self.nmask + self.init_latent * self.mask + + if self.p.scripts is not None: + from modules import scripts + mba = scripts.MaskBlendArgs(current_latent, self.nmask, self.init_latent, self.mask, blended_latent, denoiser=self, sigma=sigma) + self.p.scripts.on_mask_blend(self.p, mba) + blended_latent = mba.blended_latent + + return blended_latent # Blend in the original latents (before) if self.mask_before_denoising and self.mask is not None: From 2abc417834d752e43a283f8603bfddfb1c80b30f Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Wed, 6 Dec 2023 22:25:53 -0700 Subject: [PATCH 225/342] Re-implemented soft inpainting via a script. Also fixed some mistakes with the previous hooks, removed unnecessary formatting changes, removed code that I had forgotten to. --- modules/processing.py | 23 +-- modules/scripts.py | 4 +- modules/soft_inpainting.py | 308 ---------------------------- scripts/soft_inpainting.py | 401 +++++++++++++++++++++++++++++++++++++ 4 files changed, 413 insertions(+), 323 deletions(-) delete mode 100644 modules/soft_inpainting.py create mode 100644 scripts/soft_inpainting.py diff --git a/modules/processing.py b/modules/processing.py index 5a1a90afe..f8d85bdf5 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -879,14 +879,13 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: if p.scripts is not None: ps = scripts.PostSampleArgs(samples_ddim) p.scripts.post_sample(p, ps) - samples_ddim = pp.samples + samples_ddim = ps.samples if getattr(samples_ddim, 'already_decoded', False): x_samples_ddim = samples_ddim else: if opts.sd_vae_decode_method != 'Full': p.extra_generation_params['VAE Decoder'] = opts.sd_vae_decode_method - x_samples_ddim = decode_latent_batch(p.sd_model, samples_ddim, target_device=devices.cpu, check_for_nans=True) x_samples_ddim = torch.stack(x_samples_ddim).float() @@ -944,7 +943,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: if p.scripts is not None: ppmo = scripts.PostProcessMaskOverlayArgs(i, mask_for_overlay, overlay_image) p.scripts.postprocess_maskoverlay(p, ppmo) - mask_for_overlay, overlay_image = pp.mask_for_overlay, pp.overlay_image + mask_for_overlay, overlay_image = ppmo.mask_for_overlay, ppmo.overlay_image if p.color_corrections is not None and i < len(p.color_corrections): if save_samples and opts.save_images_before_color_correction: @@ -959,7 +958,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: original_denoised_image = image.copy() if p.paste_to is not None: - original_denoised_image = uncrop(original_denoised_image, (p.overlay_image.width, p.overlay_image.height), p.paste_to) + original_denoised_image = uncrop(original_denoised_image, (overlay_image.width, overlay_image.height), p.paste_to) image = apply_overlay(image, p.paste_to, overlay_image) @@ -1512,9 +1511,6 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): if self.overlay_images is not None: self.overlay_images = self.overlay_images * self.batch_size - if self.masks_for_overlay is not None: - self.masks_for_overlay = self.masks_for_overlay * self.batch_size - if self.color_corrections is not None and len(self.color_corrections) == 1: self.color_corrections = self.color_corrections * self.batch_size @@ -1565,14 +1561,15 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): samples = self.sampler.sample_img2img(self, self.init_latent, x, conditioning, unconditional_conditioning, image_conditioning=self.image_conditioning) - blended_samples = samples * self.nmask + self.init_latent * self.mask + if self.mask is not None: + blended_samples = samples * self.nmask + self.init_latent * self.mask - if self.scripts is not None: - mba = scripts.MaskBlendArgs(self, samples, self.nmask, self.init_latent, self.mask, blended_samples, sigma=None, is_final_blend=True) - self.scripts.on_mask_blend(self, mba) - blended_samples = mba.blended_latent + if self.scripts is not None: + mba = scripts.MaskBlendArgs(samples, self.nmask, self.init_latent, self.mask, blended_samples) + self.scripts.on_mask_blend(self, mba) + blended_samples = mba.blended_latent - samples = blended_samples + samples = blended_samples del x devices.torch_gc() diff --git a/modules/scripts.py b/modules/scripts.py index 92a07c564..b6fcf96e0 100644 --- a/modules/scripts.py +++ b/modules/scripts.py @@ -12,12 +12,12 @@ from modules import shared, paths, script_callbacks, extensions, script_loading, AlwaysVisible = object() class MaskBlendArgs: - def __init__(self, current_latent, nmask, init_latent, mask, blended_samples, denoiser=None, sigma=None): + def __init__(self, current_latent, nmask, init_latent, mask, blended_latent, denoiser=None, sigma=None): self.current_latent = current_latent self.nmask = nmask self.init_latent = init_latent self.mask = mask - self.blended_samples = blended_samples + self.blended_latent = blended_latent self.denoiser = denoiser self.is_final_blend = denoiser is None diff --git a/modules/soft_inpainting.py b/modules/soft_inpainting.py deleted file mode 100644 index b36ac8fa1..000000000 --- a/modules/soft_inpainting.py +++ /dev/null @@ -1,308 +0,0 @@ -class SoftInpaintingSettings: - def __init__(self, mask_blend_power, mask_blend_scale, inpaint_detail_preservation): - self.mask_blend_power = mask_blend_power - self.mask_blend_scale = mask_blend_scale - self.inpaint_detail_preservation = inpaint_detail_preservation - - def add_generation_params(self, dest): - dest[enabled_gen_param_label] = True - dest[gen_param_labels.mask_blend_power] = self.mask_blend_power - dest[gen_param_labels.mask_blend_scale] = self.mask_blend_scale - dest[gen_param_labels.inpaint_detail_preservation] = self.inpaint_detail_preservation - - -# ------------------- Methods ------------------- - - -def latent_blend(soft_inpainting, a, b, t): - """ - Interpolates two latent image representations according to the parameter t, - where the interpolated vectors' magnitudes are also interpolated separately. - The "detail_preservation" factor biases the magnitude interpolation towards - the larger of the two magnitudes. - """ - import torch - - # NOTE: We use inplace operations wherever possible. - - # [4][w][h] to [1][4][w][h] - t2 = t.unsqueeze(0) - # [4][w][h] to [1][1][w][h] - the [4] seem redundant. - t3 = t[0].unsqueeze(0).unsqueeze(0) - - one_minus_t2 = 1 - t2 - one_minus_t3 = 1 - t3 - - # Linearly interpolate the image vectors. - a_scaled = a * one_minus_t2 - b_scaled = b * t2 - image_interp = a_scaled - image_interp.add_(b_scaled) - result_type = image_interp.dtype - del a_scaled, b_scaled, t2, one_minus_t2 - - # Calculate the magnitude of the interpolated vectors. (We will remove this magnitude.) - # 64-bit operations are used here to allow large exponents. - current_magnitude = torch.norm(image_interp, p=2, dim=1, keepdim=True).to(torch.float64).add_(0.00001) - - # Interpolate the powered magnitudes, then un-power them (bring them back to a power of 1). - a_magnitude = torch.norm(a, p=2, dim=1, keepdim=True).to(torch.float64).pow_(soft_inpainting.inpaint_detail_preservation) * one_minus_t3 - b_magnitude = torch.norm(b, p=2, dim=1, keepdim=True).to(torch.float64).pow_(soft_inpainting.inpaint_detail_preservation) * t3 - desired_magnitude = a_magnitude - desired_magnitude.add_(b_magnitude).pow_(1 / soft_inpainting.inpaint_detail_preservation) - del a_magnitude, b_magnitude, t3, one_minus_t3 - - # Change the linearly interpolated image vectors' magnitudes to the value we want. - # This is the last 64-bit operation. - image_interp_scaling_factor = desired_magnitude - image_interp_scaling_factor.div_(current_magnitude) - image_interp_scaling_factor = image_interp_scaling_factor.to(result_type) - image_interp_scaled = image_interp - image_interp_scaled.mul_(image_interp_scaling_factor) - del current_magnitude - del desired_magnitude - del image_interp - del image_interp_scaling_factor - del result_type - - return image_interp_scaled - - -def get_modified_nmask(soft_inpainting, nmask, sigma): - """ - Converts a negative mask representing the transparency of the original latent vectors being overlayed - to a mask that is scaled according to the denoising strength for this step. - - Where: - 0 = fully opaque, infinite density, fully masked - 1 = fully transparent, zero density, fully unmasked - - We bring this transparency to a power, as this allows one to simulate N number of blending operations - where N can be any positive real value. Using this one can control the balance of influence between - the denoiser and the original latents according to the sigma value. - - NOTE: "mask" is not used - """ - import torch - # todo: Why is sigma 2D? Both values are the same. - return torch.pow(nmask, (sigma[0] ** soft_inpainting.mask_blend_power) * soft_inpainting.mask_blend_scale) - - -def apply_adaptive_masks( - latent_orig, - latent_processed, - overlay_images, - masks_for_overlay, - width, height, - paste_to): - import torch - import numpy as np - import modules.processing as proc - import modules.images as images - from PIL import Image, ImageOps, ImageFilter - - # TODO: Bias the blending according to the latent mask, add adjustable parameter for bias control. - # latent_mask = p.nmask[0].float().cpu() - # convert the original mask into a form we use to scale distances for thresholding - # mask_scalar = 1-(torch.clamp(latent_mask, min=0, max=1) ** (p.mask_blend_scale / 2)) - # mask_scalar = mask_scalar / (1.00001-mask_scalar) - # mask_scalar = mask_scalar.numpy() - - latent_distance = torch.norm(latent_processed - latent_orig, p=2, dim=1) - - kernel, kernel_center = images.get_gaussian_kernel(stddev_radius=1.5, max_radius=2) - - for i, (distance_map, overlay_image) in enumerate(zip(latent_distance, overlay_images)): - converted_mask = distance_map.float().cpu().numpy() - converted_mask = images.weighted_histogram_filter(converted_mask, kernel, kernel_center, - percentile_min=0.9, percentile_max=1, min_width=1) - converted_mask = images.weighted_histogram_filter(converted_mask, kernel, kernel_center, - percentile_min=0.25, percentile_max=0.75, min_width=1) - - # The distance at which opacity of original decreases to 50% - # half_weighted_distance = 1 # * mask_scalar - # converted_mask = converted_mask / half_weighted_distance - - converted_mask = 1 / (1 + converted_mask ** 2) - converted_mask = images.smootherstep(converted_mask) - converted_mask = 1 - converted_mask - converted_mask = 255. * converted_mask - converted_mask = converted_mask.astype(np.uint8) - converted_mask = Image.fromarray(converted_mask) - converted_mask = images.resize_image(2, converted_mask, width, height) - converted_mask = proc.create_binary_mask(converted_mask, round=False) - - # Remove aliasing artifacts using a gaussian blur. - converted_mask = converted_mask.filter(ImageFilter.GaussianBlur(radius=4)) - - # Expand the mask to fit the whole image if needed. - if paste_to is not None: - converted_mask = proc. uncrop(converted_mask, - (overlay_image.width, overlay_image.height), - paste_to) - - masks_for_overlay[i] = converted_mask - - image_masked = Image.new('RGBa', (overlay_image.width, overlay_image.height)) - image_masked.paste(overlay_image.convert("RGBA").convert("RGBa"), - mask=ImageOps.invert(converted_mask.convert('L'))) - - overlay_images[i] = image_masked.convert('RGBA') - -def apply_masks( - soft_inpainting, - nmask, - overlay_images, - masks_for_overlay, - width, height, - paste_to): - import torch - import numpy as np - import modules.processing as proc - import modules.images as images - from PIL import Image, ImageOps, ImageFilter - - converted_mask = nmask[0].float() - converted_mask = torch.clamp(converted_mask, min=0, max=1).pow_(soft_inpainting.mask_blend_scale / 2) - converted_mask = 255. * converted_mask - converted_mask = converted_mask.cpu().numpy().astype(np.uint8) - converted_mask = Image.fromarray(converted_mask) - converted_mask = images.resize_image(2, converted_mask, width, height) - converted_mask = proc.create_binary_mask(converted_mask, round=False) - - # Remove aliasing artifacts using a gaussian blur. - converted_mask = converted_mask.filter(ImageFilter.GaussianBlur(radius=4)) - - # Expand the mask to fit the whole image if needed. - if paste_to is not None: - converted_mask = proc.uncrop(converted_mask, - (width, height), - paste_to) - - for i, overlay_image in enumerate(overlay_images): - masks_for_overlay[i] = converted_mask - - image_masked = Image.new('RGBa', (overlay_image.width, overlay_image.height)) - image_masked.paste(overlay_image.convert("RGBA").convert("RGBa"), - mask=ImageOps.invert(converted_mask.convert('L'))) - - overlay_images[i] = image_masked.convert('RGBA') - - -# ------------------- Constants ------------------- - - -default = SoftInpaintingSettings(1, 0.5, 4) - -enabled_ui_label = "Soft inpainting" -enabled_gen_param_label = "Soft inpainting enabled" -enabled_el_id = "soft_inpainting_enabled" - -ui_labels = SoftInpaintingSettings( - "Schedule bias", - "Preservation strength", - "Transition contrast boost") - -ui_info = SoftInpaintingSettings( - "Shifts when preservation of original content occurs during denoising.", - "How strongly partially masked content should be preserved.", - "Amplifies the contrast that may be lost in partially masked regions.") - -gen_param_labels = SoftInpaintingSettings( - "Soft inpainting schedule bias", - "Soft inpainting preservation strength", - "Soft inpainting transition contrast boost") - -el_ids = SoftInpaintingSettings( - "mask_blend_power", - "mask_blend_scale", - "inpaint_detail_preservation") - - -# ------------------- UI ------------------- - - -def gradio_ui(): - import gradio as gr - from modules.ui_components import InputAccordion - - with InputAccordion(False, label=enabled_ui_label, elem_id=enabled_el_id) as soft_inpainting_enabled: - with gr.Group(): - gr.Markdown( - """ - Soft inpainting allows you to **seamlessly blend original content with inpainted content** according to the mask opacity. - **High _Mask blur_** values are recommended! - """) - - result = SoftInpaintingSettings( - gr.Slider(label=ui_labels.mask_blend_power, - info=ui_info.mask_blend_power, - minimum=0, - maximum=8, - step=0.1, - value=default.mask_blend_power, - elem_id=el_ids.mask_blend_power), - gr.Slider(label=ui_labels.mask_blend_scale, - info=ui_info.mask_blend_scale, - minimum=0, - maximum=8, - step=0.05, - value=default.mask_blend_scale, - elem_id=el_ids.mask_blend_scale), - gr.Slider(label=ui_labels.inpaint_detail_preservation, - info=ui_info.inpaint_detail_preservation, - minimum=1, - maximum=32, - step=0.5, - value=default.inpaint_detail_preservation, - elem_id=el_ids.inpaint_detail_preservation)) - - with gr.Accordion("Help", open=False): - gr.Markdown( - f""" - ### {ui_labels.mask_blend_power} - - The blending strength of original content is scaled proportionally with the decreasing noise level values at each step (sigmas). - This ensures that the influence of the denoiser and original content preservation is roughly balanced at each step. - This balance can be shifted using this parameter, controlling whether earlier or later steps have stronger preservation. - - - **Below 1**: Stronger preservation near the end (with low sigma) - - **1**: Balanced (proportional to sigma) - - **Above 1**: Stronger preservation in the beginning (with high sigma) - """) - gr.Markdown( - f""" - ### {ui_labels.mask_blend_scale} - - Skews whether partially masked image regions should be more likely to preserve the original content or favor inpainted content. - This may need to be adjusted depending on the {ui_labels.mask_blend_power}, CFG Scale, prompt and Denoising strength. - - - **Low values**: Favors generated content. - - **High values**: Favors original content. - """) - gr.Markdown( - f""" - ### {ui_labels.inpaint_detail_preservation} - - This parameter controls how the original latent vectors and denoised latent vectors are interpolated. - With higher values, the magnitude of the resulting blended vector will be closer to the maximum of the two interpolated vectors. - This can prevent the loss of contrast that occurs with linear interpolation. - - - **Low values**: Softer blending, details may fade. - - **High values**: Stronger contrast, may over-saturate colors. - """) - - return ( - [ - soft_inpainting_enabled, - result.mask_blend_power, - result.mask_blend_scale, - result.inpaint_detail_preservation - ], - [ - (soft_inpainting_enabled, enabled_gen_param_label), - (result.mask_blend_power, gen_param_labels.mask_blend_power), - (result.mask_blend_scale, gen_param_labels.mask_blend_scale), - (result.inpaint_detail_preservation, gen_param_labels.inpaint_detail_preservation) - ] - ) diff --git a/scripts/soft_inpainting.py b/scripts/soft_inpainting.py new file mode 100644 index 000000000..47e0269bf --- /dev/null +++ b/scripts/soft_inpainting.py @@ -0,0 +1,401 @@ +import gradio as gr +from modules.ui_components import InputAccordion +import modules.scripts as scripts + + +class SoftInpaintingSettings: + def __init__(self, mask_blend_power, mask_blend_scale, inpaint_detail_preservation): + self.mask_blend_power = mask_blend_power + self.mask_blend_scale = mask_blend_scale + self.inpaint_detail_preservation = inpaint_detail_preservation + + def add_generation_params(self, dest): + dest[enabled_gen_param_label] = True + dest[gen_param_labels.mask_blend_power] = self.mask_blend_power + dest[gen_param_labels.mask_blend_scale] = self.mask_blend_scale + dest[gen_param_labels.inpaint_detail_preservation] = self.inpaint_detail_preservation + + +# ------------------- Methods ------------------- + + +def latent_blend(soft_inpainting, a, b, t): + """ + Interpolates two latent image representations according to the parameter t, + where the interpolated vectors' magnitudes are also interpolated separately. + The "detail_preservation" factor biases the magnitude interpolation towards + the larger of the two magnitudes. + """ + import torch + + # NOTE: We use inplace operations wherever possible. + + # [4][w][h] to [1][4][w][h] + t2 = t.unsqueeze(0) + # [4][w][h] to [1][1][w][h] - the [4] seem redundant. + t3 = t[0].unsqueeze(0).unsqueeze(0) + + one_minus_t2 = 1 - t2 + one_minus_t3 = 1 - t3 + + # Linearly interpolate the image vectors. + a_scaled = a * one_minus_t2 + b_scaled = b * t2 + image_interp = a_scaled + image_interp.add_(b_scaled) + result_type = image_interp.dtype + del a_scaled, b_scaled, t2, one_minus_t2 + + # Calculate the magnitude of the interpolated vectors. (We will remove this magnitude.) + # 64-bit operations are used here to allow large exponents. + current_magnitude = torch.norm(image_interp, p=2, dim=1, keepdim=True).to(torch.float64).add_(0.00001) + + # Interpolate the powered magnitudes, then un-power them (bring them back to a power of 1). + a_magnitude = torch.norm(a, p=2, dim=1, keepdim=True).to(torch.float64).pow_( + soft_inpainting.inpaint_detail_preservation) * one_minus_t3 + b_magnitude = torch.norm(b, p=2, dim=1, keepdim=True).to(torch.float64).pow_( + soft_inpainting.inpaint_detail_preservation) * t3 + desired_magnitude = a_magnitude + desired_magnitude.add_(b_magnitude).pow_(1 / soft_inpainting.inpaint_detail_preservation) + del a_magnitude, b_magnitude, t3, one_minus_t3 + + # Change the linearly interpolated image vectors' magnitudes to the value we want. + # This is the last 64-bit operation. + image_interp_scaling_factor = desired_magnitude + image_interp_scaling_factor.div_(current_magnitude) + image_interp_scaling_factor = image_interp_scaling_factor.to(result_type) + image_interp_scaled = image_interp + image_interp_scaled.mul_(image_interp_scaling_factor) + del current_magnitude + del desired_magnitude + del image_interp + del image_interp_scaling_factor + del result_type + + return image_interp_scaled + + +def get_modified_nmask(soft_inpainting, nmask, sigma): + """ + Converts a negative mask representing the transparency of the original latent vectors being overlayed + to a mask that is scaled according to the denoising strength for this step. + + Where: + 0 = fully opaque, infinite density, fully masked + 1 = fully transparent, zero density, fully unmasked + + We bring this transparency to a power, as this allows one to simulate N number of blending operations + where N can be any positive real value. Using this one can control the balance of influence between + the denoiser and the original latents according to the sigma value. + + NOTE: "mask" is not used + """ + import torch + return torch.pow(nmask, (sigma ** soft_inpainting.mask_blend_power) * soft_inpainting.mask_blend_scale) + + +def apply_adaptive_masks( + latent_orig, + latent_processed, + overlay_images, + width, height, + paste_to): + import torch + import numpy as np + import modules.processing as proc + import modules.images as images + from PIL import Image, ImageOps, ImageFilter + + # TODO: Bias the blending according to the latent mask, add adjustable parameter for bias control. + # latent_mask = p.nmask[0].float().cpu() + # convert the original mask into a form we use to scale distances for thresholding + # mask_scalar = 1-(torch.clamp(latent_mask, min=0, max=1) ** (p.mask_blend_scale / 2)) + # mask_scalar = mask_scalar / (1.00001-mask_scalar) + # mask_scalar = mask_scalar.numpy() + + latent_distance = torch.norm(latent_processed - latent_orig, p=2, dim=1) + + kernel, kernel_center = images.get_gaussian_kernel(stddev_radius=1.5, max_radius=2) + + masks_for_overlay = [] + + for i, (distance_map, overlay_image) in enumerate(zip(latent_distance, overlay_images)): + converted_mask = distance_map.float().cpu().numpy() + converted_mask = images.weighted_histogram_filter(converted_mask, kernel, kernel_center, + percentile_min=0.9, percentile_max=1, min_width=1) + converted_mask = images.weighted_histogram_filter(converted_mask, kernel, kernel_center, + percentile_min=0.25, percentile_max=0.75, min_width=1) + + # The distance at which opacity of original decreases to 50% + # half_weighted_distance = 1 # * mask_scalar + # converted_mask = converted_mask / half_weighted_distance + + converted_mask = 1 / (1 + converted_mask ** 2) + converted_mask = images.smootherstep(converted_mask) + converted_mask = 1 - converted_mask + converted_mask = 255. * converted_mask + converted_mask = converted_mask.astype(np.uint8) + converted_mask = Image.fromarray(converted_mask) + converted_mask = images.resize_image(2, converted_mask, width, height) + converted_mask = proc.create_binary_mask(converted_mask, round=False) + + # Remove aliasing artifacts using a gaussian blur. + converted_mask = converted_mask.filter(ImageFilter.GaussianBlur(radius=4)) + + # Expand the mask to fit the whole image if needed. + if paste_to is not None: + converted_mask = proc.uncrop(converted_mask, + (overlay_image.width, overlay_image.height), + paste_to) + + masks_for_overlay.append(converted_mask) + + image_masked = Image.new('RGBa', (overlay_image.width, overlay_image.height)) + image_masked.paste(overlay_image.convert("RGBA").convert("RGBa"), + mask=ImageOps.invert(converted_mask.convert('L'))) + + overlay_images[i] = image_masked.convert('RGBA') + + return masks_for_overlay + + +def apply_masks( + soft_inpainting, + nmask, + overlay_images, + width, height, + paste_to): + import torch + import numpy as np + import modules.processing as proc + import modules.images as images + from PIL import Image, ImageOps, ImageFilter + + converted_mask = nmask[0].float() + converted_mask = torch.clamp(converted_mask, min=0, max=1).pow_(soft_inpainting.mask_blend_scale / 2) + converted_mask = 255. * converted_mask + converted_mask = converted_mask.cpu().numpy().astype(np.uint8) + converted_mask = Image.fromarray(converted_mask) + converted_mask = images.resize_image(2, converted_mask, width, height) + converted_mask = proc.create_binary_mask(converted_mask, round=False) + + # Remove aliasing artifacts using a gaussian blur. + converted_mask = converted_mask.filter(ImageFilter.GaussianBlur(radius=4)) + + # Expand the mask to fit the whole image if needed. + if paste_to is not None: + converted_mask = proc.uncrop(converted_mask, + (width, height), + paste_to) + + masks_for_overlay = [] + + for i, overlay_image in enumerate(overlay_images): + masks_for_overlay[i] = converted_mask + + image_masked = Image.new('RGBa', (overlay_image.width, overlay_image.height)) + image_masked.paste(overlay_image.convert("RGBA").convert("RGBa"), + mask=ImageOps.invert(converted_mask.convert('L'))) + + overlay_images[i] = image_masked.convert('RGBA') + + return masks_for_overlay + + +# ------------------- Constants ------------------- + + +default = SoftInpaintingSettings(1, 0.5, 4) + +enabled_ui_label = "Soft inpainting" +enabled_gen_param_label = "Soft inpainting enabled" +enabled_el_id = "soft_inpainting_enabled" + +ui_labels = SoftInpaintingSettings( + "Schedule bias", + "Preservation strength", + "Transition contrast boost") + +ui_info = SoftInpaintingSettings( + "Shifts when preservation of original content occurs during denoising.", + "How strongly partially masked content should be preserved.", + "Amplifies the contrast that may be lost in partially masked regions.") + +gen_param_labels = SoftInpaintingSettings( + "Soft inpainting schedule bias", + "Soft inpainting preservation strength", + "Soft inpainting transition contrast boost") + +el_ids = SoftInpaintingSettings( + "mask_blend_power", + "mask_blend_scale", + "inpaint_detail_preservation") + + +class Script(scripts.Script): + + def __init__(self): + self.masks_for_overlay = None + self.overlay_images = None + + def title(self): + return "Soft Inpainting" + + def show(self, is_img2img): + return scripts.AlwaysVisible if is_img2img else False + + def ui(self, is_img2img): + if not is_img2img: + return + + with InputAccordion(False, label=enabled_ui_label, elem_id=enabled_el_id) as soft_inpainting_enabled: + with gr.Group(): + gr.Markdown( + """ + Soft inpainting allows you to **seamlessly blend original content with inpainted content** according to the mask opacity. + **High _Mask blur_** values are recommended! + """) + + result = SoftInpaintingSettings( + gr.Slider(label=ui_labels.mask_blend_power, + info=ui_info.mask_blend_power, + minimum=0, + maximum=8, + step=0.1, + value=default.mask_blend_power, + elem_id=el_ids.mask_blend_power), + gr.Slider(label=ui_labels.mask_blend_scale, + info=ui_info.mask_blend_scale, + minimum=0, + maximum=8, + step=0.05, + value=default.mask_blend_scale, + elem_id=el_ids.mask_blend_scale), + gr.Slider(label=ui_labels.inpaint_detail_preservation, + info=ui_info.inpaint_detail_preservation, + minimum=1, + maximum=32, + step=0.5, + value=default.inpaint_detail_preservation, + elem_id=el_ids.inpaint_detail_preservation)) + + with gr.Accordion("Help", open=False): + gr.Markdown( + f""" + ### {ui_labels.mask_blend_power} + + The blending strength of original content is scaled proportionally with the decreasing noise level values at each step (sigmas). + This ensures that the influence of the denoiser and original content preservation is roughly balanced at each step. + This balance can be shifted using this parameter, controlling whether earlier or later steps have stronger preservation. + + - **Below 1**: Stronger preservation near the end (with low sigma) + - **1**: Balanced (proportional to sigma) + - **Above 1**: Stronger preservation in the beginning (with high sigma) + """) + gr.Markdown( + f""" + ### {ui_labels.mask_blend_scale} + + Skews whether partially masked image regions should be more likely to preserve the original content or favor inpainted content. + This may need to be adjusted depending on the {ui_labels.mask_blend_power}, CFG Scale, prompt and Denoising strength. + + - **Low values**: Favors generated content. + - **High values**: Favors original content. + """) + gr.Markdown( + f""" + ### {ui_labels.inpaint_detail_preservation} + + This parameter controls how the original latent vectors and denoised latent vectors are interpolated. + With higher values, the magnitude of the resulting blended vector will be closer to the maximum of the two interpolated vectors. + This can prevent the loss of contrast that occurs with linear interpolation. + + - **Low values**: Softer blending, details may fade. + - **High values**: Stronger contrast, may over-saturate colors. + """) + + self.infotext_fields = [(soft_inpainting_enabled, enabled_gen_param_label), + (result.mask_blend_power, gen_param_labels.mask_blend_power), + (result.mask_blend_scale, gen_param_labels.mask_blend_scale), + (result.inpaint_detail_preservation, gen_param_labels.inpaint_detail_preservation)] + + self.paste_field_names = [] + for _, field_name in self.infotext_fields: + self.paste_field_names.append(field_name) + + return [soft_inpainting_enabled, + result.mask_blend_power, + result.mask_blend_scale, + result.inpaint_detail_preservation] + + def process(self, p, enabled, power, scale, detail_preservation): + if not enabled: + return + + # Shut off the rounding it normally does. + p.mask_round = False + + settings = SoftInpaintingSettings(power, scale, detail_preservation) + + # p.extra_generation_params["Mask rounding"] = False + settings.add_generation_params(p.extra_generation_params) + + def on_mask_blend(self, p, mba: scripts.MaskBlendArgs, enabled, power, scale, detail_preservation): + if not enabled: + return + + if mba.sigma is None: + mba.blended_latent = mba.current_latent + return + + settings = SoftInpaintingSettings(power, scale, detail_preservation) + + # todo: Why is sigma 2D? Both values are the same. + mba.blended_latent = latent_blend(settings, + mba.init_latent, + mba.current_latent, + get_modified_nmask(settings, mba.nmask, mba.sigma[0])) + + def post_sample(self, p, ps: scripts.PostSampleArgs, enabled, power, scale, detail_preservation): + if not enabled: + return + + settings = SoftInpaintingSettings(power, scale, detail_preservation) + + from modules import images + from modules.shared import opts + + # since the original code puts holes in the existing overlay images, + # we have to rebuild them. + self.overlay_images = [] + for img in p.init_images: + + image = images.flatten(img, opts.img2img_background_color) + + if p.paste_to is None and p.resize_mode != 3: + image = images.resize_image(p.resize_mode, image, p.width, p.height) + + self.overlay_images.append(image.convert('RGBA')) + + if getattr(ps.samples, 'already_decoded', False): + self.masks_for_overlay = apply_masks(soft_inpainting=settings, + nmask=p.nmask, + overlay_images=self.overlay_images, + width=p.width, + height=p.height, + paste_to=p.paste_to) + else: + self.masks_for_overlay = apply_adaptive_masks(latent_orig=p.init_latent, + latent_processed=ps.samples, + overlay_images=self.overlay_images, + width=p.width, + height=p.height, + paste_to=p.paste_to) + + + def postprocess_maskoverlay(self, p, ppmo: scripts.PostProcessMaskOverlayArgs, enabled, power, scale, detail_preservation): + if not enabled: + return + + ppmo.mask_for_overlay = self.masks_for_overlay[ppmo.index] + ppmo.overlay_image = self.overlay_images[ppmo.index] \ No newline at end of file From 8dbacc7d018774a3bc801cc57617795274a15087 Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Thu, 7 Dec 2023 14:30:30 -0700 Subject: [PATCH 226/342] Fixed "No newline at end of file". --- scripts/soft_inpainting.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/scripts/soft_inpainting.py b/scripts/soft_inpainting.py index 47e0269bf..6d0cf8479 100644 --- a/scripts/soft_inpainting.py +++ b/scripts/soft_inpainting.py @@ -398,4 +398,4 @@ class Script(scripts.Script): return ppmo.mask_for_overlay = self.masks_for_overlay[ppmo.index] - ppmo.overlay_image = self.overlay_images[ppmo.index] \ No newline at end of file + ppmo.overlay_image = self.overlay_images[ppmo.index] From 56604f08a18588e8e6b57d7c3f9c61d6624846f8 Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Thu, 7 Dec 2023 14:53:44 -0700 Subject: [PATCH 227/342] Moved image filters used by soft inpainting into soft_inpainting.py from images.py --- modules/images.py | 190 ---------------------------------- scripts/soft_inpainting.py | 205 +++++++++++++++++++++++++++++++++++-- 2 files changed, 199 insertions(+), 196 deletions(-) diff --git a/modules/images.py b/modules/images.py index 949534986..16f9ae7cc 100644 --- a/modules/images.py +++ b/modules/images.py @@ -792,193 +792,3 @@ def flatten(img, bgcolor): return img.convert('RGB') - -def weighted_histogram_filter(img, kernel, kernel_center, percentile_min=0.0, percentile_max=1.0, min_width=1.0): - """ - Generalization convolution filter capable of applying - weighted mean, median, maximum, and minimum filters - parametrically using an arbitrary kernel. - - Args: - img (nparray): - The image, a 2-D array of floats, to which the filter is being applied. - kernel (nparray): - The kernel, a 2-D array of floats. - kernel_center (nparray): - The kernel center coordinate, a 1-D array with two elements. - percentile_min (float): - The lower bound of the histogram window used by the filter, - from 0 to 1. - percentile_max (float): - The upper bound of the histogram window used by the filter, - from 0 to 1. - min_width (float): - The minimum size of the histogram window bounds, in weight units. - Must be greater than 0. - - Returns: - (nparray): A filtered copy of the input image "img", a 2-D array of floats. - """ - - # Converts an index tuple into a vector. - def vec(x): - return np.array(x) - - kernel_min = -kernel_center - kernel_max = vec(kernel.shape) - kernel_center - - def weighted_histogram_filter_single(idx): - idx = vec(idx) - min_index = np.maximum(0, idx + kernel_min) - max_index = np.minimum(vec(img.shape), idx + kernel_max) - window_shape = max_index - min_index - - class WeightedElement: - """ - An element of the histogram, its weight - and bounds. - """ - def __init__(self, value, weight): - self.value: float = value - self.weight: float = weight - self.window_min: float = 0.0 - self.window_max: float = 1.0 - - # Collect the values in the image as WeightedElements, - # weighted by their corresponding kernel values. - values = [] - for window_tup in np.ndindex(tuple(window_shape)): - window_index = vec(window_tup) - image_index = window_index + min_index - centered_kernel_index = image_index - idx - kernel_index = centered_kernel_index + kernel_center - element = WeightedElement(img[tuple(image_index)], kernel[tuple(kernel_index)]) - values.append(element) - - def sort_key(x: WeightedElement): - return x.value - - values.sort(key=sort_key) - - # Calculate the height of the stack (sum) - # and each sample's range they occupy in the stack - sum = 0 - for i in range(len(values)): - values[i].window_min = sum - sum += values[i].weight - values[i].window_max = sum - - # Calculate what range of this stack ("window") - # we want to get the weighted average across. - window_min = sum * percentile_min - window_max = sum * percentile_max - window_width = window_max - window_min - - # Ensure the window is within the stack and at least a certain size. - if window_width < min_width: - window_center = (window_min + window_max) / 2 - window_min = window_center - min_width / 2 - window_max = window_center + min_width / 2 - - if window_max > sum: - window_max = sum - window_min = sum - min_width - - if window_min < 0: - window_min = 0 - window_max = min_width - - value = 0 - value_weight = 0 - - # Get the weighted average of all the samples - # that overlap with the window, weighted - # by the size of their overlap. - for i in range(len(values)): - if window_min >= values[i].window_max: - continue - if window_max <= values[i].window_min: - break - - s = max(window_min, values[i].window_min) - e = min(window_max, values[i].window_max) - w = e - s - - value += values[i].value * w - value_weight += w - - return value / value_weight if value_weight != 0 else 0 - - img_out = img.copy() - - # Apply the kernel operation over each pixel. - for index in np.ndindex(img.shape): - img_out[index] = weighted_histogram_filter_single(index) - - return img_out - -def smoothstep(x): - """ - The smoothstep function, input should be clamped to 0-1 range. - Turns a diagonal line (f(x) = x) into a sigmoid-like curve. - """ - return x * x * (3 - 2 * x) - -def smootherstep(x): - """ - The smootherstep function, input should be clamped to 0-1 range. - Turns a diagonal line (f(x) = x) into a sigmoid-like curve. - """ - return x * x * x * (x * (6 * x - 15) + 10) - - -def get_gaussian_kernel(stddev_radius=1.0, max_radius=2): - """ - Creates a Gaussian kernel with thresholded edges. - - Args: - stddev_radius (float): - Standard deviation of the gaussian kernel, in pixels. - max_radius (int): - The size of the filter kernel. The number of pixels is (max_radius*2+1) ** 2. - The kernel is thresholded so that any values one pixel beyond this radius - is weighted at 0. - - Returns: - (nparray, nparray): A kernel array (shape: (N, N)), its center coordinate (shape: (2)) - """ - # Evaluates a 0-1 normalized gaussian function for a given square distance from the mean. - def gaussian(sqr_mag): - return math.exp(-sqr_mag / (stddev_radius * stddev_radius)) - - # Helper function for converting a tuple to an array. - def vec(x): - return np.array(x) - - """ - Since a gaussian is unbounded, we need to limit ourselves - to a finite range. - We taper the ends off at the end of that range so they equal zero - while preserving the maximum value of 1 at the mean. - """ - zero_radius = max_radius + 1.0 - gauss_zero = gaussian(zero_radius * zero_radius) - gauss_kernel_scale = 1 / (1 - gauss_zero) - - def gaussian_kernel_func(coordinate): - x = coordinate[0] ** 2.0 + coordinate[1] ** 2.0 - x = gaussian(x) - x -= gauss_zero - x *= gauss_kernel_scale - x = max(0.0, x) - return x - - size = max_radius * 2 + 1 - kernel_center = max_radius - kernel = np.zeros((size, size)) - - for index in np.ndindex(kernel.shape): - kernel[index] = gaussian_kernel_func(vec(index) - kernel_center) - - return kernel, kernel_center - diff --git a/scripts/soft_inpainting.py b/scripts/soft_inpainting.py index 6d0cf8479..1f451b553 100644 --- a/scripts/soft_inpainting.py +++ b/scripts/soft_inpainting.py @@ -1,4 +1,6 @@ +import numpy as np import gradio as gr +import math from modules.ui_components import InputAccordion import modules.scripts as scripts @@ -101,7 +103,6 @@ def apply_adaptive_masks( width, height, paste_to): import torch - import numpy as np import modules.processing as proc import modules.images as images from PIL import Image, ImageOps, ImageFilter @@ -115,15 +116,15 @@ def apply_adaptive_masks( latent_distance = torch.norm(latent_processed - latent_orig, p=2, dim=1) - kernel, kernel_center = images.get_gaussian_kernel(stddev_radius=1.5, max_radius=2) + kernel, kernel_center = get_gaussian_kernel(stddev_radius=1.5, max_radius=2) masks_for_overlay = [] for i, (distance_map, overlay_image) in enumerate(zip(latent_distance, overlay_images)): converted_mask = distance_map.float().cpu().numpy() - converted_mask = images.weighted_histogram_filter(converted_mask, kernel, kernel_center, + converted_mask = weighted_histogram_filter(converted_mask, kernel, kernel_center, percentile_min=0.9, percentile_max=1, min_width=1) - converted_mask = images.weighted_histogram_filter(converted_mask, kernel, kernel_center, + converted_mask = weighted_histogram_filter(converted_mask, kernel, kernel_center, percentile_min=0.25, percentile_max=0.75, min_width=1) # The distance at which opacity of original decreases to 50% @@ -131,7 +132,7 @@ def apply_adaptive_masks( # converted_mask = converted_mask / half_weighted_distance converted_mask = 1 / (1 + converted_mask ** 2) - converted_mask = images.smootherstep(converted_mask) + converted_mask = smootherstep(converted_mask) converted_mask = 1 - converted_mask converted_mask = 255. * converted_mask converted_mask = converted_mask.astype(np.uint8) @@ -166,7 +167,6 @@ def apply_masks( width, height, paste_to): import torch - import numpy as np import modules.processing as proc import modules.images as images from PIL import Image, ImageOps, ImageFilter @@ -202,6 +202,196 @@ def apply_masks( return masks_for_overlay +def weighted_histogram_filter(img, kernel, kernel_center, percentile_min=0.0, percentile_max=1.0, min_width=1.0): + """ + Generalization convolution filter capable of applying + weighted mean, median, maximum, and minimum filters + parametrically using an arbitrary kernel. + + Args: + img (nparray): + The image, a 2-D array of floats, to which the filter is being applied. + kernel (nparray): + The kernel, a 2-D array of floats. + kernel_center (nparray): + The kernel center coordinate, a 1-D array with two elements. + percentile_min (float): + The lower bound of the histogram window used by the filter, + from 0 to 1. + percentile_max (float): + The upper bound of the histogram window used by the filter, + from 0 to 1. + min_width (float): + The minimum size of the histogram window bounds, in weight units. + Must be greater than 0. + + Returns: + (nparray): A filtered copy of the input image "img", a 2-D array of floats. + """ + + # Converts an index tuple into a vector. + def vec(x): + return np.array(x) + + kernel_min = -kernel_center + kernel_max = vec(kernel.shape) - kernel_center + + def weighted_histogram_filter_single(idx): + idx = vec(idx) + min_index = np.maximum(0, idx + kernel_min) + max_index = np.minimum(vec(img.shape), idx + kernel_max) + window_shape = max_index - min_index + + class WeightedElement: + """ + An element of the histogram, its weight + and bounds. + """ + def __init__(self, value, weight): + self.value: float = value + self.weight: float = weight + self.window_min: float = 0.0 + self.window_max: float = 1.0 + + # Collect the values in the image as WeightedElements, + # weighted by their corresponding kernel values. + values = [] + for window_tup in np.ndindex(tuple(window_shape)): + window_index = vec(window_tup) + image_index = window_index + min_index + centered_kernel_index = image_index - idx + kernel_index = centered_kernel_index + kernel_center + element = WeightedElement(img[tuple(image_index)], kernel[tuple(kernel_index)]) + values.append(element) + + def sort_key(x: WeightedElement): + return x.value + + values.sort(key=sort_key) + + # Calculate the height of the stack (sum) + # and each sample's range they occupy in the stack + sum = 0 + for i in range(len(values)): + values[i].window_min = sum + sum += values[i].weight + values[i].window_max = sum + + # Calculate what range of this stack ("window") + # we want to get the weighted average across. + window_min = sum * percentile_min + window_max = sum * percentile_max + window_width = window_max - window_min + + # Ensure the window is within the stack and at least a certain size. + if window_width < min_width: + window_center = (window_min + window_max) / 2 + window_min = window_center - min_width / 2 + window_max = window_center + min_width / 2 + + if window_max > sum: + window_max = sum + window_min = sum - min_width + + if window_min < 0: + window_min = 0 + window_max = min_width + + value = 0 + value_weight = 0 + + # Get the weighted average of all the samples + # that overlap with the window, weighted + # by the size of their overlap. + for i in range(len(values)): + if window_min >= values[i].window_max: + continue + if window_max <= values[i].window_min: + break + + s = max(window_min, values[i].window_min) + e = min(window_max, values[i].window_max) + w = e - s + + value += values[i].value * w + value_weight += w + + return value / value_weight if value_weight != 0 else 0 + + img_out = img.copy() + + # Apply the kernel operation over each pixel. + for index in np.ndindex(img.shape): + img_out[index] = weighted_histogram_filter_single(index) + + return img_out + +def smoothstep(x): + """ + The smoothstep function, input should be clamped to 0-1 range. + Turns a diagonal line (f(x) = x) into a sigmoid-like curve. + """ + return x * x * (3 - 2 * x) + +def smootherstep(x): + """ + The smootherstep function, input should be clamped to 0-1 range. + Turns a diagonal line (f(x) = x) into a sigmoid-like curve. + """ + return x * x * x * (x * (6 * x - 15) + 10) + + +def get_gaussian_kernel(stddev_radius=1.0, max_radius=2): + """ + Creates a Gaussian kernel with thresholded edges. + + Args: + stddev_radius (float): + Standard deviation of the gaussian kernel, in pixels. + max_radius (int): + The size of the filter kernel. The number of pixels is (max_radius*2+1) ** 2. + The kernel is thresholded so that any values one pixel beyond this radius + is weighted at 0. + + Returns: + (nparray, nparray): A kernel array (shape: (N, N)), its center coordinate (shape: (2)) + """ + # Evaluates a 0-1 normalized gaussian function for a given square distance from the mean. + def gaussian(sqr_mag): + return math.exp(-sqr_mag / (stddev_radius * stddev_radius)) + + # Helper function for converting a tuple to an array. + def vec(x): + return np.array(x) + + """ + Since a gaussian is unbounded, we need to limit ourselves + to a finite range. + We taper the ends off at the end of that range so they equal zero + while preserving the maximum value of 1 at the mean. + """ + zero_radius = max_radius + 1.0 + gauss_zero = gaussian(zero_radius * zero_radius) + gauss_kernel_scale = 1 / (1 - gauss_zero) + + def gaussian_kernel_func(coordinate): + x = coordinate[0] ** 2.0 + coordinate[1] ** 2.0 + x = gaussian(x) + x -= gauss_zero + x *= gauss_kernel_scale + x = max(0.0, x) + return x + + size = max_radius * 2 + 1 + kernel_center = max_radius + kernel = np.zeros((size, size)) + + for index in np.ndindex(kernel.shape): + kernel[index] = gaussian_kernel_func(vec(index) - kernel_center) + + return kernel, kernel_center + + # ------------------- Constants ------------------- @@ -232,6 +422,9 @@ el_ids = SoftInpaintingSettings( "inpaint_detail_preservation") +# ----- + + class Script(scripts.Script): def __init__(self): From 0ef4a4cb2365051b1e308f0136a0d8c01d071569 Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Thu, 7 Dec 2023 14:54:26 -0700 Subject: [PATCH 228/342] Fixed error that occurs when using vanilla samplers (somehow). --- modules/processing.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index f8d85bdf5..bea01ec68 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -937,8 +937,8 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: p.scripts.postprocess_image(p, pp) image = pp.image - mask_for_overlay = p.mask_for_overlay - overlay_image = p.overlay_images[i] if p.overlay_images is not None and i < len(p.overlay_images) else None + mask_for_overlay = getattr(p, "mask_for_overlay", None) + overlay_image = p.overlay_images[i] if getattr(p, "overlay_images", None) is not None and i < len(p.overlay_images) else None if p.scripts is not None: ppmo = scripts.PostProcessMaskOverlayArgs(i, mask_for_overlay, overlay_image) From f284ae23bcdfa212cf4763659c06e124ec5b1456 Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Thu, 7 Dec 2023 20:19:35 -0700 Subject: [PATCH 229/342] Added parameters for the composite stage, fixed batched generation. --- scripts/soft_inpainting.py | 196 +++++++++++++++++++++++++++++-------- 1 file changed, 154 insertions(+), 42 deletions(-) diff --git a/scripts/soft_inpainting.py b/scripts/soft_inpainting.py index 1f451b553..1b21aee9d 100644 --- a/scripts/soft_inpainting.py +++ b/scripts/soft_inpainting.py @@ -6,22 +6,34 @@ import modules.scripts as scripts class SoftInpaintingSettings: - def __init__(self, mask_blend_power, mask_blend_scale, inpaint_detail_preservation): + def __init__(self, + mask_blend_power, + mask_blend_scale, + inpaint_detail_preservation, + composite_mask_influence, + composite_difference_threshold, + composite_difference_contrast): self.mask_blend_power = mask_blend_power self.mask_blend_scale = mask_blend_scale self.inpaint_detail_preservation = inpaint_detail_preservation + self.composite_mask_influence = composite_mask_influence + self.composite_difference_threshold = composite_difference_threshold + self.composite_difference_contrast = composite_difference_contrast def add_generation_params(self, dest): dest[enabled_gen_param_label] = True dest[gen_param_labels.mask_blend_power] = self.mask_blend_power dest[gen_param_labels.mask_blend_scale] = self.mask_blend_scale dest[gen_param_labels.inpaint_detail_preservation] = self.inpaint_detail_preservation + dest[gen_param_labels.composite_mask_influence] = self.composite_mask_influence + dest[gen_param_labels.composite_difference_threshold] = self.composite_difference_threshold + dest[gen_param_labels.composite_difference_contrast] = self.composite_difference_contrast # ------------------- Methods ------------------- -def latent_blend(soft_inpainting, a, b, t): +def latent_blend(settings, a, b, t): """ Interpolates two latent image representations according to the parameter t, where the interpolated vectors' magnitudes are also interpolated separately. @@ -54,11 +66,11 @@ def latent_blend(soft_inpainting, a, b, t): # Interpolate the powered magnitudes, then un-power them (bring them back to a power of 1). a_magnitude = torch.norm(a, p=2, dim=1, keepdim=True).to(torch.float64).pow_( - soft_inpainting.inpaint_detail_preservation) * one_minus_t3 + settings.inpaint_detail_preservation) * one_minus_t3 b_magnitude = torch.norm(b, p=2, dim=1, keepdim=True).to(torch.float64).pow_( - soft_inpainting.inpaint_detail_preservation) * t3 + settings.inpaint_detail_preservation) * t3 desired_magnitude = a_magnitude - desired_magnitude.add_(b_magnitude).pow_(1 / soft_inpainting.inpaint_detail_preservation) + desired_magnitude.add_(b_magnitude).pow_(1 / settings.inpaint_detail_preservation) del a_magnitude, b_magnitude, t3, one_minus_t3 # Change the linearly interpolated image vectors' magnitudes to the value we want. @@ -77,7 +89,7 @@ def latent_blend(soft_inpainting, a, b, t): return image_interp_scaled -def get_modified_nmask(soft_inpainting, nmask, sigma): +def get_modified_nmask(settings, nmask, sigma): """ Converts a negative mask representing the transparency of the original latent vectors being overlayed to a mask that is scaled according to the denoising strength for this step. @@ -93,10 +105,12 @@ def get_modified_nmask(soft_inpainting, nmask, sigma): NOTE: "mask" is not used """ import torch - return torch.pow(nmask, (sigma ** soft_inpainting.mask_blend_power) * soft_inpainting.mask_blend_scale) + return torch.pow(nmask, (sigma ** settings.mask_blend_power) * settings.mask_blend_scale) def apply_adaptive_masks( + settings:SoftInpaintingSettings, + nmask, latent_orig, latent_processed, overlay_images, @@ -108,11 +122,13 @@ def apply_adaptive_masks( from PIL import Image, ImageOps, ImageFilter # TODO: Bias the blending according to the latent mask, add adjustable parameter for bias control. - # latent_mask = p.nmask[0].float().cpu() + latent_mask = nmask[0].float() # convert the original mask into a form we use to scale distances for thresholding - # mask_scalar = 1-(torch.clamp(latent_mask, min=0, max=1) ** (p.mask_blend_scale / 2)) - # mask_scalar = mask_scalar / (1.00001-mask_scalar) - # mask_scalar = mask_scalar.numpy() + mask_scalar = 1-(torch.clamp(latent_mask, min=0, max=1) ** (settings.mask_blend_scale / 2)) + mask_scalar = (0.5 * (1-settings.composite_mask_influence) + + mask_scalar * settings.composite_mask_influence) + mask_scalar = mask_scalar / (1.00001-mask_scalar) + mask_scalar = mask_scalar.cpu().numpy() latent_distance = torch.norm(latent_processed - latent_orig, p=2, dim=1) @@ -128,10 +144,10 @@ def apply_adaptive_masks( percentile_min=0.25, percentile_max=0.75, min_width=1) # The distance at which opacity of original decreases to 50% - # half_weighted_distance = 1 # * mask_scalar - # converted_mask = converted_mask / half_weighted_distance + half_weighted_distance = settings.composite_difference_threshold * mask_scalar + converted_mask = converted_mask / half_weighted_distance - converted_mask = 1 / (1 + converted_mask ** 2) + converted_mask = 1 / (1 + converted_mask ** settings.composite_difference_contrast) converted_mask = smootherstep(converted_mask) converted_mask = 1 - converted_mask converted_mask = 255. * converted_mask @@ -161,7 +177,7 @@ def apply_adaptive_masks( def apply_masks( - soft_inpainting, + settings, nmask, overlay_images, width, height, @@ -172,7 +188,7 @@ def apply_masks( from PIL import Image, ImageOps, ImageFilter converted_mask = nmask[0].float() - converted_mask = torch.clamp(converted_mask, min=0, max=1).pow_(soft_inpainting.mask_blend_scale / 2) + converted_mask = torch.clamp(converted_mask, min=0, max=1).pow_(settings.mask_blend_scale / 2) converted_mask = 255. * converted_mask converted_mask = converted_mask.cpu().numpy().astype(np.uint8) converted_mask = Image.fromarray(converted_mask) @@ -395,7 +411,7 @@ def get_gaussian_kernel(stddev_radius=1.0, max_radius=2): # ------------------- Constants ------------------- -default = SoftInpaintingSettings(1, 0.5, 4) +default = SoftInpaintingSettings(1, 0.5, 4, 0, 0.5, 2) enabled_ui_label = "Soft inpainting" enabled_gen_param_label = "Soft inpainting enabled" @@ -404,25 +420,37 @@ enabled_el_id = "soft_inpainting_enabled" ui_labels = SoftInpaintingSettings( "Schedule bias", "Preservation strength", - "Transition contrast boost") + "Transition contrast boost", + "Mask influence", + "Difference threshold", + "Difference contrast") ui_info = SoftInpaintingSettings( "Shifts when preservation of original content occurs during denoising.", "How strongly partially masked content should be preserved.", - "Amplifies the contrast that may be lost in partially masked regions.") + "Amplifies the contrast that may be lost in partially masked regions.", + "How strongly the original mask should bias the difference threshold.", + "How much an image region can change before the original pixels are not blended in anymore.", + "How sharp the transition should be between blended and not blended.") gen_param_labels = SoftInpaintingSettings( "Soft inpainting schedule bias", "Soft inpainting preservation strength", - "Soft inpainting transition contrast boost") + "Soft inpainting transition contrast boost", + "Soft inpainting mask influence", + "Soft inpainting difference threshold", + "Soft inpainting difference contrast") el_ids = SoftInpaintingSettings( "mask_blend_power", "mask_blend_scale", - "inpaint_detail_preservation") + "inpaint_detail_preservation", + "composite_mask_influence", + "composite_difference_threshold", + "composite_difference_contrast") -# ----- +# ------------------- Script ------------------- class Script(scripts.Script): @@ -449,28 +477,62 @@ class Script(scripts.Script): **High _Mask blur_** values are recommended! """) - result = SoftInpaintingSettings( + power = \ gr.Slider(label=ui_labels.mask_blend_power, info=ui_info.mask_blend_power, minimum=0, maximum=8, step=0.1, value=default.mask_blend_power, - elem_id=el_ids.mask_blend_power), + elem_id=el_ids.mask_blend_power) + scale = \ gr.Slider(label=ui_labels.mask_blend_scale, info=ui_info.mask_blend_scale, minimum=0, maximum=8, step=0.05, value=default.mask_blend_scale, - elem_id=el_ids.mask_blend_scale), + elem_id=el_ids.mask_blend_scale) + detail = \ gr.Slider(label=ui_labels.inpaint_detail_preservation, info=ui_info.inpaint_detail_preservation, minimum=1, maximum=32, step=0.5, value=default.inpaint_detail_preservation, - elem_id=el_ids.inpaint_detail_preservation)) + elem_id=el_ids.inpaint_detail_preservation) + + gr.Markdown( + """ + ### Pixel Composite Settings + """) + + mask_inf = \ + gr.Slider(label=ui_labels.composite_mask_influence, + info=ui_info.composite_mask_influence, + minimum=0, + maximum=1, + step=0.05, + value=default.composite_mask_influence, + elem_id=el_ids.composite_mask_influence) + + dif_thresh = \ + gr.Slider(label=ui_labels.composite_difference_threshold, + info=ui_info.composite_difference_threshold, + minimum=0, + maximum=8, + step=0.25, + value=default.composite_difference_threshold, + elem_id=el_ids.composite_difference_threshold) + + dif_contr = \ + gr.Slider(label=ui_labels.composite_difference_contrast, + info=ui_info.composite_difference_contrast, + minimum=0, + maximum=8, + step=0.25, + value=default.composite_difference_contrast, + elem_id=el_ids.composite_difference_contrast) with gr.Accordion("Help", open=False): gr.Markdown( @@ -507,41 +569,86 @@ class Script(scripts.Script): - **High values**: Stronger contrast, may over-saturate colors. """) + gr.Markdown( + """ + ## Pixel Composite Settings + + Masks are generated based on how much a part of the image changed after denoising. + These masks are used to blend the original and final images together. + If the difference is low, the original pixels are used instead of the pixels returned by the inpainting process. + """) + + gr.Markdown( + f""" + ### {ui_labels.composite_mask_influence} + + This parameter controls how much the mask should bias this sensitivity to difference. + + - **0**: Ignore the mask, only consider differences in image content. + - **1**: Follow the mask closely despite image content changes. + """) + + gr.Markdown( + f""" + ### {ui_labels.composite_difference_threshold} + + This value represents the difference at which the opacity of the original pixels will have less than 50% opacity. + + - **Low values**: Two images patches must be almost the same in order to retain original pixels. + - **High values**: Two images patches can be very different and still retain original pixels. + """) + + gr.Markdown( + f""" + ### {ui_labels.composite_difference_contrast} + + This value represents the difference at which the opacity of the original pixels will have less than 50% opacity. + + - **Low values**: Two images patches must be almost the same in order to retain original pixels. + - **High values**: Two images patches can be very different and still retain original pixels. + """) + self.infotext_fields = [(soft_inpainting_enabled, enabled_gen_param_label), - (result.mask_blend_power, gen_param_labels.mask_blend_power), - (result.mask_blend_scale, gen_param_labels.mask_blend_scale), - (result.inpaint_detail_preservation, gen_param_labels.inpaint_detail_preservation)] + (power, gen_param_labels.mask_blend_power), + (scale, gen_param_labels.mask_blend_scale), + (detail, gen_param_labels.inpaint_detail_preservation), + (mask_inf, gen_param_labels.composite_mask_influence), + (dif_thresh, gen_param_labels.composite_difference_threshold), + (dif_contr, gen_param_labels.composite_difference_contrast)] self.paste_field_names = [] for _, field_name in self.infotext_fields: self.paste_field_names.append(field_name) return [soft_inpainting_enabled, - result.mask_blend_power, - result.mask_blend_scale, - result.inpaint_detail_preservation] + power, + scale, + detail, + mask_inf, + dif_thresh, + dif_contr] - def process(self, p, enabled, power, scale, detail_preservation): + def process(self, p, enabled, power, scale, detail_preservation, mask_inf, dif_thresh, dif_contr): if not enabled: return # Shut off the rounding it normally does. p.mask_round = False - settings = SoftInpaintingSettings(power, scale, detail_preservation) + settings = SoftInpaintingSettings(power, scale, detail_preservation, mask_inf, dif_thresh, dif_contr) # p.extra_generation_params["Mask rounding"] = False settings.add_generation_params(p.extra_generation_params) - def on_mask_blend(self, p, mba: scripts.MaskBlendArgs, enabled, power, scale, detail_preservation): + def on_mask_blend(self, p, mba: scripts.MaskBlendArgs, enabled, power, scale, detail_preservation, mask_inf, dif_thresh, dif_contr): if not enabled: return - if mba.sigma is None: + if mba.is_final_blend: mba.blended_latent = mba.current_latent return - settings = SoftInpaintingSettings(power, scale, detail_preservation) + settings = SoftInpaintingSettings(power, scale, detail_preservation, mask_inf, dif_thresh, dif_contr) # todo: Why is sigma 2D? Both values are the same. mba.blended_latent = latent_blend(settings, @@ -549,11 +656,11 @@ class Script(scripts.Script): mba.current_latent, get_modified_nmask(settings, mba.nmask, mba.sigma[0])) - def post_sample(self, p, ps: scripts.PostSampleArgs, enabled, power, scale, detail_preservation): + def post_sample(self, p, ps: scripts.PostSampleArgs, enabled, power, scale, detail_preservation, mask_inf, dif_thresh, dif_contr): if not enabled: return - settings = SoftInpaintingSettings(power, scale, detail_preservation) + settings = SoftInpaintingSettings(power, scale, detail_preservation, mask_inf, dif_thresh, dif_contr) from modules import images from modules.shared import opts @@ -570,15 +677,20 @@ class Script(scripts.Script): self.overlay_images.append(image.convert('RGBA')) + if len(p.init_images) == 1: + self.overlay_images = self.overlay_images * p.batch_size + if getattr(ps.samples, 'already_decoded', False): - self.masks_for_overlay = apply_masks(soft_inpainting=settings, + self.masks_for_overlay = apply_masks(settings=settings, nmask=p.nmask, overlay_images=self.overlay_images, width=p.width, height=p.height, paste_to=p.paste_to) else: - self.masks_for_overlay = apply_adaptive_masks(latent_orig=p.init_latent, + self.masks_for_overlay = apply_adaptive_masks(settings=settings, + nmask=p.nmask, + latent_orig=p.init_latent, latent_processed=ps.samples, overlay_images=self.overlay_images, width=p.width, @@ -586,7 +698,7 @@ class Script(scripts.Script): paste_to=p.paste_to) - def postprocess_maskoverlay(self, p, ppmo: scripts.PostProcessMaskOverlayArgs, enabled, power, scale, detail_preservation): + def postprocess_maskoverlay(self, p, ppmo: scripts.PostProcessMaskOverlayArgs, enabled, power, scale, detail_preservation, mask_inf, dif_thresh, dif_contr): if not enabled: return From fc3e246c0f4f292c33b181a902cd934629ff0d7a Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Thu, 7 Dec 2023 20:28:38 -0700 Subject: [PATCH 230/342] Fixed complaint about whitespace, updated help section for a parameter. --- scripts/soft_inpainting.py | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/scripts/soft_inpainting.py b/scripts/soft_inpainting.py index 1b21aee9d..6fb5cfbd0 100644 --- a/scripts/soft_inpainting.py +++ b/scripts/soft_inpainting.py @@ -572,7 +572,7 @@ class Script(scripts.Script): gr.Markdown( """ ## Pixel Composite Settings - + Masks are generated based on how much a part of the image changed after denoising. These masks are used to blend the original and final images together. If the difference is low, the original pixels are used instead of the pixels returned by the inpainting process. @@ -602,10 +602,10 @@ class Script(scripts.Script): f""" ### {ui_labels.composite_difference_contrast} - This value represents the difference at which the opacity of the original pixels will have less than 50% opacity. + This value represents the contrast between the opacity of the original and inpainted content. - - **Low values**: Two images patches must be almost the same in order to retain original pixels. - - **High values**: Two images patches can be very different and still retain original pixels. + - **Low values**: The blend will be more gradual and have longer transitions, but may cause ghosting. + - **High values**: Ghosting will be less common, but transitions may be very sudden. """) self.infotext_fields = [(soft_inpainting_enabled, enabled_gen_param_label), From 659f62e120b210e3043712ff928e8b7b6cd6cf61 Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Thu, 7 Dec 2023 21:39:54 -0700 Subject: [PATCH 231/342] Fixed grammar error. --- scripts/soft_inpainting.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/scripts/soft_inpainting.py b/scripts/soft_inpainting.py index 6fb5cfbd0..51f9ca2fe 100644 --- a/scripts/soft_inpainting.py +++ b/scripts/soft_inpainting.py @@ -592,7 +592,7 @@ class Script(scripts.Script): f""" ### {ui_labels.composite_difference_threshold} - This value represents the difference at which the opacity of the original pixels will have less than 50% opacity. + This value represents the difference at which the original pixels will have less than 50% opacity. - **Low values**: Two images patches must be almost the same in order to retain original pixels. - **High values**: Two images patches can be very different and still retain original pixels. From 16bdcce92d5b482d50cdc32a8f308040d320b6c9 Mon Sep 17 00:00:00 2001 From: Rene Kroon Date: Fri, 8 Dec 2023 21:19:29 +0100 Subject: [PATCH 232/342] #13354: solve lora loading issue --- extensions-builtin/Lora/networks.py | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/extensions-builtin/Lora/networks.py b/extensions-builtin/Lora/networks.py index 7f814706a..629bf8537 100644 --- a/extensions-builtin/Lora/networks.py +++ b/extensions-builtin/Lora/networks.py @@ -159,7 +159,8 @@ def load_network(name, network_on_disk): bundle_embeddings = {} for key_network, weight in sd.items(): - key_network_without_network_parts, network_part = key_network.split(".", 1) + key_network_without_network_parts, _, network_part = key_network.partition(".") + if key_network_without_network_parts == "bundle_emb": emb_name, vec_name = network_part.split(".", 1) emb_dict = bundle_embeddings.get(emb_name, {}) From b2414476ef164ba55cff2508c58b73d23bbc3000 Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Fri, 8 Dec 2023 17:32:41 -0700 Subject: [PATCH 233/342] soft_inpainting now appears in the "inpaint" section, and will not activate unless inpainting is activated. --- scripts/soft_inpainting.py | 43 ++++++++++++++++++++++++++++++++++---- 1 file changed, 39 insertions(+), 4 deletions(-) diff --git a/scripts/soft_inpainting.py b/scripts/soft_inpainting.py index 51f9ca2fe..f10a1e562 100644 --- a/scripts/soft_inpainting.py +++ b/scripts/soft_inpainting.py @@ -32,6 +32,19 @@ class SoftInpaintingSettings: # ------------------- Methods ------------------- +def processing_uses_inpainting(p): + # TODO: Figure out a better way to determine if inpainting is being used by p + if getattr(p, "image_mask", None) is not None: + return True + + if getattr(p, "mask", None) is not None: + return True + + if getattr(p, "nmask", None) is not None: + return True + + return False + def latent_blend(settings, a, b, t): """ @@ -454,8 +467,8 @@ el_ids = SoftInpaintingSettings( class Script(scripts.Script): - def __init__(self): + self.section = "inpaint" self.masks_for_overlay = None self.overlay_images = None @@ -632,6 +645,9 @@ class Script(scripts.Script): if not enabled: return + if not processing_uses_inpainting(p): + return + # Shut off the rounding it normally does. p.mask_round = False @@ -644,6 +660,9 @@ class Script(scripts.Script): if not enabled: return + if not processing_uses_inpainting(p): + return + if mba.is_final_blend: mba.blended_latent = mba.current_latent return @@ -660,11 +679,18 @@ class Script(scripts.Script): if not enabled: return - settings = SoftInpaintingSettings(power, scale, detail_preservation, mask_inf, dif_thresh, dif_contr) + if not processing_uses_inpainting(p): + return + + nmask = getattr(p, "nmask", None) + if nmask is None: + return from modules import images from modules.shared import opts + settings = SoftInpaintingSettings(power, scale, detail_preservation, mask_inf, dif_thresh, dif_contr) + # since the original code puts holes in the existing overlay images, # we have to rebuild them. self.overlay_images = [] @@ -682,14 +708,14 @@ class Script(scripts.Script): if getattr(ps.samples, 'already_decoded', False): self.masks_for_overlay = apply_masks(settings=settings, - nmask=p.nmask, + nmask=nmask, overlay_images=self.overlay_images, width=p.width, height=p.height, paste_to=p.paste_to) else: self.masks_for_overlay = apply_adaptive_masks(settings=settings, - nmask=p.nmask, + nmask=nmask, latent_orig=p.init_latent, latent_processed=ps.samples, overlay_images=self.overlay_images, @@ -702,5 +728,14 @@ class Script(scripts.Script): if not enabled: return + if not processing_uses_inpainting(p): + return + + if self.masks_for_overlay is None: + return + + if self.overlay_images is None: + return + ppmo.mask_for_overlay = self.masks_for_overlay[ppmo.index] ppmo.overlay_image = self.overlay_images[ppmo.index] From f1ff932cafa2bf34fa35f41072f21a8ea5474d84 Mon Sep 17 00:00:00 2001 From: CodeHatchling Date: Fri, 8 Dec 2023 17:33:11 -0700 Subject: [PATCH 234/342] Formatted soft_inpainting. --- scripts/soft_inpainting.py | 26 ++++++++++++++++---------- 1 file changed, 16 insertions(+), 10 deletions(-) diff --git a/scripts/soft_inpainting.py b/scripts/soft_inpainting.py index f10a1e562..d90243442 100644 --- a/scripts/soft_inpainting.py +++ b/scripts/soft_inpainting.py @@ -122,7 +122,7 @@ def get_modified_nmask(settings, nmask, sigma): def apply_adaptive_masks( - settings:SoftInpaintingSettings, + settings: SoftInpaintingSettings, nmask, latent_orig, latent_processed, @@ -137,10 +137,10 @@ def apply_adaptive_masks( # TODO: Bias the blending according to the latent mask, add adjustable parameter for bias control. latent_mask = nmask[0].float() # convert the original mask into a form we use to scale distances for thresholding - mask_scalar = 1-(torch.clamp(latent_mask, min=0, max=1) ** (settings.mask_blend_scale / 2)) - mask_scalar = (0.5 * (1-settings.composite_mask_influence) + mask_scalar = 1 - (torch.clamp(latent_mask, min=0, max=1) ** (settings.mask_blend_scale / 2)) + mask_scalar = (0.5 * (1 - settings.composite_mask_influence) + mask_scalar * settings.composite_mask_influence) - mask_scalar = mask_scalar / (1.00001-mask_scalar) + mask_scalar = mask_scalar / (1.00001 - mask_scalar) mask_scalar = mask_scalar.cpu().numpy() latent_distance = torch.norm(latent_processed - latent_orig, p=2, dim=1) @@ -152,9 +152,9 @@ def apply_adaptive_masks( for i, (distance_map, overlay_image) in enumerate(zip(latent_distance, overlay_images)): converted_mask = distance_map.float().cpu().numpy() converted_mask = weighted_histogram_filter(converted_mask, kernel, kernel_center, - percentile_min=0.9, percentile_max=1, min_width=1) + percentile_min=0.9, percentile_max=1, min_width=1) converted_mask = weighted_histogram_filter(converted_mask, kernel, kernel_center, - percentile_min=0.25, percentile_max=0.75, min_width=1) + percentile_min=0.25, percentile_max=0.75, min_width=1) # The distance at which opacity of original decreases to 50% half_weighted_distance = settings.composite_difference_threshold * mask_scalar @@ -276,6 +276,7 @@ def weighted_histogram_filter(img, kernel, kernel_center, percentile_min=0.0, pe An element of the histogram, its weight and bounds. """ + def __init__(self, value, weight): self.value: float = value self.weight: float = weight @@ -355,6 +356,7 @@ def weighted_histogram_filter(img, kernel, kernel_center, percentile_min=0.0, pe return img_out + def smoothstep(x): """ The smoothstep function, input should be clamped to 0-1 range. @@ -362,6 +364,7 @@ def smoothstep(x): """ return x * x * (3 - 2 * x) + def smootherstep(x): """ The smootherstep function, input should be clamped to 0-1 range. @@ -385,6 +388,7 @@ def get_gaussian_kernel(stddev_radius=1.0, max_radius=2): Returns: (nparray, nparray): A kernel array (shape: (N, N)), its center coordinate (shape: (2)) """ + # Evaluates a 0-1 normalized gaussian function for a given square distance from the mean. def gaussian(sqr_mag): return math.exp(-sqr_mag / (stddev_radius * stddev_radius)) @@ -656,7 +660,8 @@ class Script(scripts.Script): # p.extra_generation_params["Mask rounding"] = False settings.add_generation_params(p.extra_generation_params) - def on_mask_blend(self, p, mba: scripts.MaskBlendArgs, enabled, power, scale, detail_preservation, mask_inf, dif_thresh, dif_contr): + def on_mask_blend(self, p, mba: scripts.MaskBlendArgs, enabled, power, scale, detail_preservation, mask_inf, + dif_thresh, dif_contr): if not enabled: return @@ -675,7 +680,8 @@ class Script(scripts.Script): mba.current_latent, get_modified_nmask(settings, mba.nmask, mba.sigma[0])) - def post_sample(self, p, ps: scripts.PostSampleArgs, enabled, power, scale, detail_preservation, mask_inf, dif_thresh, dif_contr): + def post_sample(self, p, ps: scripts.PostSampleArgs, enabled, power, scale, detail_preservation, mask_inf, + dif_thresh, dif_contr): if not enabled: return @@ -723,8 +729,8 @@ class Script(scripts.Script): height=p.height, paste_to=p.paste_to) - - def postprocess_maskoverlay(self, p, ppmo: scripts.PostProcessMaskOverlayArgs, enabled, power, scale, detail_preservation, mask_inf, dif_thresh, dif_contr): + def postprocess_maskoverlay(self, p, ppmo: scripts.PostProcessMaskOverlayArgs, enabled, power, scale, + detail_preservation, mask_inf, dif_thresh, dif_contr): if not enabled: return From 59429793440fb3cb1624ddcc702c6f9807373203 Mon Sep 17 00:00:00 2001 From: Nuullll Date: Sat, 9 Dec 2023 18:09:45 +0800 Subject: [PATCH 235/342] Fix ControlNet --- modules/xpu_specific.py | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/modules/xpu_specific.py b/modules/xpu_specific.py index ec1ad100a..9bb0a5615 100644 --- a/modules/xpu_specific.py +++ b/modules/xpu_specific.py @@ -51,3 +51,9 @@ if has_xpu: CondFunc('torch.bmm', lambda orig_func, input, mat2, out=None: orig_func(input.to(mat2.dtype), mat2, out=out), lambda orig_func, input, mat2, out=None: input.dtype != mat2.dtype) + CondFunc('torch.cat', + lambda orig_func, tensors, dim=0, out=None: orig_func([t.to(tensors[0].dtype) for t in tensors], dim=dim, out=out), + lambda orig_func, tensors, dim=0, out=None: not all(t.dtype == tensors[0].dtype for t in tensors)) + CondFunc('torch.nn.functional.scaled_dot_product_attention', + lambda orig_func, query, key, value, attn_mask=None, dropout_p=0.0, is_causal=False: orig_func(query, key.to(query.dtype), value.to(query.dtype), attn_mask, dropout_p, is_causal), + lambda orig_func, query, key, value, attn_mask=None, dropout_p=0.0, is_causal=False: query.dtype != key.dtype or query.dtype != value.dtype) \ No newline at end of file From 049d5642e58d572ee8657ac754e72d019eea0e6c Mon Sep 17 00:00:00 2001 From: Nuullll Date: Sat, 9 Dec 2023 18:11:26 +0800 Subject: [PATCH 236/342] Fix format --- modules/xpu_specific.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/xpu_specific.py b/modules/xpu_specific.py index 9bb0a5615..d8da94a0e 100644 --- a/modules/xpu_specific.py +++ b/modules/xpu_specific.py @@ -56,4 +56,4 @@ if has_xpu: lambda orig_func, tensors, dim=0, out=None: not all(t.dtype == tensors[0].dtype for t in tensors)) CondFunc('torch.nn.functional.scaled_dot_product_attention', lambda orig_func, query, key, value, attn_mask=None, dropout_p=0.0, is_causal=False: orig_func(query, key.to(query.dtype), value.to(query.dtype), attn_mask, dropout_p, is_causal), - lambda orig_func, query, key, value, attn_mask=None, dropout_p=0.0, is_causal=False: query.dtype != key.dtype or query.dtype != value.dtype) \ No newline at end of file + lambda orig_func, query, key, value, attn_mask=None, dropout_p=0.0, is_causal=False: query.dtype != key.dtype or query.dtype != value.dtype) From 39ec4cfea9040bc94e639eb4aa8ab8ed37a68f01 Mon Sep 17 00:00:00 2001 From: kaalibro Date: Sat, 9 Dec 2023 19:12:59 +0600 Subject: [PATCH 237/342] Re-add setting lost as part of e294e46 --- modules/shared_options.py | 1 + 1 file changed, 1 insertion(+) diff --git a/modules/shared_options.py b/modules/shared_options.py index e5de0d018..acb6e2d48 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -256,6 +256,7 @@ options_templates.update(options_section(('ui_prompt_editing', "Prompt editing", "keyedit_precision_extra": OptionInfo(0.05, "Precision for when editing the prompt with Ctrl+up/down", gr.Slider, {"minimum": 0.01, "maximum": 0.2, "step": 0.001}), "keyedit_delimiters": OptionInfo(r".,\/!?%^*;:{}=`~() ", "Word delimiters when editing the prompt with Ctrl+up/down"), "keyedit_delimiters_whitespace": OptionInfo(["Tab", "Carriage Return", "Line Feed"], "Ctrl+up/down whitespace delimiters", gr.CheckboxGroup, lambda: {"choices": ["Tab", "Carriage Return", "Line Feed"]}), + "keyedit_move": OptionInfo(True, "Alt+left/right moves prompt elements"), "disable_token_counters": OptionInfo(False, "Disable prompt token counters").needs_reload_ui(), })) From 9c201550ddae0b33367adfb99bcbb57ba9b207a9 Mon Sep 17 00:00:00 2001 From: kaalibro Date: Sat, 9 Dec 2023 21:04:45 +0600 Subject: [PATCH 238/342] Add keyboard shortcuts for generation (Removed Alt+Enter) Ctrl+Enter to start/restart generation (New) Alt/Option+Enter to skip generation (New) Ctrl+Alt/Option+Enter to interrupt generation --- modules/ui_toprow.py | 4 ++-- script.js | 23 +++++++++++++++++++---- 2 files changed, 21 insertions(+), 6 deletions(-) diff --git a/modules/ui_toprow.py b/modules/ui_toprow.py index 88838f977..c3865e3d9 100644 --- a/modules/ui_toprow.py +++ b/modules/ui_toprow.py @@ -79,11 +79,11 @@ class Toprow: def create_prompts(self): with gr.Column(elem_id=f"{self.id_part}_prompt_container", elem_classes=["prompt-container-compact"] if self.is_compact else [], scale=6): with gr.Row(elem_id=f"{self.id_part}_prompt_row", elem_classes=["prompt-row"]): - self.prompt = gr.Textbox(label="Prompt", elem_id=f"{self.id_part}_prompt", show_label=False, lines=3, placeholder="Prompt (press Ctrl+Enter or Alt+Enter to generate)", elem_classes=["prompt"]) + self.prompt = gr.Textbox(label="Prompt", elem_id=f"{self.id_part}_prompt", show_label=False, lines=3, placeholder="Prompt\n(Press Ctrl+Enter to generate, Alt+Enter to skip, Ctrl+Alt+Enter to interrupt)", elem_classes=["prompt"]) self.prompt_img = gr.File(label="", elem_id=f"{self.id_part}_prompt_image", file_count="single", type="binary", visible=False) with gr.Row(elem_id=f"{self.id_part}_neg_prompt_row", elem_classes=["prompt-row"]): - self.negative_prompt = gr.Textbox(label="Negative prompt", elem_id=f"{self.id_part}_neg_prompt", show_label=False, lines=3, placeholder="Negative prompt (press Ctrl+Enter or Alt+Enter to generate)", elem_classes=["prompt"]) + self.negative_prompt = gr.Textbox(label="Negative prompt", elem_id=f"{self.id_part}_neg_prompt", show_label=False, lines=3, placeholder="Negative prompt\n(Press Ctrl+Enter to generate, Alt+Enter to skip, Ctrl+Alt+Enter to interrupt)", elem_classes=["prompt"]) self.prompt_img.change( fn=modules.images.image_data, diff --git a/script.js b/script.js index c0e678ea7..69598f454 100644 --- a/script.js +++ b/script.js @@ -121,16 +121,21 @@ document.addEventListener("DOMContentLoaded", function() { }); /** - * Add a ctrl+enter as a shortcut to start a generation + * Add keyboard shortcuts: + * Ctrl+Enter to start/restart a generation + * Alt/Option+Enter to skip a generation + * Alt/Option+Ctrl+Enter to interrupt a generation */ document.addEventListener('keydown', function(e) { const isEnter = e.key === 'Enter' || e.keyCode === 13; - const isModifierKey = e.metaKey || e.ctrlKey || e.altKey; + const isCtrlKey = e.metaKey || e.ctrlKey; + const isAltKey = e.altKey; - const interruptButton = get_uiCurrentTabContent().querySelector('button[id$=_interrupt]'); const generateButton = get_uiCurrentTabContent().querySelector('button[id$=_generate]'); + const interruptButton = get_uiCurrentTabContent().querySelector('button[id$=_interrupt]'); + const skipButton = get_uiCurrentTabContent().querySelector('button[id$=_skip]'); - if (isEnter && isModifierKey) { + if (isCtrlKey && isEnter && !isAltKey) { if (interruptButton.style.display === 'block') { interruptButton.click(); const callback = (mutationList) => { @@ -150,6 +155,16 @@ document.addEventListener('keydown', function(e) { } e.preventDefault(); } + + if (isAltKey && isEnter && !isCtrlKey) { + skipButton.click(); + e.preventDefault(); + } + + if (isAltKey && isCtrlKey && isEnter) { + interruptButton.click(); + e.preventDefault(); + } }); /** From 1a79a5049bdfef285235e83f37b201e39dd54f81 Mon Sep 17 00:00:00 2001 From: kaalibro Date: Sat, 9 Dec 2023 22:35:31 +0600 Subject: [PATCH 239/342] Assign id for "extra_options". Replace numeric field with slider in Settings. --- .../extra-options-section/scripts/extra_options_section.py | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/extensions-builtin/extra-options-section/scripts/extra_options_section.py b/extensions-builtin/extra-options-section/scripts/extra_options_section.py index a903df625..b9867fe63 100644 --- a/extensions-builtin/extra-options-section/scripts/extra_options_section.py +++ b/extensions-builtin/extra-options-section/scripts/extra_options_section.py @@ -23,11 +23,12 @@ class ExtraOptionsSection(scripts.Script): self.setting_names = [] self.infotext_fields = [] extra_options = shared.opts.extra_options_img2img if is_img2img else shared.opts.extra_options_txt2img + elem_id_tabname = "extra_options_" + ("img2img" if is_img2img else "txt2img") mapping = {k: v for v, k in generation_parameters_copypaste.infotext_to_setting_name_mapping} with gr.Blocks() as interface: - with gr.Accordion("Options", open=False) if shared.opts.extra_options_accordion and extra_options else gr.Group(): + with gr.Accordion("Options", open=False, elem_id=elem_id_tabname) if shared.opts.extra_options_accordion and extra_options else gr.Group(elem_id=elem_id_tabname): row_count = math.ceil(len(extra_options) / shared.opts.extra_options_cols) @@ -70,7 +71,7 @@ This page allows you to add some settings to the main interface of txt2img and i """), "extra_options_txt2img": shared.OptionInfo([], "Settings for txt2img", ui_components.DropdownMulti, lambda: {"choices": list(shared.opts.data_labels.keys())}).js("info", "settingsHintsShowQuicksettings").info("setting entries that also appear in txt2img interfaces").needs_reload_ui(), "extra_options_img2img": shared.OptionInfo([], "Settings for img2img", ui_components.DropdownMulti, lambda: {"choices": list(shared.opts.data_labels.keys())}).js("info", "settingsHintsShowQuicksettings").info("setting entries that also appear in img2img interfaces").needs_reload_ui(), - "extra_options_cols": shared.OptionInfo(1, "Number of columns for added settings", gr.Number, {"precision": 0}).needs_reload_ui(), + "extra_options_cols": shared.OptionInfo(1, "Number of columns for added settings", gr.Slider, {"step": 1, "minimum": 1, "maximum": 6}).needs_reload_ui(), "extra_options_accordion": shared.OptionInfo(False, "Place added settings into an accordion").needs_reload_ui() })) From 5381405eaa1e809e5cfb97522bd4c19d3c946079 Mon Sep 17 00:00:00 2001 From: drhead <1313496+drhead@users.noreply.github.com> Date: Sat, 9 Dec 2023 14:09:28 -0500 Subject: [PATCH 240/342] re-derive sqrt alpha bar and sqrt one minus alphabar This is the only place these values are ever referenced outside of training code so this change is very justifiable and more consistent. --- modules/sd_samplers_timesteps.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/sd_samplers_timesteps.py b/modules/sd_samplers_timesteps.py index b17a8f93c..c4bd5c127 100644 --- a/modules/sd_samplers_timesteps.py +++ b/modules/sd_samplers_timesteps.py @@ -36,7 +36,7 @@ class CompVisTimestepsVDenoiser(torch.nn.Module): self.inner_model = model def predict_eps_from_z_and_v(self, x_t, t, v): - return self.inner_model.sqrt_alphas_cumprod[t.to(torch.int), None, None, None] * v + self.inner_model.sqrt_one_minus_alphas_cumprod[t.to(torch.int), None, None, None] * x_t + return torch.sqrt(self.inner_model.alphas_cumprod)[t.to(torch.int), None, None, None] * v + torch.sqrt(1 - self.inner_model.alphas_cumprod)[t.to(torch.int), None, None, None] * x_t def forward(self, input, timesteps, **kwargs): model_output = self.inner_model.apply_model(input, timesteps, **kwargs) From 23a0e60b9bf90a80f8af9732cc6495fbfce2ea21 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Sun, 10 Dec 2023 14:03:41 +0900 Subject: [PATCH 241/342] fix save styles --- modules/styles.py | 25 +++++-------------------- 1 file changed, 5 insertions(+), 20 deletions(-) diff --git a/modules/styles.py b/modules/styles.py index 7fb6c2e11..07588945e 100644 --- a/modules/styles.py +++ b/modules/styles.py @@ -155,10 +155,8 @@ class StyleDatabase: row["name"], prompt, negative_prompt, path ) - def get_style_paths(self) -> list(): - """ - Returns a list of all distinct paths, including the default path, of - files that styles are loaded from.""" + def get_style_paths(self) -> set: + """Returns a set of all distinct paths of files that styles are loaded from.""" # Update any styles without a path to the default path for style in list(self.styles.values()): if not style.path: @@ -172,9 +170,9 @@ class StyleDatabase: style_paths.add(style.path) # Remove any paths for styles that are just list dividers - style_paths.remove("do_not_save") + style_paths.discard("do_not_save") - return list(style_paths) + return style_paths def get_style_prompts(self, styles): return [self.styles.get(x, self.no_style).prompt for x in styles] @@ -196,20 +194,7 @@ class StyleDatabase: # The path argument is deprecated, but kept for backwards compatibility _ = path - # Update any styles without a path to the default path - for style in list(self.styles.values()): - if not style.path: - self.styles[style.name] = style._replace(path=self.default_path) - - # Create a list of all distinct paths, including the default path - style_paths = set() - style_paths.add(self.default_path) - for _, style in self.styles.items(): - if style.path: - style_paths.add(style.path) - - # Remove any paths for styles that are just list dividers - style_paths.remove("do_not_save") + style_paths = self.get_style_paths() csv_names = [os.path.split(path)[1].lower() for path in style_paths] From 8b74389e76a7678e972583ef16100e90e1519e55 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Sun, 10 Dec 2023 15:48:16 +0900 Subject: [PATCH 242/342] fix styles.csv filename --- modules/styles.py | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) diff --git a/modules/styles.py b/modules/styles.py index 07588945e..81d9800d1 100644 --- a/modules/styles.py +++ b/modules/styles.py @@ -98,10 +98,8 @@ class StyleDatabase: self.path = path folder, file = os.path.split(self.path) - self.default_file = file.split("*")[0] + ".csv" - if self.default_file == ".csv": - self.default_file = "styles.csv" - self.default_path = os.path.join(folder, self.default_file) + filename, _, ext = file.partition('*') + self.default_path = os.path.join(folder, filename + ext) self.prompt_fields = [field for field in PromptStyle._fields if field != "path"] From 6b8143a84e112f029ee1868b6ab98b1d2c773ead Mon Sep 17 00:00:00 2001 From: kaalibro Date: Sun, 10 Dec 2023 15:35:06 +0600 Subject: [PATCH 243/342] Number of columns slider: max count set to 20, add description info --- .../extra-options-section/scripts/extra_options_section.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/extensions-builtin/extra-options-section/scripts/extra_options_section.py b/extensions-builtin/extra-options-section/scripts/extra_options_section.py index b9867fe63..ac2c3de46 100644 --- a/extensions-builtin/extra-options-section/scripts/extra_options_section.py +++ b/extensions-builtin/extra-options-section/scripts/extra_options_section.py @@ -71,7 +71,7 @@ This page allows you to add some settings to the main interface of txt2img and i """), "extra_options_txt2img": shared.OptionInfo([], "Settings for txt2img", ui_components.DropdownMulti, lambda: {"choices": list(shared.opts.data_labels.keys())}).js("info", "settingsHintsShowQuicksettings").info("setting entries that also appear in txt2img interfaces").needs_reload_ui(), "extra_options_img2img": shared.OptionInfo([], "Settings for img2img", ui_components.DropdownMulti, lambda: {"choices": list(shared.opts.data_labels.keys())}).js("info", "settingsHintsShowQuicksettings").info("setting entries that also appear in img2img interfaces").needs_reload_ui(), - "extra_options_cols": shared.OptionInfo(1, "Number of columns for added settings", gr.Slider, {"step": 1, "minimum": 1, "maximum": 6}).needs_reload_ui(), + "extra_options_cols": shared.OptionInfo(1, "Number of columns for added settings", gr.Slider, {"step": 1, "minimum": 1, "maximum": 20}).info("displayed amount will depend on the actual browser window width").needs_reload_ui(), "extra_options_accordion": shared.OptionInfo(False, "Place added settings into an accordion").needs_reload_ui() })) From 1d42babd324b933bae317cb427fe0513138954f4 Mon Sep 17 00:00:00 2001 From: kaalibro Date: Sun, 10 Dec 2023 16:28:56 +0600 Subject: [PATCH 244/342] Replace Ctrl+Alt+Enter with Esc --- modules/ui_toprow.py | 4 ++-- script.js | 17 +++++++++++------ 2 files changed, 13 insertions(+), 8 deletions(-) diff --git a/modules/ui_toprow.py b/modules/ui_toprow.py index c3865e3d9..9caf8faa2 100644 --- a/modules/ui_toprow.py +++ b/modules/ui_toprow.py @@ -79,11 +79,11 @@ class Toprow: def create_prompts(self): with gr.Column(elem_id=f"{self.id_part}_prompt_container", elem_classes=["prompt-container-compact"] if self.is_compact else [], scale=6): with gr.Row(elem_id=f"{self.id_part}_prompt_row", elem_classes=["prompt-row"]): - self.prompt = gr.Textbox(label="Prompt", elem_id=f"{self.id_part}_prompt", show_label=False, lines=3, placeholder="Prompt\n(Press Ctrl+Enter to generate, Alt+Enter to skip, Ctrl+Alt+Enter to interrupt)", elem_classes=["prompt"]) + self.prompt = gr.Textbox(label="Prompt", elem_id=f"{self.id_part}_prompt", show_label=False, lines=3, placeholder="Prompt\n(Press Ctrl+Enter to generate, Alt+Enter to skip, Esc to interrupt)", elem_classes=["prompt"]) self.prompt_img = gr.File(label="", elem_id=f"{self.id_part}_prompt_image", file_count="single", type="binary", visible=False) with gr.Row(elem_id=f"{self.id_part}_neg_prompt_row", elem_classes=["prompt-row"]): - self.negative_prompt = gr.Textbox(label="Negative prompt", elem_id=f"{self.id_part}_neg_prompt", show_label=False, lines=3, placeholder="Negative prompt\n(Press Ctrl+Enter to generate, Alt+Enter to skip, Ctrl+Alt+Enter to interrupt)", elem_classes=["prompt"]) + self.negative_prompt = gr.Textbox(label="Negative prompt", elem_id=f"{self.id_part}_neg_prompt", show_label=False, lines=3, placeholder="Negative prompt\n(Press Ctrl+Enter to generate, Alt+Enter to skip, Esc to interrupt)", elem_classes=["prompt"]) self.prompt_img.change( fn=modules.images.image_data, diff --git a/script.js b/script.js index 69598f454..44950090a 100644 --- a/script.js +++ b/script.js @@ -124,18 +124,19 @@ document.addEventListener("DOMContentLoaded", function() { * Add keyboard shortcuts: * Ctrl+Enter to start/restart a generation * Alt/Option+Enter to skip a generation - * Alt/Option+Ctrl+Enter to interrupt a generation + * Esc to interrupt a generation */ document.addEventListener('keydown', function(e) { const isEnter = e.key === 'Enter' || e.keyCode === 13; const isCtrlKey = e.metaKey || e.ctrlKey; const isAltKey = e.altKey; + const isEsc = e.key === 'Escape'; const generateButton = get_uiCurrentTabContent().querySelector('button[id$=_generate]'); const interruptButton = get_uiCurrentTabContent().querySelector('button[id$=_interrupt]'); const skipButton = get_uiCurrentTabContent().querySelector('button[id$=_skip]'); - if (isCtrlKey && isEnter && !isAltKey) { + if (isCtrlKey && isEnter) { if (interruptButton.style.display === 'block') { interruptButton.click(); const callback = (mutationList) => { @@ -156,14 +157,18 @@ document.addEventListener('keydown', function(e) { e.preventDefault(); } - if (isAltKey && isEnter && !isCtrlKey) { + if (isAltKey && isEnter) { skipButton.click(); e.preventDefault(); } - if (isAltKey && isCtrlKey && isEnter) { - interruptButton.click(); - e.preventDefault(); + if (isEsc) { + if (!globalPopup || globalPopup.style.display === "none") { + interruptButton.click(); + e.preventDefault(); + } else { + closePopup(); + } } }); From cee1a4065162982e18f32761259d9107538c2d93 Mon Sep 17 00:00:00 2001 From: kaalibro Date: Sun, 10 Dec 2023 17:06:12 +0600 Subject: [PATCH 245/342] Fix linter issues --- script.js | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/script.js b/script.js index 44950090a..354154b01 100644 --- a/script.js +++ b/script.js @@ -163,11 +163,13 @@ document.addEventListener('keydown', function(e) { } if (isEsc) { + const globalPopup = document.querySelector('.global-popup'); if (!globalPopup || globalPopup.style.display === "none") { interruptButton.click(); e.preventDefault(); } else { - closePopup(); + if (!globalPopup) return; + globalPopup.style.display = "none"; } } }); From 6513470f0db1aed1b0a5200634e8e02f7c05e932 Mon Sep 17 00:00:00 2001 From: kaalibro Date: Mon, 11 Dec 2023 18:06:08 +0600 Subject: [PATCH 246/342] Remove unnecessary 'else', add 'lightboxModal' check --- script.js | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) diff --git a/script.js b/script.js index 354154b01..be1bc317e 100644 --- a/script.js +++ b/script.js @@ -164,12 +164,11 @@ document.addEventListener('keydown', function(e) { if (isEsc) { const globalPopup = document.querySelector('.global-popup'); - if (!globalPopup || globalPopup.style.display === "none") { + const lightboxModal = document.querySelector('#lightboxModal'); + if (!globalPopup || globalPopup.style.display === 'none') { + if (document.activeElement === lightboxModal) return; interruptButton.click(); e.preventDefault(); - } else { - if (!globalPopup) return; - globalPopup.style.display = "none"; } } }); From cc41cc4349514bbfeb9f37445c931a050b076bd6 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Wed, 13 Dec 2023 02:06:56 +0900 Subject: [PATCH 247/342] on mouse hover show / hide modal image viewer icons --- style.css | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) diff --git a/style.css b/style.css index ee39a57b7..ec449bdea 100644 --- a/style.css +++ b/style.css @@ -749,6 +749,22 @@ table.popup-table .link{ display: none; } +@media (pointer: fine) { + .modalPrev:hover, + .modalNext:hover, + .modalControls:hover ~ .modalPrev, + .modalControls:hover ~ .modalNext, + .modalControls:hover .cursor { + opacity: 1; + } + + .modalPrev, + .modalNext, + .modalControls .cursor { + opacity: 0; + } +} + /* context menu (ie for the generate button) */ #context-menu{ From bda86f0fd9653657c146f7c1128f92771d16ad4e Mon Sep 17 00:00:00 2001 From: Hina <102651522+HinaHyugaHime@users.noreply.github.com> Date: Tue, 12 Dec 2023 19:39:14 -0600 Subject: [PATCH 248/342] Update webui.sh --- webui.sh | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) diff --git a/webui.sh b/webui.sh index 3d0f87eed..046ecf9d0 100755 --- a/webui.sh +++ b/webui.sh @@ -131,7 +131,7 @@ case "$gpu_info" in if [[ $(bc <<< "$pyv <= 3.10") -eq 1 ]] then # Navi users will still use torch 1.13 because 2.0 does not seem to work. - export TORCH_COMMAND="pip install torch==1.13.1+rocm5.2 torchvision==0.14.1+rocm5.2 --index-url https://download.pytorch.org/whl/rocm5.2" + export TORCH_COMMAND="pip install --pre torch torchvision --index-url https://download.pytorch.org/whl/nightly/rocm5.6" else printf "\e[1m\e[31mERROR: RX 5000 series GPUs must be using at max python 3.10, aborting...\e[0m" exit 1 @@ -141,9 +141,8 @@ case "$gpu_info" in *"Navi 2"*) export HSA_OVERRIDE_GFX_VERSION=10.3.0 ;; *"Navi 3"*) [[ -z "${TORCH_COMMAND}" ]] && \ - export TORCH_COMMAND="pip install --pre torch torchvision --index-url https://download.pytorch.org/whl/nightly/rocm5.6" - # Navi 3 needs at least 5.5 which is only on the nightly chain, previous versions are no longer online (torch==2.1.0.dev-20230614+rocm5.5 torchvision==0.16.0.dev-20230614+rocm5.5 torchaudio==2.1.0.dev-20230614+rocm5.5) - # so switch to nightly rocm5.6 without explicit versions this time + export TORCH_COMMAND="pip install --pre torch torchvision --index-url https://download.pytorch.org/whl/nightly/rocm5.7" + ;; *"Renoir"*) export HSA_OVERRIDE_GFX_VERSION=9.0.0 printf "\n%s\n" "${delimiter}" From 89cfbc3bbe401fe1655afb07edbae34ec6af7aca Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Wed, 13 Dec 2023 12:22:13 +0200 Subject: [PATCH 249/342] Allow pasting in WIDTHxHEIGHT strings into the width/height fields --- javascript/ui.js | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+) diff --git a/javascript/ui.js b/javascript/ui.js index 410fc44e3..18c9f891a 100644 --- a/javascript/ui.js +++ b/javascript/ui.js @@ -215,9 +215,33 @@ function restoreProgressImg2img() { } +/** + * Configure the width and height elements on `tabname` to accept + * pasting of resolutions in the form of "width x height". + */ +function setupResolutionPasting(tabname) { + var width = gradioApp().querySelector(`#${tabname}_width input[type=number]`); + var height = gradioApp().querySelector(`#${tabname}_height input[type=number]`); + for (const el of [width, height]) { + el.addEventListener('paste', function(event) { + var pasteData = event.clipboardData.getData('text/plain'); + var parsed = pasteData.match(/^\s*(\d+)\D+(\d+)\s*$/); + if (parsed) { + width.value = parsed[1]; + height.value = parsed[2]; + updateInput(width); + updateInput(height); + event.preventDefault(); + } + }); + } +} + onUiLoaded(function() { showRestoreProgressButton('txt2img', localGet("txt2img_task_id")); showRestoreProgressButton('img2img', localGet("img2img_task_id")); + setupResolutionPasting('txt2img'); + setupResolutionPasting('img2img'); }); From 735c9e8059384d4f640e5582413c30871f83eac5 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Thu, 14 Dec 2023 01:38:32 +0800 Subject: [PATCH 250/342] Fix network_oft --- extensions-builtin/Lora/network_oft.py | 21 +++++++++++---------- 1 file changed, 11 insertions(+), 10 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index 05c378118..44465f7aa 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -53,12 +53,17 @@ class NetworkModuleOFT(network.NetworkModule): self.constraint = None self.block_size, self.num_blocks = factorization(self.out_dim, self.dim) - def calc_updown_kb(self, orig_weight, multiplier): + def calc_updown(self, orig_weight): + I = torch.eye(self.block_size, device=self.oft_blocks.device) oft_blocks = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) - oft_blocks = oft_blocks - oft_blocks.transpose(1, 2) # ensure skew-symmetric orthogonal matrix + if self.is_kohya: + block_Q = oft_blocks - oft_blocks.transpose(1, 2) # ensure skew-symmetric orthogonal matrix + norm_Q = torch.norm(block_Q.flatten()) + new_norm_Q = torch.clamp(norm_Q, max=self.constraint) + block_Q = block_Q * ((new_norm_Q + 1e-8) / (norm_Q + 1e-8)) + oft_blocks = torch.matmul(I + block_Q, (I - block_Q).float().inverse()) R = oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) - R = R * multiplier + torch.eye(self.block_size, device=orig_weight.device) # This errors out for MultiheadAttention, might need to be handled up-stream merged_weight = rearrange(orig_weight, '(k n) ... -> k n ...', k=self.num_blocks, n=self.block_size) @@ -70,15 +75,10 @@ class NetworkModuleOFT(network.NetworkModule): merged_weight = rearrange(merged_weight, 'k m ... -> (k m) ...') updown = merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight + print(torch.norm(updown)) output_shape = orig_weight.shape return self.finalize_updown(updown, orig_weight, output_shape) - def calc_updown(self, orig_weight): - # if alpha is a very small number as in coft, calc_scale() will return a almost zero number so we ignore it - multiplier = self.multiplier() - return self.calc_updown_kb(orig_weight, multiplier) - - # override to remove the multiplier/scale factor; it's already multiplied in get_weight def finalize_updown(self, updown, orig_weight, output_shape, ex_bias=None): if self.bias is not None: updown = updown.reshape(self.bias.shape) @@ -94,4 +94,5 @@ class NetworkModuleOFT(network.NetworkModule): if ex_bias is not None: ex_bias = ex_bias * self.multiplier() - return updown, ex_bias + # Ignore calc_scale, which is not used in OFT. + return updown * self.multiplier(), ex_bias From 265bc26c21264d63956e8f30f1ce31dec917fc76 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Thu, 14 Dec 2023 01:43:24 +0800 Subject: [PATCH 251/342] Use self.scale instead of custom finalize --- extensions-builtin/Lora/network_oft.py | 20 ++------------------ 1 file changed, 2 insertions(+), 18 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index 44465f7aa..e3ae61a22 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -21,6 +21,8 @@ class NetworkModuleOFT(network.NetworkModule): self.lin_module = None self.org_module: list[torch.Module] = [self.sd_module] + self.scale = 1.0 + # kohya-ss if "oft_blocks" in weights.w.keys(): self.is_kohya = True @@ -78,21 +80,3 @@ class NetworkModuleOFT(network.NetworkModule): print(torch.norm(updown)) output_shape = orig_weight.shape return self.finalize_updown(updown, orig_weight, output_shape) - - def finalize_updown(self, updown, orig_weight, output_shape, ex_bias=None): - if self.bias is not None: - updown = updown.reshape(self.bias.shape) - updown += self.bias.to(orig_weight.device, dtype=orig_weight.dtype) - updown = updown.reshape(output_shape) - - if len(output_shape) == 4: - updown = updown.reshape(output_shape) - - if orig_weight.size().numel() == updown.size().numel(): - updown = updown.reshape(orig_weight.shape) - - if ex_bias is not None: - ex_bias = ex_bias * self.multiplier() - - # Ignore calc_scale, which is not used in OFT. - return updown * self.multiplier(), ex_bias From 8fc67f3851babd4575d3312b931d5e7c2b0c78c6 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Thu, 14 Dec 2023 01:44:49 +0800 Subject: [PATCH 252/342] remove debug print --- extensions-builtin/Lora/network_oft.py | 1 - 1 file changed, 1 deletion(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index e3ae61a22..ff4eb59b1 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -77,6 +77,5 @@ class NetworkModuleOFT(network.NetworkModule): merged_weight = rearrange(merged_weight, 'k m ... -> (k m) ...') updown = merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight - print(torch.norm(updown)) output_shape = orig_weight.shape return self.finalize_updown(updown, orig_weight, output_shape) From 3772a82a70769fe1aac884a75bf5a3313fb83328 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Thu, 14 Dec 2023 01:47:13 +0800 Subject: [PATCH 253/342] better naming and correct order for device. --- extensions-builtin/Lora/network_oft.py | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index ff4eb59b1..fa647020f 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -56,14 +56,15 @@ class NetworkModuleOFT(network.NetworkModule): self.block_size, self.num_blocks = factorization(self.out_dim, self.dim) def calc_updown(self, orig_weight): - I = torch.eye(self.block_size, device=self.oft_blocks.device) oft_blocks = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) + eye = torch.eye(self.block_size, device=self.oft_blocks.device) + if self.is_kohya: block_Q = oft_blocks - oft_blocks.transpose(1, 2) # ensure skew-symmetric orthogonal matrix norm_Q = torch.norm(block_Q.flatten()) new_norm_Q = torch.clamp(norm_Q, max=self.constraint) block_Q = block_Q * ((new_norm_Q + 1e-8) / (norm_Q + 1e-8)) - oft_blocks = torch.matmul(I + block_Q, (I - block_Q).float().inverse()) + oft_blocks = torch.matmul(eye + block_Q, (eye - block_Q).float().inverse()) R = oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) From 6ef0ff39f2a35a02e5380e522e1dff3eafd7ccfc Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Thu, 14 Dec 2023 09:39:57 +0300 Subject: [PATCH 254/342] Merge pull request #14300 from AUTOMATIC1111/oft_fixes Fix wrong implementation in network_oft --- extensions-builtin/Lora/network_oft.py | 37 ++++++++------------------ 1 file changed, 11 insertions(+), 26 deletions(-) diff --git a/extensions-builtin/Lora/network_oft.py b/extensions-builtin/Lora/network_oft.py index 05c378118..fa647020f 100644 --- a/extensions-builtin/Lora/network_oft.py +++ b/extensions-builtin/Lora/network_oft.py @@ -21,6 +21,8 @@ class NetworkModuleOFT(network.NetworkModule): self.lin_module = None self.org_module: list[torch.Module] = [self.sd_module] + self.scale = 1.0 + # kohya-ss if "oft_blocks" in weights.w.keys(): self.is_kohya = True @@ -53,12 +55,18 @@ class NetworkModuleOFT(network.NetworkModule): self.constraint = None self.block_size, self.num_blocks = factorization(self.out_dim, self.dim) - def calc_updown_kb(self, orig_weight, multiplier): + def calc_updown(self, orig_weight): oft_blocks = self.oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) - oft_blocks = oft_blocks - oft_blocks.transpose(1, 2) # ensure skew-symmetric orthogonal matrix + eye = torch.eye(self.block_size, device=self.oft_blocks.device) + + if self.is_kohya: + block_Q = oft_blocks - oft_blocks.transpose(1, 2) # ensure skew-symmetric orthogonal matrix + norm_Q = torch.norm(block_Q.flatten()) + new_norm_Q = torch.clamp(norm_Q, max=self.constraint) + block_Q = block_Q * ((new_norm_Q + 1e-8) / (norm_Q + 1e-8)) + oft_blocks = torch.matmul(eye + block_Q, (eye - block_Q).float().inverse()) R = oft_blocks.to(orig_weight.device, dtype=orig_weight.dtype) - R = R * multiplier + torch.eye(self.block_size, device=orig_weight.device) # This errors out for MultiheadAttention, might need to be handled up-stream merged_weight = rearrange(orig_weight, '(k n) ... -> k n ...', k=self.num_blocks, n=self.block_size) @@ -72,26 +80,3 @@ class NetworkModuleOFT(network.NetworkModule): updown = merged_weight.to(orig_weight.device, dtype=orig_weight.dtype) - orig_weight output_shape = orig_weight.shape return self.finalize_updown(updown, orig_weight, output_shape) - - def calc_updown(self, orig_weight): - # if alpha is a very small number as in coft, calc_scale() will return a almost zero number so we ignore it - multiplier = self.multiplier() - return self.calc_updown_kb(orig_weight, multiplier) - - # override to remove the multiplier/scale factor; it's already multiplied in get_weight - def finalize_updown(self, updown, orig_weight, output_shape, ex_bias=None): - if self.bias is not None: - updown = updown.reshape(self.bias.shape) - updown += self.bias.to(orig_weight.device, dtype=orig_weight.dtype) - updown = updown.reshape(output_shape) - - if len(output_shape) == 4: - updown = updown.reshape(output_shape) - - if orig_weight.size().numel() == updown.size().numel(): - updown = updown.reshape(orig_weight.shape) - - if ex_bias is not None: - ex_bias = ex_bias * self.multiplier() - - return updown, ex_bias From c7cd9b441d9061f33b7b88be519fb4c6e5b8bc1e Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Thu, 14 Dec 2023 09:41:18 +0300 Subject: [PATCH 255/342] Merge pull request #14296 from akx/paste-resolution Allow pasting in WIDTHxHEIGHT strings into the width/height fields --- javascript/ui.js | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+) diff --git a/javascript/ui.js b/javascript/ui.js index 410fc44e3..18c9f891a 100644 --- a/javascript/ui.js +++ b/javascript/ui.js @@ -215,9 +215,33 @@ function restoreProgressImg2img() { } +/** + * Configure the width and height elements on `tabname` to accept + * pasting of resolutions in the form of "width x height". + */ +function setupResolutionPasting(tabname) { + var width = gradioApp().querySelector(`#${tabname}_width input[type=number]`); + var height = gradioApp().querySelector(`#${tabname}_height input[type=number]`); + for (const el of [width, height]) { + el.addEventListener('paste', function(event) { + var pasteData = event.clipboardData.getData('text/plain'); + var parsed = pasteData.match(/^\s*(\d+)\D+(\d+)\s*$/); + if (parsed) { + width.value = parsed[1]; + height.value = parsed[2]; + updateInput(width); + updateInput(height); + event.preventDefault(); + } + }); + } +} + onUiLoaded(function() { showRestoreProgressButton('txt2img', localGet("txt2img_task_id")); showRestoreProgressButton('img2img', localGet("img2img_task_id")); + setupResolutionPasting('txt2img'); + setupResolutionPasting('img2img'); }); From b55f09c4e13c082590bc64cd792a0b7bd46c1c0d Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Thu, 14 Dec 2023 09:46:05 +0300 Subject: [PATCH 256/342] Merge pull request #14270 from kaalibro/extra-options-elem-id Assign id for "extra_options". Replace numeric field with slider. --- .../extra-options-section/scripts/extra_options_section.py | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/extensions-builtin/extra-options-section/scripts/extra_options_section.py b/extensions-builtin/extra-options-section/scripts/extra_options_section.py index a903df625..ac2c3de46 100644 --- a/extensions-builtin/extra-options-section/scripts/extra_options_section.py +++ b/extensions-builtin/extra-options-section/scripts/extra_options_section.py @@ -23,11 +23,12 @@ class ExtraOptionsSection(scripts.Script): self.setting_names = [] self.infotext_fields = [] extra_options = shared.opts.extra_options_img2img if is_img2img else shared.opts.extra_options_txt2img + elem_id_tabname = "extra_options_" + ("img2img" if is_img2img else "txt2img") mapping = {k: v for v, k in generation_parameters_copypaste.infotext_to_setting_name_mapping} with gr.Blocks() as interface: - with gr.Accordion("Options", open=False) if shared.opts.extra_options_accordion and extra_options else gr.Group(): + with gr.Accordion("Options", open=False, elem_id=elem_id_tabname) if shared.opts.extra_options_accordion and extra_options else gr.Group(elem_id=elem_id_tabname): row_count = math.ceil(len(extra_options) / shared.opts.extra_options_cols) @@ -70,7 +71,7 @@ This page allows you to add some settings to the main interface of txt2img and i """), "extra_options_txt2img": shared.OptionInfo([], "Settings for txt2img", ui_components.DropdownMulti, lambda: {"choices": list(shared.opts.data_labels.keys())}).js("info", "settingsHintsShowQuicksettings").info("setting entries that also appear in txt2img interfaces").needs_reload_ui(), "extra_options_img2img": shared.OptionInfo([], "Settings for img2img", ui_components.DropdownMulti, lambda: {"choices": list(shared.opts.data_labels.keys())}).js("info", "settingsHintsShowQuicksettings").info("setting entries that also appear in img2img interfaces").needs_reload_ui(), - "extra_options_cols": shared.OptionInfo(1, "Number of columns for added settings", gr.Number, {"precision": 0}).needs_reload_ui(), + "extra_options_cols": shared.OptionInfo(1, "Number of columns for added settings", gr.Slider, {"step": 1, "minimum": 1, "maximum": 20}).info("displayed amount will depend on the actual browser window width").needs_reload_ui(), "extra_options_accordion": shared.OptionInfo(False, "Place added settings into an accordion").needs_reload_ui() })) From 888b928f0da7da4a2dfa4519e95ac17a3e5562f7 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Thu, 14 Dec 2023 09:48:14 +0300 Subject: [PATCH 257/342] Merge pull request #14276 from AUTOMATIC1111/fix-styles Fix styles --- modules/styles.py | 31 +++++++------------------------ 1 file changed, 7 insertions(+), 24 deletions(-) diff --git a/modules/styles.py b/modules/styles.py index 7fb6c2e11..81d9800d1 100644 --- a/modules/styles.py +++ b/modules/styles.py @@ -98,10 +98,8 @@ class StyleDatabase: self.path = path folder, file = os.path.split(self.path) - self.default_file = file.split("*")[0] + ".csv" - if self.default_file == ".csv": - self.default_file = "styles.csv" - self.default_path = os.path.join(folder, self.default_file) + filename, _, ext = file.partition('*') + self.default_path = os.path.join(folder, filename + ext) self.prompt_fields = [field for field in PromptStyle._fields if field != "path"] @@ -155,10 +153,8 @@ class StyleDatabase: row["name"], prompt, negative_prompt, path ) - def get_style_paths(self) -> list(): - """ - Returns a list of all distinct paths, including the default path, of - files that styles are loaded from.""" + def get_style_paths(self) -> set: + """Returns a set of all distinct paths of files that styles are loaded from.""" # Update any styles without a path to the default path for style in list(self.styles.values()): if not style.path: @@ -172,9 +168,9 @@ class StyleDatabase: style_paths.add(style.path) # Remove any paths for styles that are just list dividers - style_paths.remove("do_not_save") + style_paths.discard("do_not_save") - return list(style_paths) + return style_paths def get_style_prompts(self, styles): return [self.styles.get(x, self.no_style).prompt for x in styles] @@ -196,20 +192,7 @@ class StyleDatabase: # The path argument is deprecated, but kept for backwards compatibility _ = path - # Update any styles without a path to the default path - for style in list(self.styles.values()): - if not style.path: - self.styles[style.name] = style._replace(path=self.default_path) - - # Create a list of all distinct paths, including the default path - style_paths = set() - style_paths.add(self.default_path) - for _, style in self.styles.items(): - if style.path: - style_paths.add(style.path) - - # Remove any paths for styles that are just list dividers - style_paths.remove("do_not_save") + style_paths = self.get_style_paths() csv_names = [os.path.split(path)[1].lower() for path in style_paths] From 5cb1ce470df8332872af3dfa1067b761062d4608 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Thu, 14 Dec 2023 09:48:36 +0300 Subject: [PATCH 258/342] Merge pull request #14266 from kaalibro/dev Re-add setting lost as part of e294e46 --- modules/shared_options.py | 1 + 1 file changed, 1 insertion(+) diff --git a/modules/shared_options.py b/modules/shared_options.py index e5de0d018..acb6e2d48 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -256,6 +256,7 @@ options_templates.update(options_section(('ui_prompt_editing', "Prompt editing", "keyedit_precision_extra": OptionInfo(0.05, "Precision for when editing the prompt with Ctrl+up/down", gr.Slider, {"minimum": 0.01, "maximum": 0.2, "step": 0.001}), "keyedit_delimiters": OptionInfo(r".,\/!?%^*;:{}=`~() ", "Word delimiters when editing the prompt with Ctrl+up/down"), "keyedit_delimiters_whitespace": OptionInfo(["Tab", "Carriage Return", "Line Feed"], "Ctrl+up/down whitespace delimiters", gr.CheckboxGroup, lambda: {"choices": ["Tab", "Carriage Return", "Line Feed"]}), + "keyedit_move": OptionInfo(True, "Alt+left/right moves prompt elements"), "disable_token_counters": OptionInfo(False, "Disable prompt token counters").needs_reload_ui(), })) From b7e0d4a7e171ee1cef73684b8423fe4a20ca7e34 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Thu, 14 Dec 2023 09:52:23 +0300 Subject: [PATCH 259/342] Merge pull request #14229 from Nuullll/ipex-embedding [IPEX] Fix embedding and ControlNet --- modules/xpu_specific.py | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/modules/xpu_specific.py b/modules/xpu_specific.py index d933c7903..d8da94a0e 100644 --- a/modules/xpu_specific.py +++ b/modules/xpu_specific.py @@ -48,3 +48,12 @@ if has_xpu: CondFunc('torch.nn.modules.conv.Conv2d.forward', lambda orig_func, self, input: orig_func(self, input.to(self.weight.data.dtype)), lambda orig_func, self, input: input.dtype != self.weight.data.dtype) + CondFunc('torch.bmm', + lambda orig_func, input, mat2, out=None: orig_func(input.to(mat2.dtype), mat2, out=out), + lambda orig_func, input, mat2, out=None: input.dtype != mat2.dtype) + CondFunc('torch.cat', + lambda orig_func, tensors, dim=0, out=None: orig_func([t.to(tensors[0].dtype) for t in tensors], dim=dim, out=out), + lambda orig_func, tensors, dim=0, out=None: not all(t.dtype == tensors[0].dtype for t in tensors)) + CondFunc('torch.nn.functional.scaled_dot_product_attention', + lambda orig_func, query, key, value, attn_mask=None, dropout_p=0.0, is_causal=False: orig_func(query, key.to(query.dtype), value.to(query.dtype), attn_mask, dropout_p, is_causal), + lambda orig_func, query, key, value, attn_mask=None, dropout_p=0.0, is_causal=False: query.dtype != key.dtype or query.dtype != value.dtype) From f8871dedcfe3a67689ef333aea2fdf05a9aaffa2 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Thu, 14 Dec 2023 09:59:48 +0300 Subject: [PATCH 260/342] Merge pull request #14230 from AUTOMATIC1111/add-option-Live-preview-in-full-page-image-viewer add option: Live preview in full page image viewer --- javascript/imageviewer.js | 2 +- modules/shared_options.py | 1 + 2 files changed, 2 insertions(+), 1 deletion(-) diff --git a/javascript/imageviewer.js b/javascript/imageviewer.js index e4dae91bc..625c5d148 100644 --- a/javascript/imageviewer.js +++ b/javascript/imageviewer.js @@ -34,7 +34,7 @@ function updateOnBackgroundChange() { if (modalImage && modalImage.offsetParent) { let currentButton = selected_gallery_button(); let preview = gradioApp().querySelectorAll('.livePreview > img'); - if (preview.length > 0) { + if (opts.js_live_preview_in_modal_lightbox && preview.length > 0) { // show preview image if available modalImage.src = preview[preview.length - 1].src; } else if (currentButton?.children?.length > 0 && modalImage.src != currentButton.children[0].src) { diff --git a/modules/shared_options.py b/modules/shared_options.py index acb6e2d48..41097d8e6 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -331,6 +331,7 @@ options_templates.update(options_section(('ui', "Live previews", "ui"), { "live_preview_content": OptionInfo("Prompt", "Live preview subject", gr.Radio, {"choices": ["Combined", "Prompt", "Negative prompt"]}), "live_preview_refresh_period": OptionInfo(1000, "Progressbar and preview update period").info("in milliseconds"), "live_preview_fast_interrupt": OptionInfo(False, "Return image with chosen live preview method on interrupt").info("makes interrupts faster"), + "js_live_preview_in_modal_lightbox": OptionInfo(True, "Show Live preview in full page image viewer"), })) options_templates.update(options_section(('sampler-params', "Sampler parameters", "sd"), { From eb52c803b849cdd1fc137db4568eca5bb8373f58 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Thu, 14 Dec 2023 10:03:14 +0300 Subject: [PATCH 261/342] Merge pull request #14216 from wfjsw/state-dict-ref-comparison change state dict comparison to ref compare --- modules/sd_disable_initialization.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/sd_disable_initialization.py b/modules/sd_disable_initialization.py index 8863107ae..273a7edd8 100644 --- a/modules/sd_disable_initialization.py +++ b/modules/sd_disable_initialization.py @@ -215,7 +215,7 @@ class LoadStateDictOnMeta(ReplaceHelper): would be on the meta device. """ - if state_dict == sd: + if state_dict is sd: state_dict = {k: v.to(device="meta", dtype=v.dtype) for k, v in state_dict.items()} original(module, state_dict, strict=strict) From 2be85f8fe03533bf3b1ad562ea58ee8227ba3b99 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Thu, 14 Dec 2023 10:08:03 +0300 Subject: [PATCH 262/342] Merge pull request #14237 from ReneKroon/dev #13354 : solve lora loading issue --- extensions-builtin/Lora/networks.py | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/extensions-builtin/Lora/networks.py b/extensions-builtin/Lora/networks.py index 7f814706a..629bf8537 100644 --- a/extensions-builtin/Lora/networks.py +++ b/extensions-builtin/Lora/networks.py @@ -159,7 +159,8 @@ def load_network(name, network_on_disk): bundle_embeddings = {} for key_network, weight in sd.items(): - key_network_without_network_parts, network_part = key_network.split(".", 1) + key_network_without_network_parts, _, network_part = key_network.partition(".") + if key_network_without_network_parts == "bundle_emb": emb_name, vec_name = network_part.split(".", 1) emb_dict = bundle_embeddings.get(emb_name, {}) From 3c0c27757944ae17a7fa4c2323ee9ae2d434dbce Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Thu, 14 Dec 2023 19:36:17 +0900 Subject: [PATCH 263/342] default False js_live_preview_in_modal_lightbox --- modules/shared_options.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/shared_options.py b/modules/shared_options.py index 41097d8e6..d2e86ff10 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -331,7 +331,7 @@ options_templates.update(options_section(('ui', "Live previews", "ui"), { "live_preview_content": OptionInfo("Prompt", "Live preview subject", gr.Radio, {"choices": ["Combined", "Prompt", "Negative prompt"]}), "live_preview_refresh_period": OptionInfo(1000, "Progressbar and preview update period").info("in milliseconds"), "live_preview_fast_interrupt": OptionInfo(False, "Return image with chosen live preview method on interrupt").info("makes interrupts faster"), - "js_live_preview_in_modal_lightbox": OptionInfo(True, "Show Live preview in full page image viewer"), + "js_live_preview_in_modal_lightbox": OptionInfo(False, "Show Live preview in full page image viewer"), })) options_templates.update(options_section(('sampler-params', "Sampler parameters", "sd"), { From 0c5427960b3a4ffe6d673c28e8e135b26f015717 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Fri, 15 Dec 2023 17:11:59 +0900 Subject: [PATCH 264/342] make modal toolbar and icon opacity adjustable --- modules/shared_gradio_themes.py | 4 ++++ modules/shared_options.py | 2 ++ style.css | 4 ++-- 3 files changed, 8 insertions(+), 2 deletions(-) diff --git a/modules/shared_gradio_themes.py b/modules/shared_gradio_themes.py index 822db0a95..b6dc31450 100644 --- a/modules/shared_gradio_themes.py +++ b/modules/shared_gradio_themes.py @@ -65,3 +65,7 @@ def reload_gradio_theme(theme_name=None): except Exception as e: errors.display(e, "changing gradio theme") shared.gradio_theme = gr.themes.Default(**default_theme_args) + + # append additional values gradio_theme + shared.gradio_theme.sd_webui_modal_lightbox_toolbar_opacity = shared.opts.sd_webui_modal_lightbox_toolbar_opacity + shared.gradio_theme.sd_webui_modal_lightbox_icon_opacity = shared.opts.sd_webui_modal_lightbox_icon_opacity diff --git a/modules/shared_options.py b/modules/shared_options.py index e5de0d018..86e7636cc 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -266,6 +266,8 @@ options_templates.update(options_section(('ui_gallery', "Gallery", "ui"), { "js_modal_lightbox_initially_zoomed": OptionInfo(True, "Full page image viewer: show images zoomed in by default"), "js_modal_lightbox_gamepad": OptionInfo(False, "Full page image viewer: navigate with gamepad"), "js_modal_lightbox_gamepad_repeat": OptionInfo(250, "Full page image viewer: gamepad repeat period").info("in milliseconds"), + "sd_webui_modal_lightbox_icon_opacity": OptionInfo(1, "Full page image viewer: control icon unfocused opacity", gr.Slider, {"minimum": 0.0, "maximum": 1, "step": 0.01}, onchange=shared.reload_gradio_theme).info('for mouse only').needs_reload_ui(), + "sd_webui_modal_lightbox_toolbar_opacity": OptionInfo(0.9, "Full page image viewer: tool bar opacity", gr.Slider, {"minimum": 0.0, "maximum": 1, "step": 0.01}, onchange=shared.reload_gradio_theme).info('for mouse only').needs_reload_ui(), "gallery_height": OptionInfo("", "Gallery height", gr.Textbox).info("can be any valid CSS value, for example 768px or 20em").needs_reload_ui(), })) diff --git a/style.css b/style.css index ec449bdea..6d4c8a0d5 100644 --- a/style.css +++ b/style.css @@ -679,7 +679,7 @@ table.popup-table .link{ transition: 0.2s ease background-color; } .modalControls:hover { - background-color:rgba(0,0,0,0.9); + background-color:rgba(0,0,0, var(--sd-webui-modal-lightbox-toolbar-opacity)); } .modalClose { margin-left: auto; @@ -761,7 +761,7 @@ table.popup-table .link{ .modalPrev, .modalNext, .modalControls .cursor { - opacity: 0; + opacity: var(--sd-webui-modal-lightbox-icon-opacity); } } From 1242ba08e19f3d317bdc5924db2b73d0c9569a7f Mon Sep 17 00:00:00 2001 From: gayshub Date: Fri, 15 Dec 2023 16:57:17 +0800 Subject: [PATCH 265/342] add allow specify the task id and get the location of task in the queue of pending task --- modules/api/api.py | 20 ++++++++++++++++++-- modules/api/models.py | 2 ++ modules/processing.py | 2 ++ modules/progress.py | 21 +++++++++++++++++++-- 4 files changed, 41 insertions(+), 4 deletions(-) diff --git a/modules/api/api.py b/modules/api/api.py index e6edffe71..5d000ae8b 100644 --- a/modules/api/api.py +++ b/modules/api/api.py @@ -33,7 +33,7 @@ from typing import Dict, List, Any import piexif import piexif.helper from contextlib import closing - +from modules.progress import create_task_id, add_task_to_queue, start_task, finish_task, current_task def script_name_to_index(name, scripts): try: @@ -337,6 +337,10 @@ class Api: return script_args def text2imgapi(self, txt2imgreq: models.StableDiffusionTxt2ImgProcessingAPI): + task_id = create_task_id("text2img") + if txt2imgreq.force_task_id != None: + task_id = txt2imgreq.force_task_id + script_runner = scripts.scripts_txt2img if not script_runner.scripts: script_runner.initialize_scripts(False) @@ -363,6 +367,8 @@ class Api: send_images = args.pop('send_images', True) args.pop('save_images', None) + add_task_to_queue(task_id) + with self.queue_lock: with closing(StableDiffusionProcessingTxt2Img(sd_model=shared.sd_model, **args)) as p: p.is_api = True @@ -372,12 +378,14 @@ class Api: try: shared.state.begin(job="scripts_txt2img") + start_task(task_id) if selectable_scripts is not None: p.script_args = script_args processed = scripts.scripts_txt2img.run(p, *p.script_args) # Need to pass args as list here else: p.script_args = tuple(script_args) # Need to pass args as tuple here processed = process_images(p) + finish_task(task_id) finally: shared.state.end() shared.total_tqdm.clear() @@ -387,6 +395,10 @@ class Api: return models.TextToImageResponse(images=b64images, parameters=vars(txt2imgreq), info=processed.js()) def img2imgapi(self, img2imgreq: models.StableDiffusionImg2ImgProcessingAPI): + task_id = create_task_id("img2img") + if img2imgreq.force_task_id != None: + task_id = img2imgreq.force_task_id + init_images = img2imgreq.init_images if init_images is None: raise HTTPException(status_code=404, detail="Init image not found") @@ -423,6 +435,8 @@ class Api: send_images = args.pop('send_images', True) args.pop('save_images', None) + add_task_to_queue(task_id) + with self.queue_lock: with closing(StableDiffusionProcessingImg2Img(sd_model=shared.sd_model, **args)) as p: p.init_images = [decode_base64_to_image(x) for x in init_images] @@ -433,12 +447,14 @@ class Api: try: shared.state.begin(job="scripts_img2img") + start_task(task_id) if selectable_scripts is not None: p.script_args = script_args processed = scripts.scripts_img2img.run(p, *p.script_args) # Need to pass args as list here else: p.script_args = tuple(script_args) # Need to pass args as tuple here processed = process_images(p) + finish_task(task_id) finally: shared.state.end() shared.total_tqdm.clear() @@ -514,7 +530,7 @@ class Api: if shared.state.current_image and not req.skip_current_image: current_image = encode_pil_to_base64(shared.state.current_image) - return models.ProgressResponse(progress=progress, eta_relative=eta_relative, state=shared.state.dict(), current_image=current_image, textinfo=shared.state.textinfo) + return models.ProgressResponse(progress=progress, eta_relative=eta_relative, state=shared.state.dict(), current_image=current_image, textinfo=shared.state.textinfo, current_task=current_task) def interrogateapi(self, interrogatereq: models.InterrogateRequest): image_b64 = interrogatereq.image diff --git a/modules/api/models.py b/modules/api/models.py index 6a574771c..7b7f17738 100644 --- a/modules/api/models.py +++ b/modules/api/models.py @@ -109,6 +109,7 @@ StableDiffusionTxt2ImgProcessingAPI = PydanticModelGenerator( {"key": "send_images", "type": bool, "default": True}, {"key": "save_images", "type": bool, "default": False}, {"key": "alwayson_scripts", "type": dict, "default": {}}, + {"key": "force_task_id", "type": str, "default": None}, ] ).generate_model() @@ -126,6 +127,7 @@ StableDiffusionImg2ImgProcessingAPI = PydanticModelGenerator( {"key": "send_images", "type": bool, "default": True}, {"key": "save_images", "type": bool, "default": False}, {"key": "alwayson_scripts", "type": dict, "default": {}}, + {"key": "force_task_id", "type": str, "default": None}, ] ).generate_model() diff --git a/modules/processing.py b/modules/processing.py index e124e7f0d..657cacfca 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -1023,6 +1023,7 @@ class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing): hr_sampler_name: str = None hr_prompt: str = '' hr_negative_prompt: str = '' + force_task_id: str = None cached_hr_uc = [None, None] cached_hr_c = [None, None] @@ -1358,6 +1359,7 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): inpainting_mask_invert: int = 0 initial_noise_multiplier: float = None latent_mask: Image = None + force_task_id: string = None image_mask: Any = field(default=None, init=False) diff --git a/modules/progress.py b/modules/progress.py index 69921de72..553866db1 100644 --- a/modules/progress.py +++ b/modules/progress.py @@ -8,10 +8,13 @@ from pydantic import BaseModel, Field from modules.shared import opts import modules.shared as shared - +from collections import OrderedDict +import string +import random +from typing import List current_task = None -pending_tasks = {} +pending_tasks = OrderedDict() finished_tasks = [] recorded_results = [] recorded_results_limit = 2 @@ -34,6 +37,11 @@ def finish_task(id_task): if len(finished_tasks) > 16: finished_tasks.pop(0) +def create_task_id(task_type): + N = 7 + res = ''.join(random.choices(string.ascii_uppercase + + string.digits, k=N)) + return f"task({task_type}-{res})" def record_results(id_task, res): recorded_results.append((id_task, res)) @@ -44,6 +52,9 @@ def record_results(id_task, res): def add_task_to_queue(id_job): pending_tasks[id_job] = time.time() +class PendingTasksResponse(BaseModel): + size: int = Field(title="Pending task size") + tasks: List[str] = Field(title="Pending task ids") class ProgressRequest(BaseModel): id_task: str = Field(default=None, title="Task ID", description="id of the task to get progress for") @@ -63,8 +74,14 @@ class ProgressResponse(BaseModel): def setup_progress_api(app): + app.add_api_route("/internal/pendingTasks", get_pending_tasks, methods=["GET"]) return app.add_api_route("/internal/progress", progressapi, methods=["POST"], response_model=ProgressResponse) +def get_pending_tasks(): + pending_tasks_ids = [x for x in pending_tasks] + pending_len = len(pending_tasks_ids) + return PendingTasksResponse(size=pending_len, tasks=pending_tasks_ids) + def progressapi(req: ProgressRequest): active = req.id_task == current_task From d859de37d9ec10cb6c804226328a11c87c444852 Mon Sep 17 00:00:00 2001 From: gayshub Date: Fri, 15 Dec 2023 17:48:20 +0800 Subject: [PATCH 266/342] fix the problem of ruff of github --- modules/api/api.py | 4 ++-- modules/processing.py | 2 +- modules/progress.py | 2 +- 3 files changed, 4 insertions(+), 4 deletions(-) diff --git a/modules/api/api.py b/modules/api/api.py index 5d000ae8b..1f4648067 100644 --- a/modules/api/api.py +++ b/modules/api/api.py @@ -338,7 +338,7 @@ class Api: def text2imgapi(self, txt2imgreq: models.StableDiffusionTxt2ImgProcessingAPI): task_id = create_task_id("text2img") - if txt2imgreq.force_task_id != None: + if txt2imgreq.force_task_id is None: task_id = txt2imgreq.force_task_id script_runner = scripts.scripts_txt2img @@ -396,7 +396,7 @@ class Api: def img2imgapi(self, img2imgreq: models.StableDiffusionImg2ImgProcessingAPI): task_id = create_task_id("img2img") - if img2imgreq.force_task_id != None: + if img2imgreq.force_task_id is None: task_id = img2imgreq.force_task_id init_images = img2imgreq.init_images diff --git a/modules/processing.py b/modules/processing.py index 657cacfca..5added65e 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -1359,7 +1359,7 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): inpainting_mask_invert: int = 0 initial_noise_multiplier: float = None latent_mask: Image = None - force_task_id: string = None + force_task_id: str = None image_mask: Any = field(default=None, init=False) diff --git a/modules/progress.py b/modules/progress.py index 553866db1..6946fb1bc 100644 --- a/modules/progress.py +++ b/modules/progress.py @@ -78,7 +78,7 @@ def setup_progress_api(app): return app.add_api_route("/internal/progress", progressapi, methods=["POST"], response_model=ProgressResponse) def get_pending_tasks(): - pending_tasks_ids = [x for x in pending_tasks] + pending_tasks_ids = list(pending_tasks) pending_len = len(pending_tasks_ids) return PendingTasksResponse(size=pending_len, tasks=pending_tasks_ids) From da45e73b4ffde2e2a85b64a3e3258a0625bd307e Mon Sep 17 00:00:00 2001 From: gayshub Date: Fri, 15 Dec 2023 17:57:58 +0800 Subject: [PATCH 267/342] fix the problem of ruff of github --- modules/api/api.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/api/api.py b/modules/api/api.py index 1f4648067..9fac7e604 100644 --- a/modules/api/api.py +++ b/modules/api/api.py @@ -340,7 +340,7 @@ class Api: task_id = create_task_id("text2img") if txt2imgreq.force_task_id is None: task_id = txt2imgreq.force_task_id - + script_runner = scripts.scripts_txt2img if not script_runner.scripts: script_runner.initialize_scripts(False) From 6d7e57ba6a4d686d515518b5f90e91b32fa01caf Mon Sep 17 00:00:00 2001 From: gayshub Date: Fri, 15 Dec 2023 18:03:14 +0800 Subject: [PATCH 268/342] fix the problem of ruff of github --- modules/api/api.py | 1 - 1 file changed, 1 deletion(-) diff --git a/modules/api/api.py b/modules/api/api.py index 9fac7e604..8d8e70a46 100644 --- a/modules/api/api.py +++ b/modules/api/api.py @@ -340,7 +340,6 @@ class Api: task_id = create_task_id("text2img") if txt2imgreq.force_task_id is None: task_id = txt2imgreq.force_task_id - script_runner = scripts.scripts_txt2img if not script_runner.scripts: script_runner.initialize_scripts(False) From 0dfffe53ec11b2ee097d55efc479f8e707015db9 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sat, 16 Dec 2023 09:25:08 +0300 Subject: [PATCH 269/342] Merge pull request #14307 from AUTOMATIC1111/default-Falst-js_live_preview_in_modal_lightbox default False js_live_preview_in_modal_lightbox --- modules/shared_options.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/shared_options.py b/modules/shared_options.py index 41097d8e6..d2e86ff10 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -331,7 +331,7 @@ options_templates.update(options_section(('ui', "Live previews", "ui"), { "live_preview_content": OptionInfo("Prompt", "Live preview subject", gr.Radio, {"choices": ["Combined", "Prompt", "Negative prompt"]}), "live_preview_refresh_period": OptionInfo(1000, "Progressbar and preview update period").info("in milliseconds"), "live_preview_fast_interrupt": OptionInfo(False, "Return image with chosen live preview method on interrupt").info("makes interrupts faster"), - "js_live_preview_in_modal_lightbox": OptionInfo(True, "Show Live preview in full page image viewer"), + "js_live_preview_in_modal_lightbox": OptionInfo(False, "Show Live preview in full page image viewer"), })) options_templates.update(options_section(('sampler-params', "Sampler parameters", "sd"), { From ea272152e0b50dbb2bd675ec020607f3d50c37d0 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Sat, 16 Dec 2023 15:08:08 +0800 Subject: [PATCH 270/342] Add FP8 settings into PNG info --- modules/generation_parameters_copypaste.py | 6 ++++++ modules/processing.py | 2 ++ 2 files changed, 8 insertions(+) diff --git a/modules/generation_parameters_copypaste.py b/modules/generation_parameters_copypaste.py index 4efe53e0c..dbffe4949 100644 --- a/modules/generation_parameters_copypaste.py +++ b/modules/generation_parameters_copypaste.py @@ -314,6 +314,12 @@ Steps: 20, Sampler: Euler a, CFG scale: 7, Seed: 965400086, Size: 512x512, Model if "VAE Decoder" not in res: res["VAE Decoder"] = "Full" + if "FP8 weight" not in res: + res["FP8 weight"] = "Disable" + + if "Cache FP16 weight for LoRA" not in res and res["FP8 weight"] != "Disable": + res["Cache FP16 weight for LoRA"] = False + skip = set(shared.opts.infotext_skip_pasting) res = {k: v for k, v in res.items() if k not in skip} diff --git a/modules/processing.py b/modules/processing.py index bea01ec68..179f2c0fc 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -688,6 +688,8 @@ def create_infotext(p, all_prompts, all_seeds, all_subseeds, comments=None, iter "Size": f"{p.width}x{p.height}", "Model hash": p.sd_model_hash if opts.add_model_hash_to_info else None, "Model": p.sd_model_name if opts.add_model_name_to_info else None, + "FP8 weight": opts.fp8_storage if devices.fp8 else None, + "Cache FP16 weight for LoRA": opts.cache_fp16_weight if devices.fp8 else None, "VAE hash": p.sd_vae_hash if opts.add_vae_hash_to_info else None, "VAE": p.sd_vae_name if opts.add_vae_name_to_info else None, "Variation seed": (None if p.subseed_strength == 0 else (p.all_subseeds[0] if use_main_prompt else all_subseeds[index])), From 7745db6fc02faf19117838c1e7bcc8a60b5f5e90 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sat, 16 Dec 2023 10:15:08 +0300 Subject: [PATCH 271/342] torch 2.1.2 --- modules/errors.py | 4 ++-- modules/launch_utils.py | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/modules/errors.py b/modules/errors.py index c534a5d66..48aa13a17 100644 --- a/modules/errors.py +++ b/modules/errors.py @@ -107,8 +107,8 @@ def check_versions(): import torch import gradio - expected_torch_version = "2.1.0" - expected_xformers_version = "0.0.22.post7" + expected_torch_version = "2.1.2" + expected_xformers_version = "0.0.23.post1" expected_gradio_version = "3.41.2" if version.parse(torch.__version__) < version.parse(expected_torch_version): diff --git a/modules/launch_utils.py b/modules/launch_utils.py index 2c54e2a02..dabef0f53 100644 --- a/modules/launch_utils.py +++ b/modules/launch_utils.py @@ -315,7 +315,7 @@ def requirements_met(requirements_file): def prepare_environment(): torch_index_url = os.environ.get('TORCH_INDEX_URL', "https://download.pytorch.org/whl/cu121") - torch_command = os.environ.get('TORCH_COMMAND', f"pip install torch==2.1.0 torchvision==0.16.0 --extra-index-url {torch_index_url}") + torch_command = os.environ.get('TORCH_COMMAND', f"pip install torch==2.1.2 torchvision==0.16.2 --extra-index-url {torch_index_url}") if args.use_ipex: if platform.system() == "Windows": # The "Nuullll/intel-extension-for-pytorch" wheels were built from IPEX source for Intel Arc GPU: https://github.com/intel/intel-extension-for-pytorch/tree/xpu-main @@ -338,7 +338,7 @@ def prepare_environment(): torch_command = os.environ.get('TORCH_COMMAND', f"pip install torch==2.0.0a0 intel-extension-for-pytorch==2.0.110+gitba7f6c1 --extra-index-url {torch_index_url}") requirements_file = os.environ.get('REQS_FILE', "requirements_versions.txt") - xformers_package = os.environ.get('XFORMERS_PACKAGE', 'xformers==0.0.22.post7') + xformers_package = os.environ.get('XFORMERS_PACKAGE', 'xformers==0.0.23.post1') clip_package = os.environ.get('CLIP_PACKAGE', "https://github.com/openai/CLIP/archive/d50d76daa670286dd6cacf3bcd80b5e4823fc8e1.zip") openclip_package = os.environ.get('OPENCLIP_PACKAGE', "https://github.com/mlfoundations/open_clip/archive/bb6e834e9c70d9c27d0dc3ecedeebeaeb1ffad6b.zip") From cd9ce2e31c4a264d7cde17c54d24f8ad94c9cf2c Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sat, 16 Dec 2023 10:40:20 +0300 Subject: [PATCH 272/342] Use radio for FP8 mode selection --- modules/shared_options.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/shared_options.py b/modules/shared_options.py index d470eb8ff..fa542ba8c 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -206,7 +206,7 @@ options_templates.update(options_section(('optimizations', "Optimizations", "sd" "pad_cond_uncond": OptionInfo(False, "Pad prompt/negative prompt to be same length", infotext='Pad conds').info("improves performance when prompt and negative prompt have different lengths; changes seeds"), "persistent_cond_cache": OptionInfo(True, "Persistent cond cache").info("do not recalculate conds from prompts if prompts have not changed since previous calculation"), "batch_cond_uncond": OptionInfo(True, "Batch cond/uncond").info("do both conditional and unconditional denoising in one batch; uses a bit more VRAM during sampling, but improves speed; previously this was controlled by --always-batch-cond-uncond comandline argument"), - "fp8_storage": OptionInfo("Disable", "FP8 weight", gr.Dropdown, {"choices": ["Disable", "Enable for SDXL", "Enable"]}).info("Use FP8 to store Linear/Conv layers' weight. Require pytorch>=2.1.0."), + "fp8_storage": OptionInfo("Disable", "FP8 weight", gr.Radio, {"choices": ["Disable", "Enable for SDXL", "Enable"]}).info("Use FP8 to store Linear/Conv layers' weight. Require pytorch>=2.1.0."), "cache_fp16_weight": OptionInfo(False, "Cache FP16 weight for LoRA").info("Cache fp16 weight when enabling FP8, will increase the quality of LoRA. Use more system ram."), })) From 93eae69895c34361a71dbed17348bcfd132fbc6a Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sat, 16 Dec 2023 11:00:42 +0300 Subject: [PATCH 273/342] move soft inpainting to a built-in extension --- .../soft-inpainting/scripts}/soft_inpainting.py | 0 1 file changed, 0 insertions(+), 0 deletions(-) rename {scripts => extensions-builtin/soft-inpainting/scripts}/soft_inpainting.py (100%) diff --git a/scripts/soft_inpainting.py b/extensions-builtin/soft-inpainting/scripts/soft_inpainting.py similarity index 100% rename from scripts/soft_inpainting.py rename to extensions-builtin/soft-inpainting/scripts/soft_inpainting.py From 86b3aa94e2d36a4f9d5ef1bb7c6ec995ff8eb517 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sat, 16 Dec 2023 11:04:59 +0300 Subject: [PATCH 274/342] rename pending tasks api endpoint to be more in line with others --- modules/progress.py | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/modules/progress.py b/modules/progress.py index 6946fb1bc..85255e821 100644 --- a/modules/progress.py +++ b/modules/progress.py @@ -74,9 +74,10 @@ class ProgressResponse(BaseModel): def setup_progress_api(app): - app.add_api_route("/internal/pendingTasks", get_pending_tasks, methods=["GET"]) + app.add_api_route("/internal/pending-tasks", get_pending_tasks, methods=["GET"]) return app.add_api_route("/internal/progress", progressapi, methods=["POST"], response_model=ProgressResponse) + def get_pending_tasks(): pending_tasks_ids = list(pending_tasks) pending_len = len(pending_tasks_ids) From a97832033427096072d5ea914adac3662cda4fd1 Mon Sep 17 00:00:00 2001 From: Kohaku-Blueleaf <59680068+KohakuBlueleaf@users.noreply.github.com> Date: Sat, 16 Dec 2023 19:39:43 +0800 Subject: [PATCH 275/342] Let fp8-related settings to invalidate cond_cache --- modules/processing.py | 2 ++ 1 file changed, 2 insertions(+) diff --git a/modules/processing.py b/modules/processing.py index dd97b4eee..9351e3fb0 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -431,6 +431,8 @@ class StableDiffusionProcessing: opts.sdxl_crop_top, self.width, self.height, + opts.fp8_storage, + opts.cache_fp16_weight, ) def get_conds_with_caching(self, function, required_prompts, steps, caches, extra_network_data, hires_steps=None): From 98c5fa92015837706adfd9975d5f345ab74f1c99 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Sat, 16 Dec 2023 22:14:39 +0900 Subject: [PATCH 276/342] fix extras caption BLIP #14328 --- scripts/postprocessing_caption.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/scripts/postprocessing_caption.py b/scripts/postprocessing_caption.py index 243e3ad9c..9482a03ca 100644 --- a/scripts/postprocessing_caption.py +++ b/scripts/postprocessing_caption.py @@ -25,6 +25,6 @@ class ScriptPostprocessingCeption(scripts_postprocessing.ScriptPostprocessing): captions.append(deepbooru.model.tag(pp.image)) if "BLIP" in option: - captions.append(shared.interrogator.generate_caption(pp.image)) + captions.append(shared.interrogator.interrogate(pp.image.convert("RGB"))) pp.caption = ", ".join([x for x in captions if x]) From de03882d6ca56bc81058f5120f028678a6a54aaa Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sun, 17 Dec 2023 08:55:35 +0300 Subject: [PATCH 277/342] make task ids for API work without force_task_id --- modules/api/api.py | 9 +++------ 1 file changed, 3 insertions(+), 6 deletions(-) diff --git a/modules/api/api.py b/modules/api/api.py index 9637cb81b..7154c9d5b 100644 --- a/modules/api/api.py +++ b/modules/api/api.py @@ -336,9 +336,8 @@ class Api: return script_args def text2imgapi(self, txt2imgreq: models.StableDiffusionTxt2ImgProcessingAPI): - task_id = create_task_id("text2img") - if txt2imgreq.force_task_id is None: - task_id = txt2imgreq.force_task_id + task_id = txt2imgreq.force_task_id or create_task_id("txt2img") + script_runner = scripts.scripts_txt2img if not script_runner.scripts: script_runner.initialize_scripts(False) @@ -393,9 +392,7 @@ class Api: return models.TextToImageResponse(images=b64images, parameters=vars(txt2imgreq), info=processed.js()) def img2imgapi(self, img2imgreq: models.StableDiffusionImg2ImgProcessingAPI): - task_id = create_task_id("img2img") - if img2imgreq.force_task_id is None: - task_id = img2imgreq.force_task_id + task_id = img2imgreq.force_task_id or create_task_id("img2img") init_images = img2imgreq.init_images if init_images is None: From 10945aa41a158ee03727c5ea77d4ffff6b5370f0 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Mon, 18 Dec 2023 15:27:41 +0900 Subject: [PATCH 278/342] only rewrite ui-config when there is change and a typo --- modules/ui.py | 4 +++- modules/ui_loadsave.py | 2 +- 2 files changed, 4 insertions(+), 2 deletions(-) diff --git a/modules/ui.py b/modules/ui.py index d80486dd4..f02b55116 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -1086,6 +1086,7 @@ def create_ui(): ) loadsave = ui_loadsave.UiLoadsave(cmd_opts.ui_config_file) + ui_settings_from_file = loadsave.ui_settings.copy() settings = ui_settings.UiSettings() settings.create_ui(loadsave, dummy_component) @@ -1146,7 +1147,8 @@ def create_ui(): modelmerger_ui.setup_ui(dummy_component=dummy_component, sd_model_checkpoint_component=settings.component_dict['sd_model_checkpoint']) - loadsave.dump_defaults() + if ui_settings_from_file != loadsave.ui_settings: + loadsave.dump_defaults() demo.ui_loadsave = loadsave return demo diff --git a/modules/ui_loadsave.py b/modules/ui_loadsave.py index 7826786cc..693ff75c5 100644 --- a/modules/ui_loadsave.py +++ b/modules/ui_loadsave.py @@ -144,7 +144,7 @@ class UiLoadsave: json.dump(current_ui_settings, file, indent=4, ensure_ascii=False) def dump_defaults(self): - """saves default values to a file unless tjhe file is present and there was an error loading default values at start""" + """saves default values to a file unless the file is present and there was an error loading default values at start""" if self.error_loading and os.path.exists(self.filename): return From e4b4a9c4acf0ca375a8603f7f52fde8467b2d266 Mon Sep 17 00:00:00 2001 From: Nuullll Date: Mon, 18 Dec 2023 18:00:01 +0800 Subject: [PATCH 279/342] [IPEX] Slice SDPA into smaller chunks --- modules/xpu_specific.py | 66 +++++++++++++++++++++++++++++++++++++++-- 1 file changed, 64 insertions(+), 2 deletions(-) diff --git a/modules/xpu_specific.py b/modules/xpu_specific.py index d8da94a0e..0ebdd5964 100644 --- a/modules/xpu_specific.py +++ b/modules/xpu_specific.py @@ -27,6 +27,68 @@ def torch_xpu_gc(): has_xpu = check_for_xpu() + +# Arc GPU cannot allocate a single block larger than 4GB: https://github.com/intel/compute-runtime/issues/627 +# Here we implement a slicing algorithm to split large batch size into smaller chunks, +# so that SDPA of each chunk wouldn't require any allocation larger than ARC_SINGLE_ALLOCATION_LIMIT. +# The heuristic limit (TOTAL_VRAM // 8) is tuned for Intel Arc A770 16G and Arc A750 8G, +# which is the best trade-off between VRAM usage and performance. +ARC_SINGLE_ALLOCATION_LIMIT = min(torch.xpu.get_device_properties(shared.cmd_opts.device_id).total_memory // 8, 4 * 1024 * 1024 * 1024) +orig_sdp_attn_func = torch.nn.functional.scaled_dot_product_attention +def torch_xpu_scaled_dot_product_attention( + query, key, value, attn_mask=None, dropout_p=0.0, is_causal=False, *args, **kwargs +): + # cast to same dtype first + key = key.to(query.dtype) + value = value.to(query.dtype) + + N = query.shape[:-2] # Batch size + L = query.size(-2) # Target sequence length + E = query.size(-1) # Embedding dimension of the query and key + S = key.size(-2) # Source sequence length + Ev = value.size(-1) # Embedding dimension of the value + + total_batch_size = torch.numel(torch.empty(N)) + batch_size_limit = max(1, ARC_SINGLE_ALLOCATION_LIMIT // (L * S * query.element_size())) + + if total_batch_size <= batch_size_limit: + return orig_sdp_attn_func( + query, + key, + value, + attn_mask, + dropout_p, + is_causal, + *args, **kwargs + ) + + query = torch.reshape(query, (-1, L, E)) + key = torch.reshape(key, (-1, S, E)) + value = torch.reshape(value, (-1, S, Ev)) + if attn_mask is not None: + attn_mask = attn_mask.view(-1, L, S) + chunk_count = (total_batch_size + batch_size_limit - 1) // batch_size_limit + outputs = [] + for i in range(chunk_count): + attn_mask_chunk = ( + None + if attn_mask is None + else attn_mask[i * batch_size_limit : (i + 1) * batch_size_limit, :, :] + ) + chunk_output = orig_sdp_attn_func( + query[i * batch_size_limit : (i + 1) * batch_size_limit, :, :], + key[i * batch_size_limit : (i + 1) * batch_size_limit, :, :], + value[i * batch_size_limit : (i + 1) * batch_size_limit, :, :], + attn_mask_chunk, + dropout_p, + is_causal, + *args, **kwargs + ) + outputs.append(chunk_output) + result = torch.cat(outputs, dim=0) + return torch.reshape(result, (*N, L, Ev)) + + if has_xpu: # W/A for https://github.com/intel/intel-extension-for-pytorch/issues/452: torch.Generator API doesn't support XPU device CondFunc('torch.Generator', @@ -55,5 +117,5 @@ if has_xpu: lambda orig_func, tensors, dim=0, out=None: orig_func([t.to(tensors[0].dtype) for t in tensors], dim=dim, out=out), lambda orig_func, tensors, dim=0, out=None: not all(t.dtype == tensors[0].dtype for t in tensors)) CondFunc('torch.nn.functional.scaled_dot_product_attention', - lambda orig_func, query, key, value, attn_mask=None, dropout_p=0.0, is_causal=False: orig_func(query, key.to(query.dtype), value.to(query.dtype), attn_mask, dropout_p, is_causal), - lambda orig_func, query, key, value, attn_mask=None, dropout_p=0.0, is_causal=False: query.dtype != key.dtype or query.dtype != value.dtype) + lambda orig_func, *args, **kwargs: torch_xpu_scaled_dot_product_attention(*args, **kwargs), + lambda orig_func, query, *args, **kwargs: query.is_xpu) From f586f4973a0f715e30b42242bb0e6b3f88c37d90 Mon Sep 17 00:00:00 2001 From: Nuullll Date: Mon, 18 Dec 2023 19:44:52 +0800 Subject: [PATCH 280/342] Fix device id --- modules/xpu_specific.py | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/modules/xpu_specific.py b/modules/xpu_specific.py index 0ebdd5964..f7687a66c 100644 --- a/modules/xpu_specific.py +++ b/modules/xpu_specific.py @@ -33,7 +33,7 @@ has_xpu = check_for_xpu() # so that SDPA of each chunk wouldn't require any allocation larger than ARC_SINGLE_ALLOCATION_LIMIT. # The heuristic limit (TOTAL_VRAM // 8) is tuned for Intel Arc A770 16G and Arc A750 8G, # which is the best trade-off between VRAM usage and performance. -ARC_SINGLE_ALLOCATION_LIMIT = min(torch.xpu.get_device_properties(shared.cmd_opts.device_id).total_memory // 8, 4 * 1024 * 1024 * 1024) +ARC_SINGLE_ALLOCATION_LIMIT = {} orig_sdp_attn_func = torch.nn.functional.scaled_dot_product_attention def torch_xpu_scaled_dot_product_attention( query, key, value, attn_mask=None, dropout_p=0.0, is_causal=False, *args, **kwargs @@ -49,7 +49,10 @@ def torch_xpu_scaled_dot_product_attention( Ev = value.size(-1) # Embedding dimension of the value total_batch_size = torch.numel(torch.empty(N)) - batch_size_limit = max(1, ARC_SINGLE_ALLOCATION_LIMIT // (L * S * query.element_size())) + device_id = query.device.index + if device_id not in ARC_SINGLE_ALLOCATION_LIMIT: + ARC_SINGLE_ALLOCATION_LIMIT[device_id] = min(torch.xpu.get_device_properties(device_id).total_memory // 8, 4 * 1024 * 1024 * 1024) + batch_size_limit = max(1, ARC_SINGLE_ALLOCATION_LIMIT[device_id] // (L * S * query.element_size())) if total_batch_size <= batch_size_limit: return orig_sdp_attn_func( From fe4d084390d35de49baf83c3319a72d71f540aee Mon Sep 17 00:00:00 2001 From: Muhammad Rehan Aslam <19831661+ranareehanaslam@users.noreply.github.com> Date: Mon, 18 Dec 2023 17:50:00 +0500 Subject: [PATCH 281/342] Update webui.py Added (Fixed) IPV6 Functionality When there is No Webui Argument Passed --- webui.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/webui.py b/webui.py index 9ed20b306..3b587dc41 100644 --- a/webui.py +++ b/webui.py @@ -39,7 +39,7 @@ def api_only(): print(f"Startup time: {startup_timer.summary()}.") api.launch( - server_name="0.0.0.0" if cmd_opts.listen else "127.0.0.1", + server_name = cmd_opts.server_name if cmd_opts.server_name else ("0.0.0.0" if cmd_opts.listen else "127.0.0.1"), port=cmd_opts.port if cmd_opts.port else 7861, root_path=f"/{cmd_opts.subpath}" if cmd_opts.subpath else "" ) From 0d5941edbc4c602b760d4200bd76e044c65a0e40 Mon Sep 17 00:00:00 2001 From: Muhammad Rehan Aslam <19831661+ranareehanaslam@users.noreply.github.com> Date: Tue, 19 Dec 2023 09:50:38 +0500 Subject: [PATCH 282/342] Update webui.py Co-authored-by: Aarni Koskela --- webui.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/webui.py b/webui.py index 3b587dc41..3f5b67a30 100644 --- a/webui.py +++ b/webui.py @@ -39,7 +39,7 @@ def api_only(): print(f"Startup time: {startup_timer.summary()}.") api.launch( - server_name = cmd_opts.server_name if cmd_opts.server_name else ("0.0.0.0" if cmd_opts.listen else "127.0.0.1"), + server_name=cmd_opts.server_name or ("0.0.0.0" if cmd_opts.listen else "127.0.0.1"), port=cmd_opts.port if cmd_opts.port else 7861, root_path=f"/{cmd_opts.subpath}" if cmd_opts.subpath else "" ) From 3e068de0dcec811d515402fc184f70709a785e4f Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Tue, 19 Dec 2023 18:48:49 +0900 Subject: [PATCH 283/342] reorder training preprocessing modules in extras tab using the order from before the rework 11d23e8ca55c097ecfa255a05b63f194e25f08be --- scripts/postprocessing_caption.py | 2 +- scripts/postprocessing_create_flipped_copies.py | 2 +- scripts/postprocessing_focal_crop.py | 2 +- scripts/processing_autosized_crop.py | 2 +- 4 files changed, 4 insertions(+), 4 deletions(-) diff --git a/scripts/postprocessing_caption.py b/scripts/postprocessing_caption.py index 9482a03ca..5592a8987 100644 --- a/scripts/postprocessing_caption.py +++ b/scripts/postprocessing_caption.py @@ -4,7 +4,7 @@ import gradio as gr class ScriptPostprocessingCeption(scripts_postprocessing.ScriptPostprocessing): name = "Caption" - order = 4000 + order = 4040 def ui(self): with ui_components.InputAccordion(False, label="Caption") as enable: diff --git a/scripts/postprocessing_create_flipped_copies.py b/scripts/postprocessing_create_flipped_copies.py index 3425571dc..b673003b6 100644 --- a/scripts/postprocessing_create_flipped_copies.py +++ b/scripts/postprocessing_create_flipped_copies.py @@ -6,7 +6,7 @@ import gradio as gr class ScriptPostprocessingCreateFlippedCopies(scripts_postprocessing.ScriptPostprocessing): name = "Create flipped copies" - order = 4000 + order = 4030 def ui(self): with ui_components.InputAccordion(False, label="Create flipped copies") as enable: diff --git a/scripts/postprocessing_focal_crop.py b/scripts/postprocessing_focal_crop.py index d3baf2987..cff1dbc54 100644 --- a/scripts/postprocessing_focal_crop.py +++ b/scripts/postprocessing_focal_crop.py @@ -7,7 +7,7 @@ from modules.textual_inversion import autocrop class ScriptPostprocessingFocalCrop(scripts_postprocessing.ScriptPostprocessing): name = "Auto focal point crop" - order = 4000 + order = 4010 def ui(self): with ui_components.InputAccordion(False, label="Auto focal point crop") as enable: diff --git a/scripts/processing_autosized_crop.py b/scripts/processing_autosized_crop.py index c09802264..7e6749898 100644 --- a/scripts/processing_autosized_crop.py +++ b/scripts/processing_autosized_crop.py @@ -28,7 +28,7 @@ def multicrop_pic(image: Image, mindim, maxdim, minarea, maxarea, objective, thr class ScriptPostprocessingAutosizedCrop(scripts_postprocessing.ScriptPostprocessing): name = "Auto-sized crop" - order = 4000 + order = 4020 def ui(self): with ui_components.InputAccordion(False, label="Auto-sized crop") as enable: From 9feb034e343d6d7ef63395821658fb3774b30a24 Mon Sep 17 00:00:00 2001 From: wangqyqq Date: Thu, 21 Dec 2023 20:15:51 +0800 Subject: [PATCH 284/342] support for sdxl-inpaint model --- configs/sd_xl_inpaint.yaml | 98 +++++++++++++++++++++++++++++++++++++ modules/processing.py | 19 +++++++ modules/sd_models_config.py | 6 ++- modules/sd_models_xl.py | 5 ++ 4 files changed, 127 insertions(+), 1 deletion(-) create mode 100644 configs/sd_xl_inpaint.yaml diff --git a/configs/sd_xl_inpaint.yaml b/configs/sd_xl_inpaint.yaml new file mode 100644 index 000000000..3bad37218 --- /dev/null +++ b/configs/sd_xl_inpaint.yaml @@ -0,0 +1,98 @@ +model: + target: sgm.models.diffusion.DiffusionEngine + params: + scale_factor: 0.13025 + disable_first_stage_autocast: True + + denoiser_config: + target: sgm.modules.diffusionmodules.denoiser.DiscreteDenoiser + params: + num_idx: 1000 + + weighting_config: + target: sgm.modules.diffusionmodules.denoiser_weighting.EpsWeighting + scaling_config: + target: sgm.modules.diffusionmodules.denoiser_scaling.EpsScaling + discretization_config: + target: sgm.modules.diffusionmodules.discretizer.LegacyDDPMDiscretization + + network_config: + target: sgm.modules.diffusionmodules.openaimodel.UNetModel + params: + adm_in_channels: 2816 + num_classes: sequential + use_checkpoint: True + in_channels: 9 + out_channels: 4 + model_channels: 320 + attention_resolutions: [4, 2] + num_res_blocks: 2 + channel_mult: [1, 2, 4] + num_head_channels: 64 + use_spatial_transformer: True + use_linear_in_transformer: True + transformer_depth: [1, 2, 10] # note: the first is unused (due to attn_res starting at 2) 32, 16, 8 --> 64, 32, 16 + context_dim: 2048 + spatial_transformer_attn_type: softmax-xformers + legacy: False + + conditioner_config: + target: sgm.modules.GeneralConditioner + params: + emb_models: + # crossattn cond + - is_trainable: False + input_key: txt + target: sgm.modules.encoders.modules.FrozenCLIPEmbedder + params: + layer: hidden + layer_idx: 11 + # crossattn and vector cond + - is_trainable: False + input_key: txt + target: sgm.modules.encoders.modules.FrozenOpenCLIPEmbedder2 + params: + arch: ViT-bigG-14 + version: laion2b_s39b_b160k + freeze: True + layer: penultimate + always_return_pooled: True + legacy: False + # vector cond + - is_trainable: False + input_key: original_size_as_tuple + target: sgm.modules.encoders.modules.ConcatTimestepEmbedderND + params: + outdim: 256 # multiplied by two + # vector cond + - is_trainable: False + input_key: crop_coords_top_left + target: sgm.modules.encoders.modules.ConcatTimestepEmbedderND + params: + outdim: 256 # multiplied by two + # vector cond + - is_trainable: False + input_key: target_size_as_tuple + target: sgm.modules.encoders.modules.ConcatTimestepEmbedderND + params: + outdim: 256 # multiplied by two + + first_stage_config: + target: sgm.models.autoencoder.AutoencoderKLInferenceWrapper + params: + embed_dim: 4 + monitor: val/rec_loss + ddconfig: + attn_type: vanilla-xformers + double_z: true + z_channels: 4 + resolution: 256 + in_channels: 3 + out_ch: 3 + ch: 128 + ch_mult: [1, 2, 4, 4] + num_res_blocks: 2 + attn_resolutions: [] + dropout: 0.0 + lossconfig: + target: torch.nn.Identity diff --git a/modules/processing.py b/modules/processing.py index 6f01c95f5..159548dba 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -106,6 +106,20 @@ def txt2img_image_conditioning(sd_model, x, width, height): return x.new_zeros(x.shape[0], 2*sd_model.noise_augmentor.time_embed.dim, dtype=x.dtype, device=x.device) else: + sd = sd_model.model.state_dict() + diffusion_model_input = sd.get('diffusion_model.input_blocks.0.0.weight', None) + if diffusion_model_input.shape[1] == 9: + # The "masked-image" in this case will just be all 0.5 since the entire image is masked. + image_conditioning = torch.ones(x.shape[0], 3, height, width, device=x.device) * 0.5 + image_conditioning = images_tensor_to_samples(image_conditioning, + approximation_indexes.get(opts.sd_vae_encode_method)) + + # Add the fake full 1s mask to the first dimension. + image_conditioning = torch.nn.functional.pad(image_conditioning, (0, 0, 0, 0, 1, 0), value=1.0) + image_conditioning = image_conditioning.to(x.dtype) + + return image_conditioning + # Dummy zero conditioning if we're not using inpainting or unclip models. # Still takes up a bit of memory, but no encoder call. # Pretty sure we can just make this a 1x1 image since its not going to be used besides its batch size. @@ -362,6 +376,11 @@ class StableDiffusionProcessing: if self.sampler.conditioning_key == "crossattn-adm": return self.unclip_image_conditioning(source_image) + sd = self.sampler.model_wrap.inner_model.model.state_dict() + diffusion_model_input = sd.get('diffusion_model.input_blocks.0.0.weight', None) + if diffusion_model_input.shape[1] == 9: + return self.inpainting_image_conditioning(source_image, latent_image, image_mask=image_mask) + # Dummy zero conditioning if we're not using inpainting or depth model. return latent_image.new_zeros(latent_image.shape[0], 5, 1, 1) diff --git a/modules/sd_models_config.py b/modules/sd_models_config.py index deab2f6e2..b38137eb5 100644 --- a/modules/sd_models_config.py +++ b/modules/sd_models_config.py @@ -15,6 +15,7 @@ config_sd2v = os.path.join(sd_repo_configs_path, "v2-inference-v.yaml") config_sd2_inpainting = os.path.join(sd_repo_configs_path, "v2-inpainting-inference.yaml") config_sdxl = os.path.join(sd_xl_repo_configs_path, "sd_xl_base.yaml") config_sdxl_refiner = os.path.join(sd_xl_repo_configs_path, "sd_xl_refiner.yaml") +config_sdxl_inpainting = os.path.join(sd_configs_path, "sd_xl_inpaint.yaml") config_depth_model = os.path.join(sd_repo_configs_path, "v2-midas-inference.yaml") config_unclip = os.path.join(sd_repo_configs_path, "v2-1-stable-unclip-l-inference.yaml") config_unopenclip = os.path.join(sd_repo_configs_path, "v2-1-stable-unclip-h-inference.yaml") @@ -71,7 +72,10 @@ def guess_model_config_from_state_dict(sd, filename): sd2_variations_weight = sd.get('embedder.model.ln_final.weight', None) if sd.get('conditioner.embedders.1.model.ln_final.weight', None) is not None: - return config_sdxl + if diffusion_model_input.shape[1] == 9: + return config_sdxl_inpainting + else: + return config_sdxl if sd.get('conditioner.embedders.0.model.ln_final.weight', None) is not None: return config_sdxl_refiner elif sd.get('depth_model.model.pretrained.act_postprocess3.0.project.0.bias', None) is not None: diff --git a/modules/sd_models_xl.py b/modules/sd_models_xl.py index 011233216..d8a9a73bc 100644 --- a/modules/sd_models_xl.py +++ b/modules/sd_models_xl.py @@ -34,6 +34,11 @@ def get_learned_conditioning(self: sgm.models.diffusion.DiffusionEngine, batch: def apply_model(self: sgm.models.diffusion.DiffusionEngine, x, t, cond): + sd = self.model.state_dict() + diffusion_model_input = sd.get('diffusion_model.input_blocks.0.0.weight', None) + if diffusion_model_input.shape[1] == 9: + x = torch.cat([x] + cond['c_concat'], dim=1) + return self.model(x, t, cond) From de1809bd14450cfc41623b6021c7087fb385ab6f Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Fri, 22 Dec 2023 00:08:35 +0900 Subject: [PATCH 285/342] handle axis_type is None --- scripts/xyz_grid.py | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/scripts/xyz_grid.py b/scripts/xyz_grid.py index b2250c04d..e5083874a 100644 --- a/scripts/xyz_grid.py +++ b/scripts/xyz_grid.py @@ -476,6 +476,8 @@ class Script(scripts.Script): fill_z_button.click(fn=fill, inputs=[z_type, csv_mode], outputs=[z_values, z_values_dropdown]) def select_axis(axis_type, axis_values, axis_values_dropdown, csv_mode): + axis_type = axis_type or 0 # if axle type is None set to 0 + choices = self.current_axis_options[axis_type].choices has_choices = choices is not None @@ -526,6 +528,8 @@ class Script(scripts.Script): return [x_type, x_values, x_values_dropdown, y_type, y_values, y_values_dropdown, z_type, z_values, z_values_dropdown, draw_legend, include_lone_images, include_sub_grids, no_fixed_seeds, margin_size, csv_mode] def run(self, p, x_type, x_values, x_values_dropdown, y_type, y_values, y_values_dropdown, z_type, z_values, z_values_dropdown, draw_legend, include_lone_images, include_sub_grids, no_fixed_seeds, margin_size, csv_mode): + x_type, y_type, z_type = x_type or 0, y_type or 0, z_type or 0 # if axle type is None set to 0 + if not no_fixed_seeds: modules.processing.fix_seed(p) From edfae95d90a49ea95394b772817a59dde4175222 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Sat, 23 Dec 2023 01:21:00 +0900 Subject: [PATCH 286/342] prevent crash due to Script __init__ exception --- modules/scripts.py | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/modules/scripts.py b/modules/scripts.py index b6fcf96e0..3a7669118 100644 --- a/modules/scripts.py +++ b/modules/scripts.py @@ -566,7 +566,12 @@ class ScriptRunner: auto_processing_scripts = scripts_auto_postprocessing.create_auto_preprocessing_script_data() for script_data in auto_processing_scripts + scripts_data: - script = script_data.script_class() + try: + script = script_data.script_class() + except Exception: + errors.report(f"Error # failed to initialize Script {script_data.module}: ", exc_info=True) + continue + script.filename = script_data.path script.is_txt2img = not is_img2img script.is_img2img = is_img2img From 00d4a4d4ac75903d8224e9beb1136584dd66fcd8 Mon Sep 17 00:00:00 2001 From: lanyeeee <1210347077@qq.com> Date: Tue, 26 Dec 2023 14:46:29 +0800 Subject: [PATCH 287/342] move thread-unsafe code to __init__ --- modules/api/api.py | 19 +++++++++---------- 1 file changed, 9 insertions(+), 10 deletions(-) diff --git a/modules/api/api.py b/modules/api/api.py index 7154c9d5b..f0a68c672 100644 --- a/modules/api/api.py +++ b/modules/api/api.py @@ -251,6 +251,15 @@ class Api: self.default_script_arg_txt2img = [] self.default_script_arg_img2img = [] + script_runner = scripts.scripts_img2img + if not script_runner.scripts: + script_runner.initialize_scripts(True) + ui.create_ui() + if not self.default_script_arg_txt2img: + self.default_script_arg_txt2img = self.init_default_script_args(script_runner) + if not self.default_script_arg_img2img: + self.default_script_arg_img2img = self.init_default_script_args(script_runner) + def add_api_route(self, path: str, endpoint, **kwargs): if shared.cmd_opts.api_auth: return self.app.add_api_route(path, endpoint, dependencies=[Depends(self.auth)], **kwargs) @@ -339,11 +348,6 @@ class Api: task_id = txt2imgreq.force_task_id or create_task_id("txt2img") script_runner = scripts.scripts_txt2img - if not script_runner.scripts: - script_runner.initialize_scripts(False) - ui.create_ui() - if not self.default_script_arg_txt2img: - self.default_script_arg_txt2img = self.init_default_script_args(script_runner) selectable_scripts, selectable_script_idx = self.get_selectable_script(txt2imgreq.script_name, script_runner) populate = txt2imgreq.copy(update={ # Override __init__ params @@ -403,11 +407,6 @@ class Api: mask = decode_base64_to_image(mask) script_runner = scripts.scripts_img2img - if not script_runner.scripts: - script_runner.initialize_scripts(True) - ui.create_ui() - if not self.default_script_arg_img2img: - self.default_script_arg_img2img = self.init_default_script_args(script_runner) selectable_scripts, selectable_script_idx = self.get_selectable_script(img2imgreq.script_name, script_runner) populate = img2imgreq.copy(update={ # Override __init__ params From bfe418a58d39c69ca2672e7d8a1fd7ad2b34869b Mon Sep 17 00:00:00 2001 From: wangqyqq Date: Wed, 27 Dec 2023 10:20:56 +0800 Subject: [PATCH 288/342] add some codes for robust --- modules/processing.py | 24 +++++++++++++----------- modules/sd_models_xl.py | 5 +++-- 2 files changed, 16 insertions(+), 13 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index 159548dba..c05e608ab 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -108,17 +108,18 @@ def txt2img_image_conditioning(sd_model, x, width, height): else: sd = sd_model.model.state_dict() diffusion_model_input = sd.get('diffusion_model.input_blocks.0.0.weight', None) - if diffusion_model_input.shape[1] == 9: - # The "masked-image" in this case will just be all 0.5 since the entire image is masked. - image_conditioning = torch.ones(x.shape[0], 3, height, width, device=x.device) * 0.5 - image_conditioning = images_tensor_to_samples(image_conditioning, - approximation_indexes.get(opts.sd_vae_encode_method)) + if diffusion_model_input is not None: + if diffusion_model_input.shape[1] == 9: + # The "masked-image" in this case will just be all 0.5 since the entire image is masked. + image_conditioning = torch.ones(x.shape[0], 3, height, width, device=x.device) * 0.5 + image_conditioning = images_tensor_to_samples(image_conditioning, + approximation_indexes.get(opts.sd_vae_encode_method)) - # Add the fake full 1s mask to the first dimension. - image_conditioning = torch.nn.functional.pad(image_conditioning, (0, 0, 0, 0, 1, 0), value=1.0) - image_conditioning = image_conditioning.to(x.dtype) + # Add the fake full 1s mask to the first dimension. + image_conditioning = torch.nn.functional.pad(image_conditioning, (0, 0, 0, 0, 1, 0), value=1.0) + image_conditioning = image_conditioning.to(x.dtype) - return image_conditioning + return image_conditioning # Dummy zero conditioning if we're not using inpainting or unclip models. # Still takes up a bit of memory, but no encoder call. @@ -378,8 +379,9 @@ class StableDiffusionProcessing: sd = self.sampler.model_wrap.inner_model.model.state_dict() diffusion_model_input = sd.get('diffusion_model.input_blocks.0.0.weight', None) - if diffusion_model_input.shape[1] == 9: - return self.inpainting_image_conditioning(source_image, latent_image, image_mask=image_mask) + if diffusion_model_input is not None: + if diffusion_model_input.shape[1] == 9: + return self.inpainting_image_conditioning(source_image, latent_image, image_mask=image_mask) # Dummy zero conditioning if we're not using inpainting or depth model. return latent_image.new_zeros(latent_image.shape[0], 5, 1, 1) diff --git a/modules/sd_models_xl.py b/modules/sd_models_xl.py index d8a9a73bc..162d0fee8 100644 --- a/modules/sd_models_xl.py +++ b/modules/sd_models_xl.py @@ -36,8 +36,9 @@ def get_learned_conditioning(self: sgm.models.diffusion.DiffusionEngine, batch: def apply_model(self: sgm.models.diffusion.DiffusionEngine, x, t, cond): sd = self.model.state_dict() diffusion_model_input = sd.get('diffusion_model.input_blocks.0.0.weight', None) - if diffusion_model_input.shape[1] == 9: - x = torch.cat([x] + cond['c_concat'], dim=1) + if diffusion_model_input is not None: + if diffusion_model_input.shape[1] == 9: + x = torch.cat([x] + cond['c_concat'], dim=1) return self.model(x, t, cond) From de04573438bc111f137359b8f4998780bf315275 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Thu, 28 Dec 2023 06:22:51 +0900 Subject: [PATCH 289/342] create utility truncate_path utli.truncate_path(target_path, base_path) return the target_path relative to base_path if target_path is a sub path of base_path else return the absolute path --- modules/util.py | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-) diff --git a/modules/util.py b/modules/util.py index 60afc0670..4861bcb08 100644 --- a/modules/util.py +++ b/modules/util.py @@ -2,7 +2,7 @@ import os import re from modules import shared -from modules.paths_internal import script_path +from modules.paths_internal import script_path, cwd def natural_sort_key(s, regex=re.compile('([0-9]+)')): @@ -56,3 +56,13 @@ def ldm_print(*args, **kwargs): return print(*args, **kwargs) + + +def truncate_path(target_path, base_path=cwd): + abs_target, abs_base = os.path.abspath(target_path), os.path.abspath(base_path) + try: + if os.path.commonpath([abs_target, abs_base]) == abs_base: + return os.path.relpath(abs_target, abs_base) + except ValueError: + pass + return abs_target From af2951ed53da6d357aea9232538f9ea7e1cdc648 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Thu, 28 Dec 2023 06:52:33 +0900 Subject: [PATCH 290/342] base default image output on data_path Co-Authored-By: Alberto Cano <34340962+canoalberto@users.noreply.github.com> --- modules/paths_internal.py | 1 + modules/shared_options.py | 19 ++++++++++--------- 2 files changed, 11 insertions(+), 9 deletions(-) diff --git a/modules/paths_internal.py b/modules/paths_internal.py index 89131a54f..b86ecd7f1 100644 --- a/modules/paths_internal.py +++ b/modules/paths_internal.py @@ -28,5 +28,6 @@ models_path = os.path.join(data_path, "models") extensions_dir = os.path.join(data_path, "extensions") extensions_builtin_dir = os.path.join(script_path, "extensions-builtin") config_states_dir = os.path.join(script_path, "config_states") +default_output_dir = os.path.join(data_path, "output") roboto_ttf_file = os.path.join(modules_path, 'Roboto-Regular.ttf') diff --git a/modules/shared_options.py b/modules/shared_options.py index fa542ba8c..752a4f125 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -1,7 +1,8 @@ +import os import gradio as gr -from modules import localization, ui_components, shared_items, shared, interrogate, shared_gradio_themes -from modules.paths_internal import models_path, script_path, data_path, sd_configs_path, sd_default_config, sd_model_file, default_sd_model_file, extensions_dir, extensions_builtin_dir # noqa: F401 +from modules import localization, ui_components, shared_items, shared, interrogate, shared_gradio_themes, util +from modules.paths_internal import models_path, script_path, data_path, sd_configs_path, sd_default_config, sd_model_file, default_sd_model_file, extensions_dir, extensions_builtin_dir, default_output_dir # noqa: F401 from modules.shared_cmd_options import cmd_opts from modules.options import options_section, OptionInfo, OptionHTML, categories @@ -74,14 +75,14 @@ options_templates.update(options_section(('saving-images', "Saving images/grids" options_templates.update(options_section(('saving-paths', "Paths for saving", "saving"), { "outdir_samples": OptionInfo("", "Output directory for images; if empty, defaults to three directories below", component_args=hide_dirs), - "outdir_txt2img_samples": OptionInfo("outputs/txt2img-images", 'Output directory for txt2img images', component_args=hide_dirs), - "outdir_img2img_samples": OptionInfo("outputs/img2img-images", 'Output directory for img2img images', component_args=hide_dirs), - "outdir_extras_samples": OptionInfo("outputs/extras-images", 'Output directory for images from extras tab', component_args=hide_dirs), + "outdir_txt2img_samples": OptionInfo(util.truncate_path(os.path.join(default_output_dir, 'txt2img-images')), 'Output directory for txt2img images', component_args=hide_dirs), + "outdir_img2img_samples": OptionInfo(util.truncate_path(os.path.join(default_output_dir, 'img2img-images')), 'Output directory for img2img images', component_args=hide_dirs), + "outdir_extras_samples": OptionInfo(util.truncate_path(os.path.join(default_output_dir, 'extras-images')), 'Output directory for images from extras tab', component_args=hide_dirs), "outdir_grids": OptionInfo("", "Output directory for grids; if empty, defaults to two directories below", component_args=hide_dirs), - "outdir_txt2img_grids": OptionInfo("outputs/txt2img-grids", 'Output directory for txt2img grids', component_args=hide_dirs), - "outdir_img2img_grids": OptionInfo("outputs/img2img-grids", 'Output directory for img2img grids', component_args=hide_dirs), - "outdir_save": OptionInfo("log/images", "Directory for saving images using the Save button", component_args=hide_dirs), - "outdir_init_images": OptionInfo("outputs/init-images", "Directory for saving init images when using img2img", component_args=hide_dirs), + "outdir_txt2img_grids": OptionInfo(util.truncate_path(os.path.join(default_output_dir, 'txt2img-grids')), 'Output directory for txt2img grids', component_args=hide_dirs), + "outdir_img2img_grids": OptionInfo(util.truncate_path(os.path.join(default_output_dir, 'img2img-grids')), 'Output directory for img2img grids', component_args=hide_dirs), + "outdir_save": OptionInfo(util.truncate_path(os.path.join(data_path, 'log', 'images')), "Directory for saving images using the Save button", component_args=hide_dirs), + "outdir_init_images": OptionInfo(util.truncate_path(os.path.join(default_output_dir, 'init-images')), "Directory for saving init images when using img2img", component_args=hide_dirs), })) options_templates.update(options_section(('saving-to-dirs', "Saving to a directory", "saving"), { From 892e703b59b2f867d8a202a52fab1db89882ef86 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Thu, 28 Dec 2023 06:52:41 +0900 Subject: [PATCH 291/342] webpath use truncate_path --- modules/ui_gradio_extensions.py | 11 +++-------- 1 file changed, 3 insertions(+), 8 deletions(-) diff --git a/modules/ui_gradio_extensions.py b/modules/ui_gradio_extensions.py index 0d368f8b2..a86c368ef 100644 --- a/modules/ui_gradio_extensions.py +++ b/modules/ui_gradio_extensions.py @@ -1,17 +1,12 @@ import os import gradio as gr -from modules import localization, shared, scripts -from modules.paths import script_path, data_path, cwd +from modules import localization, shared, scripts, util +from modules.paths import script_path, data_path def webpath(fn): - if fn.startswith(cwd): - web_path = os.path.relpath(fn, cwd) - else: - web_path = os.path.abspath(fn) - - return f'file={web_path}?{os.path.getmtime(fn)}' + return f'file={util.truncate_path(fn)}?{os.path.getmtime(fn)}' def javascript_html(): From dc57ec0296e768ee91290e16ab262404837c566d Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Fri, 29 Dec 2023 01:56:48 +0900 Subject: [PATCH 292/342] save info of init image --- modules/processing.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/processing.py b/modules/processing.py index 9351e3fb0..141f2f111 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -1482,7 +1482,7 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing): # Save init image if opts.save_init_img: self.init_img_hash = hashlib.md5(img.tobytes()).hexdigest() - images.save_image(img, path=opts.outdir_init_images, basename=None, forced_filename=self.init_img_hash, save_to_dirs=False) + images.save_image(img, path=opts.outdir_init_images, basename=None, forced_filename=self.init_img_hash, save_to_dirs=False, existing_info=img.info) image = images.flatten(img, opts.img2img_background_color) From bb07cb6a0df60a96827125ffc09ea182a1ed272c Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sun, 17 Dec 2023 10:22:03 +0300 Subject: [PATCH 293/342] a --- modules/api/api.py | 27 +++++++++++++ modules/api/models.py | 2 + modules/generation_parameters_copypaste.py | 19 +++++++++ modules/processing.py | 2 +- modules/processing_scripts/refiner.py | 7 ++-- modules/processing_scripts/seed.py | 13 +++--- modules/ui.py | 46 +++++++++++----------- 7 files changed, 83 insertions(+), 33 deletions(-) diff --git a/modules/api/api.py b/modules/api/api.py index 7154c9d5b..b3d709404 100644 --- a/modules/api/api.py +++ b/modules/api/api.py @@ -335,6 +335,29 @@ class Api: script_args[alwayson_script.args_from + idx] = request.alwayson_scripts[alwayson_script_name]["args"][idx] return script_args + def apply_infotext(self, request, tabname): + if not request.infotext: + return {} + + params = generation_parameters_copypaste.parse_generation_parameters(request.infotext) + + for field in generation_parameters_copypaste.paste_fields[tabname]["fields"]: + if not field.api: + continue + + value = field.function(params) if field.function else params.get(field.label) + target_type = request.__fields__[field.api].type_ + + if value is None: + continue + + if not isinstance(value, target_type): + value = target_type(value) + + setattr(request, field.api, value) + + return params + def text2imgapi(self, txt2imgreq: models.StableDiffusionTxt2ImgProcessingAPI): task_id = txt2imgreq.force_task_id or create_task_id("txt2img") @@ -342,6 +365,9 @@ class Api: if not script_runner.scripts: script_runner.initialize_scripts(False) ui.create_ui() + + infotext_params = self.apply_infotext(txt2imgreq, "txt2img") + if not self.default_script_arg_txt2img: self.default_script_arg_txt2img = self.init_default_script_args(script_runner) selectable_scripts, selectable_script_idx = self.get_selectable_script(txt2imgreq.script_name, script_runner) @@ -358,6 +384,7 @@ class Api: args.pop('script_name', None) args.pop('script_args', None) # will refeed them to the pipeline directly after initializing them args.pop('alwayson_scripts', None) + args.pop('infotext', None) script_args = self.init_script_args(txt2imgreq, self.default_script_arg_txt2img, selectable_scripts, selectable_script_idx, script_runner) diff --git a/modules/api/models.py b/modules/api/models.py index 58083a34f..16edf11cf 100644 --- a/modules/api/models.py +++ b/modules/api/models.py @@ -108,6 +108,7 @@ StableDiffusionTxt2ImgProcessingAPI = PydanticModelGenerator( {"key": "save_images", "type": bool, "default": False}, {"key": "alwayson_scripts", "type": dict, "default": {}}, {"key": "force_task_id", "type": str, "default": None}, + {"key": "infotext", "type": str, "default": None}, ] ).generate_model() @@ -126,6 +127,7 @@ StableDiffusionImg2ImgProcessingAPI = PydanticModelGenerator( {"key": "save_images", "type": bool, "default": False}, {"key": "alwayson_scripts", "type": dict, "default": {}}, {"key": "force_task_id", "type": str, "default": None}, + {"key": "infotext", "type": str, "default": None}, ] ).generate_model() diff --git a/modules/generation_parameters_copypaste.py b/modules/generation_parameters_copypaste.py index dbffe4949..4b4727c44 100644 --- a/modules/generation_parameters_copypaste.py +++ b/modules/generation_parameters_copypaste.py @@ -28,6 +28,19 @@ class ParamBinding: self.paste_field_names = paste_field_names or [] +class PasteField(tuple): + def __new__(cls, component, target, *, api=None): + return super().__new__(cls, (component, target)) + + def __init__(self, component, target, *, api=None): + super().__init__() + + self.api = api + self.component = component + self.label = target if isinstance(target, str) else None + self.function = target if callable(target) else None + + paste_fields: dict[str, dict] = {} registered_param_bindings: list[ParamBinding] = [] @@ -84,6 +97,12 @@ def image_from_url_text(filedata): def add_paste_fields(tabname, init_img, fields, override_settings_component=None): + + if fields: + for i in range(len(fields)): + if not isinstance(fields[i], PasteField): + fields[i] = PasteField(*fields[i]) + paste_fields[tabname] = {"init_img": init_img, "fields": fields, "override_settings_component": override_settings_component} # backwards compatibility for existing extensions diff --git a/modules/processing.py b/modules/processing.py index 9351e3fb0..ee2ccf460 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -1135,7 +1135,7 @@ class StableDiffusionProcessingTxt2Img(StableDiffusionProcessing): def init(self, all_prompts, all_seeds, all_subseeds): if self.enable_hr: - if self.hr_checkpoint_name: + if self.hr_checkpoint_name and self.hr_checkpoint_name != 'Use same checkpoint': self.hr_checkpoint_info = sd_models.get_closet_checkpoint_match(self.hr_checkpoint_name) if self.hr_checkpoint_info is None: diff --git a/modules/processing_scripts/refiner.py b/modules/processing_scripts/refiner.py index 29ccb78f9..cefad32b7 100644 --- a/modules/processing_scripts/refiner.py +++ b/modules/processing_scripts/refiner.py @@ -1,6 +1,7 @@ import gradio as gr from modules import scripts, sd_models +from modules.generation_parameters_copypaste import PasteField from modules.ui_common import create_refresh_button from modules.ui_components import InputAccordion @@ -31,9 +32,9 @@ class ScriptRefiner(scripts.ScriptBuiltinUI): return None if info is None else info.title self.infotext_fields = [ - (enable_refiner, lambda d: 'Refiner' in d), - (refiner_checkpoint, lambda d: lookup_checkpoint(d.get('Refiner'))), - (refiner_switch_at, 'Refiner switch at'), + PasteField(enable_refiner, lambda d: 'Refiner' in d), + PasteField(refiner_checkpoint, lambda d: lookup_checkpoint(d.get('Refiner')), api="refiner_checkpoint"), + PasteField(refiner_switch_at, 'Refiner switch at', api="refiner_switch_at"), ] return enable_refiner, refiner_checkpoint, refiner_switch_at diff --git a/modules/processing_scripts/seed.py b/modules/processing_scripts/seed.py index dc9c2da50..a3e16a12e 100644 --- a/modules/processing_scripts/seed.py +++ b/modules/processing_scripts/seed.py @@ -3,6 +3,7 @@ import json import gradio as gr from modules import scripts, ui, errors +from modules.generation_parameters_copypaste import PasteField from modules.shared import cmd_opts from modules.ui_components import ToolButton @@ -51,12 +52,12 @@ class ScriptSeed(scripts.ScriptBuiltinUI): seed_checkbox.change(lambda x: gr.update(visible=x), show_progress=False, inputs=[seed_checkbox], outputs=[seed_extras]) self.infotext_fields = [ - (self.seed, "Seed"), - (seed_checkbox, lambda d: "Variation seed" in d or "Seed resize from-1" in d), - (subseed, "Variation seed"), - (subseed_strength, "Variation seed strength"), - (seed_resize_from_w, "Seed resize from-1"), - (seed_resize_from_h, "Seed resize from-2"), + PasteField(self.seed, "Seed", api="seed"), + PasteField(seed_checkbox, lambda d: "Variation seed" in d or "Seed resize from-1" in d), + PasteField(subseed, "Variation seed", api="subseed"), + PasteField(subseed_strength, "Variation seed strength", api="subseed_strength"), + PasteField(seed_resize_from_w, "Seed resize from-1", api="seed_resize_from_h"), + PasteField(seed_resize_from_h, "Seed resize from-2", api="seed_resize_from_w"), ] self.on_after_component(lambda x: connect_reuse_seed(self.seed, reuse_seed, x.component, False), elem_id=f'generation_info_{self.tabname}') diff --git a/modules/ui.py b/modules/ui.py index d80486dd4..9db2407ed 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -28,7 +28,7 @@ import modules.textual_inversion.textual_inversion as textual_inversion import modules.shared as shared from modules import prompt_parser from modules.sd_hijack import model_hijack -from modules.generation_parameters_copypaste import image_from_url_text +from modules.generation_parameters_copypaste import image_from_url_text, PasteField create_setting_component = ui_settings.create_setting_component @@ -436,28 +436,28 @@ def create_ui(): ) txt2img_paste_fields = [ - (toprow.prompt, "Prompt"), - (toprow.negative_prompt, "Negative prompt"), - (steps, "Steps"), - (sampler_name, "Sampler"), - (cfg_scale, "CFG scale"), - (width, "Size-1"), - (height, "Size-2"), - (batch_size, "Batch size"), - (toprow.ui_styles.dropdown, lambda d: d["Styles array"] if isinstance(d.get("Styles array"), list) else gr.update()), - (denoising_strength, "Denoising strength"), - (enable_hr, lambda d: "Denoising strength" in d and ("Hires upscale" in d or "Hires upscaler" in d or "Hires resize-1" in d)), - (hr_scale, "Hires upscale"), - (hr_upscaler, "Hires upscaler"), - (hr_second_pass_steps, "Hires steps"), - (hr_resize_x, "Hires resize-1"), - (hr_resize_y, "Hires resize-2"), - (hr_checkpoint_name, "Hires checkpoint"), - (hr_sampler_name, "Hires sampler"), - (hr_sampler_container, lambda d: gr.update(visible=True) if d.get("Hires sampler", "Use same sampler") != "Use same sampler" or d.get("Hires checkpoint", "Use same checkpoint") != "Use same checkpoint" else gr.update()), - (hr_prompt, "Hires prompt"), - (hr_negative_prompt, "Hires negative prompt"), - (hr_prompts_container, lambda d: gr.update(visible=True) if d.get("Hires prompt", "") != "" or d.get("Hires negative prompt", "") != "" else gr.update()), + PasteField(toprow.prompt, "Prompt", api="prompt"), + PasteField(toprow.negative_prompt, "Negative prompt", api="negative_prompt"), + PasteField(steps, "Steps", api="steps"), + PasteField(sampler_name, "Sampler", api="sampler_name"), + PasteField(cfg_scale, "CFG scale", api="cfg_scale"), + PasteField(width, "Size-1", api="width"), + PasteField(height, "Size-2", api="height"), + PasteField(batch_size, "Batch size", api="batch_size"), + PasteField(toprow.ui_styles.dropdown, lambda d: d["Styles array"] if isinstance(d.get("Styles array"), list) else gr.update(), api="styles"), + PasteField(denoising_strength, "Denoising strength", api="denoising_strength"), + PasteField(enable_hr, lambda d: "Denoising strength" in d and ("Hires upscale" in d or "Hires upscaler" in d or "Hires resize-1" in d), api="enable_hr"), + PasteField(hr_scale, "Hires upscale", api="hr_scale"), + PasteField(hr_upscaler, "Hires upscaler", api="hr_upscaler"), + PasteField(hr_second_pass_steps, "Hires steps", api="hr_second_pass_steps"), + PasteField(hr_resize_x, "Hires resize-1", api="hr_resize_x"), + PasteField(hr_resize_y, "Hires resize-2", api="hr_resize_y"), + PasteField(hr_checkpoint_name, "Hires checkpoint", api="hr_checkpoint_name"), + PasteField(hr_sampler_name, "Hires sampler", api="hr_sampler_name"), + PasteField(hr_sampler_container, lambda d: gr.update(visible=True) if d.get("Hires sampler", "Use same sampler") != "Use same sampler" or d.get("Hires checkpoint", "Use same checkpoint") != "Use same checkpoint" else gr.update()), + PasteField(hr_prompt, "Hires prompt", api="hr_prompt"), + PasteField(hr_negative_prompt, "Hires negative prompt", api="hr_negative_prompt"), + PasteField(hr_prompts_container, lambda d: gr.update(visible=True) if d.get("Hires prompt", "") != "" or d.get("Hires negative prompt", "") != "" else gr.update()), *scripts.scripts_txt2img.infotext_fields ] parameters_copypaste.add_paste_fields("txt2img", None, txt2img_paste_fields, override_settings) From 59d060fd5ea93fcc3fdbfbd13b6e20fda06ecf94 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Sat, 30 Dec 2023 17:11:03 +0900 Subject: [PATCH 294/342] More lora not found warning --- extensions-builtin/Lora/networks.py | 8 +++++++- extensions-builtin/Lora/scripts/lora_script.py | 2 ++ 2 files changed, 9 insertions(+), 1 deletion(-) diff --git a/extensions-builtin/Lora/networks.py b/extensions-builtin/Lora/networks.py index 985b2753b..72ebd6241 100644 --- a/extensions-builtin/Lora/networks.py +++ b/extensions-builtin/Lora/networks.py @@ -1,3 +1,4 @@ +import gradio as gr import logging import os import re @@ -314,7 +315,12 @@ def load_networks(names, te_multipliers=None, unet_multipliers=None, dyn_dims=No emb_db.skipped_embeddings[name] = embedding if failed_to_load_networks: - sd_hijack.model_hijack.comments.append("Networks not found: " + ", ".join(failed_to_load_networks)) + lora_not_found_message = f'Lora not found: {", ".join(failed_to_load_networks)}' + sd_hijack.model_hijack.comments.append(lora_not_found_message) + if shared.opts.lora_not_found_warning_console: + print(f'\n{lora_not_found_message}\n') + if shared.opts.lora_not_found_gradio_warning: + gr.Warning(lora_not_found_message) purge_networks_from_memory() diff --git a/extensions-builtin/Lora/scripts/lora_script.py b/extensions-builtin/Lora/scripts/lora_script.py index ef23968c5..1518f7e5c 100644 --- a/extensions-builtin/Lora/scripts/lora_script.py +++ b/extensions-builtin/Lora/scripts/lora_script.py @@ -39,6 +39,8 @@ shared.options_templates.update(shared.options_section(('extra_networks', "Extra "lora_show_all": shared.OptionInfo(False, "Always show all networks on the Lora page").info("otherwise, those detected as for incompatible version of Stable Diffusion will be hidden"), "lora_hide_unknown_for_versions": shared.OptionInfo([], "Hide networks of unknown versions for model versions", gr.CheckboxGroup, {"choices": ["SD1", "SD2", "SDXL"]}), "lora_in_memory_limit": shared.OptionInfo(0, "Number of Lora networks to keep cached in memory", gr.Number, {"precision": 0}), + "lora_not_found_warning_console": shared.OptionInfo(False, "Lora not found warning in console"), + "lora_not_found_gradio_warning": shared.OptionInfo(False, "Lora not found warning popup in webui"), })) From ba92135a2ba9e210ce5370715e2defcb43df70d1 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sat, 30 Dec 2023 12:11:09 +0300 Subject: [PATCH 295/342] add override_settings support for infotext API --- modules/api/api.py | 10 ++++ modules/generation_parameters_copypaste.py | 66 ++++++++++++++-------- 2 files changed, 54 insertions(+), 22 deletions(-) diff --git a/modules/api/api.py b/modules/api/api.py index b3d709404..fb108486f 100644 --- a/modules/api/api.py +++ b/modules/api/api.py @@ -341,6 +341,7 @@ class Api: params = generation_parameters_copypaste.parse_generation_parameters(request.infotext) + handled_fields = {} for field in generation_parameters_copypaste.paste_fields[tabname]["fields"]: if not field.api: continue @@ -355,6 +356,15 @@ class Api: value = target_type(value) setattr(request, field.api, value) + handled_fields[field.label] = 1 + + if request.override_settings is None: + request.override_settings = {} + + overriden_settings = generation_parameters_copypaste.get_override_settings(params, skip_fields=handled_fields) + for infotext_text, setting_name, value in overriden_settings: + if setting_name not in request.override_settings: + request.override_settings[setting_name] = value return params diff --git a/modules/generation_parameters_copypaste.py b/modules/generation_parameters_copypaste.py index 4b4727c44..86a36c327 100644 --- a/modules/generation_parameters_copypaste.py +++ b/modules/generation_parameters_copypaste.py @@ -390,6 +390,48 @@ def create_override_settings_dict(text_pairs): return res +def get_override_settings(params, *, skip_fields=None): + """Returns a list of settings overrides from the infotext parameters dictionary. + + This function checks the `params` dictionary for any keys that correspond to settings in `shared.opts` and returns + a list of tuples containing the parameter name, setting name, and new value cast to correct type. + + It checks for conditions before adding an override: + - ignores settings that match the current value + - ignores parameter keys present in skip_fields argument. + + Example input: + {"Clip skip": "2"} + + Example output: + [("Clip skip", "CLIP_stop_at_last_layers", 2)] + """ + + res = [] + + mapping = [(info.infotext, k) for k, info in shared.opts.data_labels.items() if info.infotext] + for param_name, setting_name in mapping + infotext_to_setting_name_mapping: + if param_name in (skip_fields or {}): + continue + + v = params.get(param_name, None) + if v is None: + continue + + if setting_name == "sd_model_checkpoint" and shared.opts.disable_weights_auto_swap: + continue + + v = shared.opts.cast_value(setting_name, v) + current_value = getattr(shared.opts, setting_name, None) + + if v == current_value: + continue + + res.append((param_name, setting_name, v)) + + return res + + def connect_paste(button, paste_fields, input_comp, override_settings_component, tabname): def paste_func(prompt): if not prompt and not shared.cmd_opts.hide_ui_dir_config: @@ -431,29 +473,9 @@ def connect_paste(button, paste_fields, input_comp, override_settings_component, already_handled_fields = {key: 1 for _, key in paste_fields} def paste_settings(params): - vals = {} + vals = get_override_settings(params, skip_fields=already_handled_fields) - mapping = [(info.infotext, k) for k, info in shared.opts.data_labels.items() if info.infotext] - for param_name, setting_name in mapping + infotext_to_setting_name_mapping: - if param_name in already_handled_fields: - continue - - v = params.get(param_name, None) - if v is None: - continue - - if setting_name == "sd_model_checkpoint" and shared.opts.disable_weights_auto_swap: - continue - - v = shared.opts.cast_value(setting_name, v) - current_value = getattr(shared.opts, setting_name, None) - - if v == current_value: - continue - - vals[param_name] = v - - vals_pairs = [f"{k}: {v}" for k, v in vals.items()] + vals_pairs = [f"{infotext_text}: {value}" for infotext_text, setting_name, value in vals] return gr.Dropdown.update(value=vals_pairs, choices=vals_pairs, visible=bool(vals_pairs)) From 8b08b78c03f09898455d54cf099225ed5f8de1ee Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sat, 30 Dec 2023 12:27:23 +0300 Subject: [PATCH 296/342] make it so that if an option from infotext conflicts with an argument from API, the latter overrides the former --- modules/api/api.py | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/modules/api/api.py b/modules/api/api.py index fb108486f..cabccb4c0 100644 --- a/modules/api/api.py +++ b/modules/api/api.py @@ -339,6 +339,7 @@ class Api: if not request.infotext: return {} + set_fields = request.model_dump(exclude_unset=True) if hasattr(request, "request") else request.dict(exclude_unset=True) # pydantic v1/v2 have differenrt names for this params = generation_parameters_copypaste.parse_generation_parameters(request.infotext) handled_fields = {} @@ -346,6 +347,9 @@ class Api: if not field.api: continue + if field.api in set_fields: + continue + value = field.function(params) if field.function else params.get(field.label) target_type = request.__fields__[field.api].type_ @@ -376,7 +380,7 @@ class Api: script_runner.initialize_scripts(False) ui.create_ui() - infotext_params = self.apply_infotext(txt2imgreq, "txt2img") + self.apply_infotext(txt2imgreq, "txt2img") if not self.default_script_arg_txt2img: self.default_script_arg_txt2img = self.init_default_script_args(script_runner) From 0aacd4c72b4008d7153e747301fe8c5ffca57f85 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sat, 30 Dec 2023 13:33:18 +0300 Subject: [PATCH 297/342] add support for alwayson scripts for infotext API --- modules/api/api.py | 61 +++++++++++++++++++++++++++++++--------------- 1 file changed, 42 insertions(+), 19 deletions(-) diff --git a/modules/api/api.py b/modules/api/api.py index cabccb4c0..946cfe4a9 100644 --- a/modules/api/api.py +++ b/modules/api/api.py @@ -312,8 +312,13 @@ class Api: script_args[script.args_from:script.args_to] = ui_default_values return script_args - def init_script_args(self, request, default_script_args, selectable_scripts, selectable_idx, script_runner): + def init_script_args(self, request, default_script_args, selectable_scripts, selectable_idx, script_runner, *, input_script_args=None): script_args = default_script_args.copy() + + if input_script_args is not None: + for index, value in input_script_args.items(): + script_args[index] = value + # position 0 in script_arg is the idx+1 of the selectable script that is going to be run when using scripts.scripts_*2img.run() if selectable_scripts: script_args[selectable_scripts.args_from:selectable_scripts.args_to] = request.script_args @@ -335,41 +340,58 @@ class Api: script_args[alwayson_script.args_from + idx] = request.alwayson_scripts[alwayson_script_name]["args"][idx] return script_args - def apply_infotext(self, request, tabname): + def apply_infotext(self, request, tabname, *, script_runner=None, mentioned_script_args=None): if not request.infotext: return {} + possible_fields = generation_parameters_copypaste.paste_fields[tabname]["fields"] set_fields = request.model_dump(exclude_unset=True) if hasattr(request, "request") else request.dict(exclude_unset=True) # pydantic v1/v2 have differenrt names for this params = generation_parameters_copypaste.parse_generation_parameters(request.infotext) - handled_fields = {} - for field in generation_parameters_copypaste.paste_fields[tabname]["fields"]: + def get_field_value(field, params): + value = field.function(params) if field.function else params.get(field.label) + if value is None: + return None + + if field.api in request.__fields__: + target_type = request.__fields__[field.api].type_ + else: + target_type = type(field.component.value) + + if target_type == type(None): + return None + + if not isinstance(value, target_type): + value = target_type(value) + + return value + + for field in possible_fields: if not field.api: continue if field.api in set_fields: continue - value = field.function(params) if field.function else params.get(field.label) - target_type = request.__fields__[field.api].type_ - - if value is None: - continue - - if not isinstance(value, target_type): - value = target_type(value) - - setattr(request, field.api, value) - handled_fields[field.label] = 1 + value = get_field_value(field, params) + if value is not None: + setattr(request, field.api, value) if request.override_settings is None: request.override_settings = {} - overriden_settings = generation_parameters_copypaste.get_override_settings(params, skip_fields=handled_fields) - for infotext_text, setting_name, value in overriden_settings: + overriden_settings = generation_parameters_copypaste.get_override_settings(params) + for _, setting_name, value in overriden_settings: if setting_name not in request.override_settings: request.override_settings[setting_name] = value + if script_runner is not None and mentioned_script_args is not None: + indexes = {v: i for i, v in enumerate(script_runner.inputs)} + script_fields = ((field, indexes[field.component]) for field in possible_fields if field.component in indexes) + + for field, index in script_fields: + mentioned_script_args[index] = get_field_value(field, params) + return params def text2imgapi(self, txt2imgreq: models.StableDiffusionTxt2ImgProcessingAPI): @@ -380,7 +402,8 @@ class Api: script_runner.initialize_scripts(False) ui.create_ui() - self.apply_infotext(txt2imgreq, "txt2img") + infotext_script_args = {} + self.apply_infotext(txt2imgreq, "txt2img", script_runner=script_runner, mentioned_script_args=infotext_script_args) if not self.default_script_arg_txt2img: self.default_script_arg_txt2img = self.init_default_script_args(script_runner) @@ -400,7 +423,7 @@ class Api: args.pop('alwayson_scripts', None) args.pop('infotext', None) - script_args = self.init_script_args(txt2imgreq, self.default_script_arg_txt2img, selectable_scripts, selectable_script_idx, script_runner) + script_args = self.init_script_args(txt2imgreq, self.default_script_arg_txt2img, selectable_scripts, selectable_script_idx, script_runner, input_script_args=infotext_script_args) send_images = args.pop('send_images', True) args.pop('save_images', None) From 11a435b4697c2d735a117f31944c4ebe59c2504c Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sat, 30 Dec 2023 13:34:46 +0300 Subject: [PATCH 298/342] img2img support for infotext API --- modules/api/api.py | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/modules/api/api.py b/modules/api/api.py index 946cfe4a9..2c8dc2a0f 100644 --- a/modules/api/api.py +++ b/modules/api/api.py @@ -470,6 +470,10 @@ class Api: if not script_runner.scripts: script_runner.initialize_scripts(True) ui.create_ui() + + infotext_script_args = {} + self.apply_infotext(img2imgreq, "img2img", script_runner=script_runner, mentioned_script_args=infotext_script_args) + if not self.default_script_arg_img2img: self.default_script_arg_img2img = self.init_default_script_args(script_runner) selectable_scripts, selectable_script_idx = self.get_selectable_script(img2imgreq.script_name, script_runner) @@ -489,7 +493,7 @@ class Api: args.pop('script_args', None) # will refeed them to the pipeline directly after initializing them args.pop('alwayson_scripts', None) - script_args = self.init_script_args(img2imgreq, self.default_script_arg_img2img, selectable_scripts, selectable_script_idx, script_runner) + script_args = self.init_script_args(img2imgreq, self.default_script_arg_img2img, selectable_scripts, selectable_script_idx, script_runner, input_script_args=infotext_script_args) send_images = args.pop('send_images', True) args.pop('save_images', None) From 8f1826375943718463cec3af97a37886249bdb44 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sat, 30 Dec 2023 13:48:25 +0300 Subject: [PATCH 299/342] fix bad values read from infotext for API, add comment --- modules/api/api.py | 19 +++++++++++++++++-- 1 file changed, 17 insertions(+), 2 deletions(-) diff --git a/modules/api/api.py b/modules/api/api.py index 2c8dc2a0f..2918f7857 100644 --- a/modules/api/api.py +++ b/modules/api/api.py @@ -341,6 +341,13 @@ class Api: return script_args def apply_infotext(self, request, tabname, *, script_runner=None, mentioned_script_args=None): + """Processes `infotext` field from the `request`, and sets other fields of the `request` accoring to what's in infotext. + + If request already has a field set, and that field is encountered in infotext too, the value from infotext is ignored. + + Additionally, fills `mentioned_script_args` dict with index: value pairs for script arguments read from infotext. + """ + if not request.infotext: return {} @@ -361,7 +368,10 @@ class Api: if target_type == type(None): return None - if not isinstance(value, target_type): + if isinstance(value, dict) and value.get('__type__') == 'generic_update': # this is a gradio.update rather than a value + value = value.get('value') + + if value is not None and not isinstance(value, target_type): value = target_type(value) return value @@ -390,7 +400,12 @@ class Api: script_fields = ((field, indexes[field.component]) for field in possible_fields if field.component in indexes) for field, index in script_fields: - mentioned_script_args[index] = get_field_value(field, params) + value = get_field_value(field, params) + + if value is None: + continue + + mentioned_script_args[index] = value return params From f0e2e8b930115012976f7c5bae00e243a7ebbf79 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sat, 30 Dec 2023 15:12:48 +0300 Subject: [PATCH 300/342] update #14354 --- webui.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/webui.py b/webui.py index 3f5b67a30..2c417168a 100644 --- a/webui.py +++ b/webui.py @@ -39,7 +39,7 @@ def api_only(): print(f"Startup time: {startup_timer.summary()}.") api.launch( - server_name=cmd_opts.server_name or ("0.0.0.0" if cmd_opts.listen else "127.0.0.1"), + server_name=initialize_util.gradio_server_name(), port=cmd_opts.port if cmd_opts.port else 7861, root_path=f"/{cmd_opts.subpath}" if cmd_opts.subpath else "" ) From c069c2c5628728c9506dd034ef98e6335fd5bb34 Mon Sep 17 00:00:00 2001 From: lanyeeee <1210347077@qq.com> Date: Sat, 30 Dec 2023 21:32:22 +0800 Subject: [PATCH 301/342] add locks to ensure init args are thread-safe --- modules/api/api.py | 24 ++++++++++++++++-------- 1 file changed, 16 insertions(+), 8 deletions(-) diff --git a/modules/api/api.py b/modules/api/api.py index f0a68c672..45c5c5077 100644 --- a/modules/api/api.py +++ b/modules/api/api.py @@ -251,14 +251,10 @@ class Api: self.default_script_arg_txt2img = [] self.default_script_arg_img2img = [] - script_runner = scripts.scripts_img2img - if not script_runner.scripts: - script_runner.initialize_scripts(True) - ui.create_ui() - if not self.default_script_arg_txt2img: - self.default_script_arg_txt2img = self.init_default_script_args(script_runner) - if not self.default_script_arg_img2img: - self.default_script_arg_img2img = self.init_default_script_args(script_runner) + self.txt2img_script_arg_init_lock = Lock() + self.img2img_script_arg_init_lock = Lock() + + def add_api_route(self, path: str, endpoint, **kwargs): if shared.cmd_opts.api_auth: @@ -348,6 +344,12 @@ class Api: task_id = txt2imgreq.force_task_id or create_task_id("txt2img") script_runner = scripts.scripts_txt2img + with self.txt2img_script_arg_init_lock: + if not script_runner.scripts: + script_runner.initialize_scripts(False) + ui.create_ui() + if not self.default_script_arg_txt2img: + self.default_script_arg_txt2img = self.init_default_script_args(script_runner) selectable_scripts, selectable_script_idx = self.get_selectable_script(txt2imgreq.script_name, script_runner) populate = txt2imgreq.copy(update={ # Override __init__ params @@ -407,6 +409,12 @@ class Api: mask = decode_base64_to_image(mask) script_runner = scripts.scripts_img2img + with self.img2img_script_arg_init_lock: + if not script_runner.scripts: + script_runner.initialize_scripts(True) + ui.create_ui() + if not self.default_script_arg_img2img: + self.default_script_arg_img2img = self.init_default_script_args(script_runner) selectable_scripts, selectable_script_idx = self.get_selectable_script(img2imgreq.script_name, script_runner) populate = img2imgreq.copy(update={ # Override __init__ params From 31992eff9b9714c158b12cec16dfe66c76270dfa Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sat, 30 Dec 2023 16:51:02 +0300 Subject: [PATCH 302/342] make it possible again to extract styles that have whitespace at the end. --- modules/styles.py | 47 +++++++++++++++++------------------------------ 1 file changed, 17 insertions(+), 30 deletions(-) diff --git a/modules/styles.py b/modules/styles.py index 81d9800d1..026c43001 100644 --- a/modules/styles.py +++ b/modules/styles.py @@ -30,38 +30,29 @@ def apply_styles_to_prompt(prompt, styles): return prompt -def unwrap_style_text_from_prompt(style_text, prompt): - """ - Checks the prompt to see if the style text is wrapped around it. If so, - returns True plus the prompt text without the style text. Otherwise, returns - False with the original prompt. +def extract_style_text_from_prompt(style_text, prompt): + """This function extracts the text from a given prompt based on a provided style text. It checks if the style text contains the placeholder {prompt} or if it appears at the end of the prompt. If a match is found, it returns True along with the extracted text. Otherwise, it returns False and the original prompt. - Note that the "cleaned" version of the style text is only used for matching - purposes here. It isn't returned; the original style text is not modified. + extract_style_text_from_prompt("masterpiece", "1girl, art by greg, masterpiece") outputs (True, "1girl, art by greg") + extract_style_text_from_prompt("masterpiece, {prompt}", "masterpiece, 1girl, art by greg") outputs (True, "1girl, art by greg") + extract_style_text_from_prompt("masterpiece, {prompt}", "exquisite, 1girl, art by greg") outputs (False, "exquisite, 1girl, art by greg") """ - stripped_prompt = prompt - stripped_style_text = style_text + + stripped_prompt = prompt.strip() + stripped_style_text = style_text.strip() + if "{prompt}" in stripped_style_text: - # Work out whether the prompt is wrapped in the style text. If so, we - # return True and the "inner" prompt text that isn't part of the style. - try: - left, right = stripped_style_text.split("{prompt}", 2) - except ValueError as e: - # If the style text has multple "{prompt}"s, we can't split it into - # two parts. This is an error, but we can't do anything about it. - print(f"Unable to compare style text to prompt:\n{style_text}") - print(f"Error: {e}") - return False, prompt + left, right = stripped_style_text.split("{prompt}", 2) if stripped_prompt.startswith(left) and stripped_prompt.endswith(right): - prompt = stripped_prompt[len(left) : len(stripped_prompt) - len(right)] + prompt = stripped_prompt[len(left):len(stripped_prompt)-len(right)] return True, prompt else: - # Work out whether the given prompt ends with the style text. If so, we - # return True and the prompt text up to where the style text starts. if stripped_prompt.endswith(stripped_style_text): - prompt = stripped_prompt[: len(stripped_prompt) - len(stripped_style_text)] - if prompt.endswith(", "): + prompt = stripped_prompt[:len(stripped_prompt)-len(stripped_style_text)] + + if prompt.endswith(', '): prompt = prompt[:-2] + return True, prompt return False, prompt @@ -76,15 +67,11 @@ def extract_original_prompts(style: PromptStyle, prompt, negative_prompt): if not style.prompt and not style.negative_prompt: return False, prompt, negative_prompt - match_positive, extracted_positive = unwrap_style_text_from_prompt( - style.prompt, prompt - ) + match_positive, extracted_positive = extract_style_text_from_prompt(style.prompt, prompt) if not match_positive: return False, prompt, negative_prompt - match_negative, extracted_negative = unwrap_style_text_from_prompt( - style.negative_prompt, negative_prompt - ) + match_negative, extracted_negative = extract_style_text_from_prompt(style.negative_prompt, negative_prompt) if not match_negative: return False, prompt, negative_prompt From 7aa27b000a3087dcb5cc7254600064bf70cacd3e Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Mon, 25 Dec 2023 14:44:15 +0200 Subject: [PATCH 303/342] Add types to split_grid --- modules/images.py | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/modules/images.py b/modules/images.py index 16f9ae7cc..d30e8865d 100644 --- a/modules/images.py +++ b/modules/images.py @@ -64,9 +64,8 @@ def image_grid(imgs, batch_size=1, rows=None): Grid = namedtuple("Grid", ["tiles", "tile_w", "tile_h", "image_w", "image_h", "overlap"]) -def split_grid(image, tile_w=512, tile_h=512, overlap=64): - w = image.width - h = image.height +def split_grid(image: Image.Image, tile_w: int = 512, tile_h: int = 512, overlap: int = 64) -> Grid: + w, h = image.size non_overlap_width = tile_w - overlap non_overlap_height = tile_h - overlap From 12c6f37f8e4b1d1d643c9d8d5dfc763c3203c728 Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Wed, 27 Dec 2023 11:01:45 +0200 Subject: [PATCH 304/342] Add tile_count property to Grid --- modules/images.py | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/modules/images.py b/modules/images.py index d30e8865d..87a7bf221 100644 --- a/modules/images.py +++ b/modules/images.py @@ -61,7 +61,13 @@ def image_grid(imgs, batch_size=1, rows=None): return grid -Grid = namedtuple("Grid", ["tiles", "tile_w", "tile_h", "image_w", "image_h", "overlap"]) +class Grid(namedtuple("_Grid", ["tiles", "tile_w", "tile_h", "image_w", "image_h", "overlap"])): + @property + def tile_count(self) -> int: + """ + The total number of tiles in the grid. + """ + return sum(len(row[2]) for row in self.tiles) def split_grid(image: Image.Image, tile_w: int = 512, tile_h: int = 512, overlap: int = 64) -> Grid: From e472383acbb9e07dca311abe5fb16ee2675e410a Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Wed, 27 Dec 2023 11:04:33 +0200 Subject: [PATCH 305/342] Refactor esrgan_upscale to more generic upscale_with_model --- modules/esrgan_model.py | 47 +++++----------------------- modules/upscaler_utils.py | 66 +++++++++++++++++++++++++++++++++++++++ 2 files changed, 74 insertions(+), 39 deletions(-) create mode 100644 modules/upscaler_utils.py diff --git a/modules/esrgan_model.py b/modules/esrgan_model.py index 02a1727d2..c0d22a992 100644 --- a/modules/esrgan_model.py +++ b/modules/esrgan_model.py @@ -1,13 +1,12 @@ import sys -import numpy as np import torch -from PIL import Image import modules.esrgan_model_arch as arch -from modules import modelloader, images, devices +from modules import modelloader, devices from modules.shared import opts from modules.upscaler import Upscaler, UpscalerData +from modules.upscaler_utils import upscale_with_model def mod2normal(state_dict): @@ -190,40 +189,10 @@ class UpscalerESRGAN(Upscaler): return model -def upscale_without_tiling(model, img): - img = np.array(img) - img = img[:, :, ::-1] - img = np.ascontiguousarray(np.transpose(img, (2, 0, 1))) / 255 - img = torch.from_numpy(img).float() - img = img.unsqueeze(0).to(devices.device_esrgan) - with torch.no_grad(): - output = model(img) - output = output.squeeze().float().cpu().clamp_(0, 1).numpy() - output = 255. * np.moveaxis(output, 0, 2) - output = output.astype(np.uint8) - output = output[:, :, ::-1] - return Image.fromarray(output, 'RGB') - - def esrgan_upscale(model, img): - if opts.ESRGAN_tile == 0: - return upscale_without_tiling(model, img) - - grid = images.split_grid(img, opts.ESRGAN_tile, opts.ESRGAN_tile, opts.ESRGAN_tile_overlap) - newtiles = [] - scale_factor = 1 - - for y, h, row in grid.tiles: - newrow = [] - for tiledata in row: - x, w, tile = tiledata - - output = upscale_without_tiling(model, tile) - scale_factor = output.width // tile.width - - newrow.append([x * scale_factor, w * scale_factor, output]) - newtiles.append([y * scale_factor, h * scale_factor, newrow]) - - newgrid = images.Grid(newtiles, grid.tile_w * scale_factor, grid.tile_h * scale_factor, grid.image_w * scale_factor, grid.image_h * scale_factor, grid.overlap * scale_factor) - output = images.combine_grid(newgrid) - return output + return upscale_with_model( + model, + img, + tile_size=opts.ESRGAN_tile, + tile_overlap=opts.ESRGAN_tile_overlap, + ) diff --git a/modules/upscaler_utils.py b/modules/upscaler_utils.py new file mode 100644 index 000000000..8bdda51c4 --- /dev/null +++ b/modules/upscaler_utils.py @@ -0,0 +1,66 @@ +import logging +from typing import Callable + +import numpy as np +import torch +import tqdm +from PIL import Image + +from modules import devices, images + +logger = logging.getLogger(__name__) + + +def upscale_without_tiling(model, img: Image.Image): + img = np.array(img) + img = img[:, :, ::-1] + img = np.ascontiguousarray(np.transpose(img, (2, 0, 1))) / 255 + img = torch.from_numpy(img).float() + img = img.unsqueeze(0).to(devices.device_esrgan) + with torch.no_grad(): + output = model(img) + output = output.squeeze().float().cpu().clamp_(0, 1).numpy() + output = 255. * np.moveaxis(output, 0, 2) + output = output.astype(np.uint8) + output = output[:, :, ::-1] + return Image.fromarray(output, 'RGB') + + +def upscale_with_model( + model: Callable[[torch.Tensor], torch.Tensor], + img: Image.Image, + *, + tile_size: int, + tile_overlap: int = 0, + desc="tiled upscale", +) -> Image.Image: + if tile_size <= 0: + logger.debug("Upscaling %s without tiling", img) + output = upscale_without_tiling(model, img) + logger.debug("=> %s", output) + return output + + grid = images.split_grid(img, tile_size, tile_size, tile_overlap) + newtiles = [] + + with tqdm.tqdm(total=grid.tile_count, desc=desc) as p: + for y, h, row in grid.tiles: + newrow = [] + for x, w, tile in row: + logger.debug("Tile (%d, %d) %s...", x, y, tile) + output = upscale_without_tiling(model, tile) + scale_factor = output.width // tile.width + logger.debug("=> %s (scale factor %s)", output, scale_factor) + newrow.append([x * scale_factor, w * scale_factor, output]) + p.update(1) + newtiles.append([y * scale_factor, h * scale_factor, newrow]) + + newgrid = images.Grid( + newtiles, + tile_w=grid.tile_w * scale_factor, + tile_h=grid.tile_h * scale_factor, + image_w=grid.image_w * scale_factor, + image_h=grid.image_h * scale_factor, + overlap=grid.overlap * scale_factor, + ) + return images.combine_grid(newgrid) From b0f59342346b1c8b405f97c0e0bb01c6ae05c601 Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Mon, 25 Dec 2023 14:43:51 +0200 Subject: [PATCH 306/342] Use Spandrel for upscaling and face restoration architectures (aside from GFPGAN and LDSR) --- .../ScuNET/scripts/scunet_model.py | 13 +- .../ScuNET/scunet_model_arch.py | 268 ----- .../SwinIR/scripts/swinir_model.py | 126 +- .../SwinIR/swinir_model_arch.py | 867 -------------- .../SwinIR/swinir_model_arch_v2.py | 1017 ----------------- modules/codeformer/codeformer_arch.py | 276 ----- modules/codeformer/vqgan_arch.py | 435 ------- modules/codeformer_model.py | 223 ++-- modules/esrgan_model.py | 153 +-- modules/esrgan_model_arch.py | 465 -------- modules/gfpgan_model.py | 13 +- modules/launch_utils.py | 7 - modules/modelloader.py | 16 + modules/paths.py | 1 - modules/realesrgan_model.py | 153 +-- modules/sysinfo.py | 2 - modules/upscaler.py | 3 + requirements.txt | 3 +- requirements_versions.txt | 4 +- 19 files changed, 277 insertions(+), 3768 deletions(-) delete mode 100644 extensions-builtin/ScuNET/scunet_model_arch.py delete mode 100644 extensions-builtin/SwinIR/swinir_model_arch.py delete mode 100644 extensions-builtin/SwinIR/swinir_model_arch_v2.py delete mode 100644 modules/codeformer/codeformer_arch.py delete mode 100644 modules/codeformer/vqgan_arch.py delete mode 100644 modules/esrgan_model_arch.py diff --git a/extensions-builtin/ScuNET/scripts/scunet_model.py b/extensions-builtin/ScuNET/scripts/scunet_model.py index 167d2f64b..18cf8e1a0 100644 --- a/extensions-builtin/ScuNET/scripts/scunet_model.py +++ b/extensions-builtin/ScuNET/scripts/scunet_model.py @@ -7,9 +7,7 @@ from tqdm import tqdm import modules.upscaler from modules import devices, modelloader, script_callbacks, errors -from scunet_model_arch import SCUNet -from modules.modelloader import load_file_from_url from modules.shared import opts @@ -120,17 +118,10 @@ class UpscalerScuNET(modules.upscaler.Upscaler): device = devices.get_device_for('scunet') if path.startswith("http"): # TODO: this doesn't use `path` at all? - filename = load_file_from_url(self.model_url, model_dir=self.model_download_path, file_name=f"{self.name}.pth") + filename = modelloader.load_file_from_url(self.model_url, model_dir=self.model_download_path, file_name=f"{self.name}.pth") else: filename = path - model = SCUNet(in_nc=3, config=[4, 4, 4, 4, 4, 4, 4], dim=64) - model.load_state_dict(torch.load(filename), strict=True) - model.eval() - for _, v in model.named_parameters(): - v.requires_grad = False - model = model.to(device) - - return model + return modelloader.load_spandrel_model(filename, device=device) def on_ui_settings(): diff --git a/extensions-builtin/ScuNET/scunet_model_arch.py b/extensions-builtin/ScuNET/scunet_model_arch.py deleted file mode 100644 index b51a88062..000000000 --- a/extensions-builtin/ScuNET/scunet_model_arch.py +++ /dev/null @@ -1,268 +0,0 @@ -# -*- coding: utf-8 -*- -import numpy as np -import torch -import torch.nn as nn -from einops import rearrange -from einops.layers.torch import Rearrange -from timm.models.layers import trunc_normal_, DropPath - - -class WMSA(nn.Module): - """ Self-attention module in Swin Transformer - """ - - def __init__(self, input_dim, output_dim, head_dim, window_size, type): - super(WMSA, self).__init__() - self.input_dim = input_dim - self.output_dim = output_dim - self.head_dim = head_dim - self.scale = self.head_dim ** -0.5 - self.n_heads = input_dim // head_dim - self.window_size = window_size - self.type = type - self.embedding_layer = nn.Linear(self.input_dim, 3 * self.input_dim, bias=True) - - self.relative_position_params = nn.Parameter( - torch.zeros((2 * window_size - 1) * (2 * window_size - 1), self.n_heads)) - - self.linear = nn.Linear(self.input_dim, self.output_dim) - - trunc_normal_(self.relative_position_params, std=.02) - self.relative_position_params = torch.nn.Parameter( - self.relative_position_params.view(2 * window_size - 1, 2 * window_size - 1, self.n_heads).transpose(1, - 2).transpose( - 0, 1)) - - def generate_mask(self, h, w, p, shift): - """ generating the mask of SW-MSA - Args: - shift: shift parameters in CyclicShift. - Returns: - attn_mask: should be (1 1 w p p), - """ - # supporting square. - attn_mask = torch.zeros(h, w, p, p, p, p, dtype=torch.bool, device=self.relative_position_params.device) - if self.type == 'W': - return attn_mask - - s = p - shift - attn_mask[-1, :, :s, :, s:, :] = True - attn_mask[-1, :, s:, :, :s, :] = True - attn_mask[:, -1, :, :s, :, s:] = True - attn_mask[:, -1, :, s:, :, :s] = True - attn_mask = rearrange(attn_mask, 'w1 w2 p1 p2 p3 p4 -> 1 1 (w1 w2) (p1 p2) (p3 p4)') - return attn_mask - - def forward(self, x): - """ Forward pass of Window Multi-head Self-attention module. - Args: - x: input tensor with shape of [b h w c]; - attn_mask: attention mask, fill -inf where the value is True; - Returns: - output: tensor shape [b h w c] - """ - if self.type != 'W': - x = torch.roll(x, shifts=(-(self.window_size // 2), -(self.window_size // 2)), dims=(1, 2)) - - x = rearrange(x, 'b (w1 p1) (w2 p2) c -> b w1 w2 p1 p2 c', p1=self.window_size, p2=self.window_size) - h_windows = x.size(1) - w_windows = x.size(2) - # square validation - # assert h_windows == w_windows - - x = rearrange(x, 'b w1 w2 p1 p2 c -> b (w1 w2) (p1 p2) c', p1=self.window_size, p2=self.window_size) - qkv = self.embedding_layer(x) - q, k, v = rearrange(qkv, 'b nw np (threeh c) -> threeh b nw np c', c=self.head_dim).chunk(3, dim=0) - sim = torch.einsum('hbwpc,hbwqc->hbwpq', q, k) * self.scale - # Adding learnable relative embedding - sim = sim + rearrange(self.relative_embedding(), 'h p q -> h 1 1 p q') - # Using Attn Mask to distinguish different subwindows. - if self.type != 'W': - attn_mask = self.generate_mask(h_windows, w_windows, self.window_size, shift=self.window_size // 2) - sim = sim.masked_fill_(attn_mask, float("-inf")) - - probs = nn.functional.softmax(sim, dim=-1) - output = torch.einsum('hbwij,hbwjc->hbwic', probs, v) - output = rearrange(output, 'h b w p c -> b w p (h c)') - output = self.linear(output) - output = rearrange(output, 'b (w1 w2) (p1 p2) c -> b (w1 p1) (w2 p2) c', w1=h_windows, p1=self.window_size) - - if self.type != 'W': - output = torch.roll(output, shifts=(self.window_size // 2, self.window_size // 2), dims=(1, 2)) - - return output - - def relative_embedding(self): - cord = torch.tensor(np.array([[i, j] for i in range(self.window_size) for j in range(self.window_size)])) - relation = cord[:, None, :] - cord[None, :, :] + self.window_size - 1 - # negative is allowed - return self.relative_position_params[:, relation[:, :, 0].long(), relation[:, :, 1].long()] - - -class Block(nn.Module): - def __init__(self, input_dim, output_dim, head_dim, window_size, drop_path, type='W', input_resolution=None): - """ SwinTransformer Block - """ - super(Block, self).__init__() - self.input_dim = input_dim - self.output_dim = output_dim - assert type in ['W', 'SW'] - self.type = type - if input_resolution <= window_size: - self.type = 'W' - - self.ln1 = nn.LayerNorm(input_dim) - self.msa = WMSA(input_dim, input_dim, head_dim, window_size, self.type) - self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity() - self.ln2 = nn.LayerNorm(input_dim) - self.mlp = nn.Sequential( - nn.Linear(input_dim, 4 * input_dim), - nn.GELU(), - nn.Linear(4 * input_dim, output_dim), - ) - - def forward(self, x): - x = x + self.drop_path(self.msa(self.ln1(x))) - x = x + self.drop_path(self.mlp(self.ln2(x))) - return x - - -class ConvTransBlock(nn.Module): - def __init__(self, conv_dim, trans_dim, head_dim, window_size, drop_path, type='W', input_resolution=None): - """ SwinTransformer and Conv Block - """ - super(ConvTransBlock, self).__init__() - self.conv_dim = conv_dim - self.trans_dim = trans_dim - self.head_dim = head_dim - self.window_size = window_size - self.drop_path = drop_path - self.type = type - self.input_resolution = input_resolution - - assert self.type in ['W', 'SW'] - if self.input_resolution <= self.window_size: - self.type = 'W' - - self.trans_block = Block(self.trans_dim, self.trans_dim, self.head_dim, self.window_size, self.drop_path, - self.type, self.input_resolution) - self.conv1_1 = nn.Conv2d(self.conv_dim + self.trans_dim, self.conv_dim + self.trans_dim, 1, 1, 0, bias=True) - self.conv1_2 = nn.Conv2d(self.conv_dim + self.trans_dim, self.conv_dim + self.trans_dim, 1, 1, 0, bias=True) - - self.conv_block = nn.Sequential( - nn.Conv2d(self.conv_dim, self.conv_dim, 3, 1, 1, bias=False), - nn.ReLU(True), - nn.Conv2d(self.conv_dim, self.conv_dim, 3, 1, 1, bias=False) - ) - - def forward(self, x): - conv_x, trans_x = torch.split(self.conv1_1(x), (self.conv_dim, self.trans_dim), dim=1) - conv_x = self.conv_block(conv_x) + conv_x - trans_x = Rearrange('b c h w -> b h w c')(trans_x) - trans_x = self.trans_block(trans_x) - trans_x = Rearrange('b h w c -> b c h w')(trans_x) - res = self.conv1_2(torch.cat((conv_x, trans_x), dim=1)) - x = x + res - - return x - - -class SCUNet(nn.Module): - # def __init__(self, in_nc=3, config=[2, 2, 2, 2, 2, 2, 2], dim=64, drop_path_rate=0.0, input_resolution=256): - def __init__(self, in_nc=3, config=None, dim=64, drop_path_rate=0.0, input_resolution=256): - super(SCUNet, self).__init__() - if config is None: - config = [2, 2, 2, 2, 2, 2, 2] - self.config = config - self.dim = dim - self.head_dim = 32 - self.window_size = 8 - - # drop path rate for each layer - dpr = [x.item() for x in torch.linspace(0, drop_path_rate, sum(config))] - - self.m_head = [nn.Conv2d(in_nc, dim, 3, 1, 1, bias=False)] - - begin = 0 - self.m_down1 = [ConvTransBlock(dim // 2, dim // 2, self.head_dim, self.window_size, dpr[i + begin], - 'W' if not i % 2 else 'SW', input_resolution) - for i in range(config[0])] + \ - [nn.Conv2d(dim, 2 * dim, 2, 2, 0, bias=False)] - - begin += config[0] - self.m_down2 = [ConvTransBlock(dim, dim, self.head_dim, self.window_size, dpr[i + begin], - 'W' if not i % 2 else 'SW', input_resolution // 2) - for i in range(config[1])] + \ - [nn.Conv2d(2 * dim, 4 * dim, 2, 2, 0, bias=False)] - - begin += config[1] - self.m_down3 = [ConvTransBlock(2 * dim, 2 * dim, self.head_dim, self.window_size, dpr[i + begin], - 'W' if not i % 2 else 'SW', input_resolution // 4) - for i in range(config[2])] + \ - [nn.Conv2d(4 * dim, 8 * dim, 2, 2, 0, bias=False)] - - begin += config[2] - self.m_body = [ConvTransBlock(4 * dim, 4 * dim, self.head_dim, self.window_size, dpr[i + begin], - 'W' if not i % 2 else 'SW', input_resolution // 8) - for i in range(config[3])] - - begin += config[3] - self.m_up3 = [nn.ConvTranspose2d(8 * dim, 4 * dim, 2, 2, 0, bias=False), ] + \ - [ConvTransBlock(2 * dim, 2 * dim, self.head_dim, self.window_size, dpr[i + begin], - 'W' if not i % 2 else 'SW', input_resolution // 4) - for i in range(config[4])] - - begin += config[4] - self.m_up2 = [nn.ConvTranspose2d(4 * dim, 2 * dim, 2, 2, 0, bias=False), ] + \ - [ConvTransBlock(dim, dim, self.head_dim, self.window_size, dpr[i + begin], - 'W' if not i % 2 else 'SW', input_resolution // 2) - for i in range(config[5])] - - begin += config[5] - self.m_up1 = [nn.ConvTranspose2d(2 * dim, dim, 2, 2, 0, bias=False), ] + \ - [ConvTransBlock(dim // 2, dim // 2, self.head_dim, self.window_size, dpr[i + begin], - 'W' if not i % 2 else 'SW', input_resolution) - for i in range(config[6])] - - self.m_tail = [nn.Conv2d(dim, in_nc, 3, 1, 1, bias=False)] - - self.m_head = nn.Sequential(*self.m_head) - self.m_down1 = nn.Sequential(*self.m_down1) - self.m_down2 = nn.Sequential(*self.m_down2) - self.m_down3 = nn.Sequential(*self.m_down3) - self.m_body = nn.Sequential(*self.m_body) - self.m_up3 = nn.Sequential(*self.m_up3) - self.m_up2 = nn.Sequential(*self.m_up2) - self.m_up1 = nn.Sequential(*self.m_up1) - self.m_tail = nn.Sequential(*self.m_tail) - # self.apply(self._init_weights) - - def forward(self, x0): - - h, w = x0.size()[-2:] - paddingBottom = int(np.ceil(h / 64) * 64 - h) - paddingRight = int(np.ceil(w / 64) * 64 - w) - x0 = nn.ReplicationPad2d((0, paddingRight, 0, paddingBottom))(x0) - - x1 = self.m_head(x0) - x2 = self.m_down1(x1) - x3 = self.m_down2(x2) - x4 = self.m_down3(x3) - x = self.m_body(x4) - x = self.m_up3(x + x4) - x = self.m_up2(x + x3) - x = self.m_up1(x + x2) - x = self.m_tail(x + x1) - - x = x[..., :h, :w] - - return x - - def _init_weights(self, m): - if isinstance(m, nn.Linear): - trunc_normal_(m.weight, std=.02) - if m.bias is not None: - nn.init.constant_(m.bias, 0) - elif isinstance(m, nn.LayerNorm): - nn.init.constant_(m.bias, 0) - nn.init.constant_(m.weight, 1.0) diff --git a/extensions-builtin/SwinIR/scripts/swinir_model.py b/extensions-builtin/SwinIR/scripts/swinir_model.py index ae0d0e6a8..85c18b9e9 100644 --- a/extensions-builtin/SwinIR/scripts/swinir_model.py +++ b/extensions-builtin/SwinIR/scripts/swinir_model.py @@ -1,5 +1,5 @@ +import logging import sys -import platform import numpy as np import torch @@ -8,13 +8,11 @@ from tqdm import tqdm from modules import modelloader, devices, script_callbacks, shared from modules.shared import opts, state -from swinir_model_arch import SwinIR -from swinir_model_arch_v2 import Swin2SR from modules.upscaler import Upscaler, UpscalerData SWINIR_MODEL_URL = "https://github.com/JingyunLiang/SwinIR/releases/download/v0.0/003_realSR_BSRGAN_DFOWMFC_s64w8_SwinIR-L_x4_GAN.pth" -device_swinir = devices.get_device_for('swinir') +logger = logging.getLogger(__name__) class UpscalerSwinIR(Upscaler): @@ -37,26 +35,29 @@ class UpscalerSwinIR(Upscaler): scalers.append(model_data) self.scalers = scalers - def do_upscale(self, img, model_file): - use_compile = hasattr(opts, 'SWIN_torch_compile') and opts.SWIN_torch_compile \ - and int(torch.__version__.split('.')[0]) >= 2 and platform.system() != "Windows" + def do_upscale(self, img: Image.Image, model_file: str) -> Image.Image: current_config = (model_file, opts.SWIN_tile) - if use_compile and self._cached_model_config == current_config: + device = self._get_device() + + if self._cached_model_config == current_config: model = self._cached_model else: - self._cached_model = None try: model = self.load_model(model_file) except Exception as e: print(f"Failed loading SwinIR model {model_file}: {e}", file=sys.stderr) return img - model = model.to(device_swinir, dtype=devices.dtype) - if use_compile: - model = torch.compile(model) - self._cached_model = model - self._cached_model_config = current_config - img = upscale(img, model) + self._cached_model = model + self._cached_model_config = current_config + + img = upscale( + img, + model, + tile=opts.SWIN_tile, + tile_overlap=opts.SWIN_tile_overlap, + device=device, + ) devices.torch_gc() return img @@ -69,69 +70,54 @@ class UpscalerSwinIR(Upscaler): ) else: filename = path - if filename.endswith(".v2.pth"): - model = Swin2SR( - upscale=scale, - in_chans=3, - img_size=64, - window_size=8, - img_range=1.0, - depths=[6, 6, 6, 6, 6, 6], - embed_dim=180, - num_heads=[6, 6, 6, 6, 6, 6], - mlp_ratio=2, - upsampler="nearest+conv", - resi_connection="1conv", - ) - params = None - else: - model = SwinIR( - upscale=scale, - in_chans=3, - img_size=64, - window_size=8, - img_range=1.0, - depths=[6, 6, 6, 6, 6, 6, 6, 6, 6], - embed_dim=240, - num_heads=[8, 8, 8, 8, 8, 8, 8, 8, 8], - mlp_ratio=2, - upsampler="nearest+conv", - resi_connection="3conv", - ) - params = "params_ema" - pretrained_model = torch.load(filename) - if params is not None: - model.load_state_dict(pretrained_model[params], strict=True) - else: - model.load_state_dict(pretrained_model, strict=True) + model = modelloader.load_spandrel_model( + filename, + device=self._get_device(), + dtype=devices.dtype, + ) + if getattr(opts, 'SWIN_torch_compile', False): + try: + model = torch.compile(model) + except Exception: + logger.warning("Failed to compile SwinIR model, fallback to JIT", exc_info=True) return model + def _get_device(self): + return devices.get_device_for('swinir') + def upscale( - img, - model, - tile=None, - tile_overlap=None, - window_size=8, - scale=4, + img, + model, + *, + tile: int, + tile_overlap: int, + window_size=8, + scale=4, + device, ): - tile = tile or opts.SWIN_tile - tile_overlap = tile_overlap or opts.SWIN_tile_overlap - img = np.array(img) img = img[:, :, ::-1] img = np.moveaxis(img, 2, 0) / 255 img = torch.from_numpy(img).float() - img = img.unsqueeze(0).to(device_swinir, dtype=devices.dtype) + img = img.unsqueeze(0).to(device, dtype=devices.dtype) with torch.no_grad(), devices.autocast(): _, _, h_old, w_old = img.size() h_pad = (h_old // window_size + 1) * window_size - h_old w_pad = (w_old // window_size + 1) * window_size - w_old img = torch.cat([img, torch.flip(img, [2])], 2)[:, :, : h_old + h_pad, :] img = torch.cat([img, torch.flip(img, [3])], 3)[:, :, :, : w_old + w_pad] - output = inference(img, model, tile, tile_overlap, window_size, scale) + output = inference( + img, + model, + tile=tile, + tile_overlap=tile_overlap, + window_size=window_size, + scale=scale, + device=device, + ) output = output[..., : h_old * scale, : w_old * scale] output = output.data.squeeze().float().cpu().clamp_(0, 1).numpy() if output.ndim == 3: @@ -142,7 +128,16 @@ def upscale( return Image.fromarray(output, "RGB") -def inference(img, model, tile, tile_overlap, window_size, scale): +def inference( + img, + model, + *, + tile: int, + tile_overlap: int, + window_size: int, + scale: int, + device, +): # test the image tile by tile b, c, h, w = img.size() tile = min(tile, h, w) @@ -152,8 +147,8 @@ def inference(img, model, tile, tile_overlap, window_size, scale): stride = tile - tile_overlap h_idx_list = list(range(0, h - tile, stride)) + [h - tile] w_idx_list = list(range(0, w - tile, stride)) + [w - tile] - E = torch.zeros(b, c, h * sf, w * sf, dtype=devices.dtype, device=device_swinir).type_as(img) - W = torch.zeros_like(E, dtype=devices.dtype, device=device_swinir) + E = torch.zeros(b, c, h * sf, w * sf, dtype=devices.dtype, device=device).type_as(img) + W = torch.zeros_like(E, dtype=devices.dtype, device=device) with tqdm(total=len(h_idx_list) * len(w_idx_list), desc="SwinIR tiles") as pbar: for h_idx in h_idx_list: @@ -185,8 +180,7 @@ def on_ui_settings(): shared.opts.add_option("SWIN_tile", shared.OptionInfo(192, "Tile size for all SwinIR.", gr.Slider, {"minimum": 16, "maximum": 512, "step": 16}, section=('upscaling', "Upscaling"))) shared.opts.add_option("SWIN_tile_overlap", shared.OptionInfo(8, "Tile overlap, in pixels for SwinIR. Low values = visible seam.", gr.Slider, {"minimum": 0, "maximum": 48, "step": 1}, section=('upscaling', "Upscaling"))) - if int(torch.__version__.split('.')[0]) >= 2 and platform.system() != "Windows": # torch.compile() require pytorch 2.0 or above, and not on Windows - shared.opts.add_option("SWIN_torch_compile", shared.OptionInfo(False, "Use torch.compile to accelerate SwinIR.", gr.Checkbox, {"interactive": True}, section=('upscaling', "Upscaling")).info("Takes longer on first run")) + shared.opts.add_option("SWIN_torch_compile", shared.OptionInfo(False, "Use torch.compile to accelerate SwinIR.", gr.Checkbox, {"interactive": True}, section=('upscaling', "Upscaling")).info("Takes longer on first run")) script_callbacks.on_ui_settings(on_ui_settings) diff --git a/extensions-builtin/SwinIR/swinir_model_arch.py b/extensions-builtin/SwinIR/swinir_model_arch.py deleted file mode 100644 index 93b932747..000000000 --- a/extensions-builtin/SwinIR/swinir_model_arch.py +++ /dev/null @@ -1,867 +0,0 @@ -# ----------------------------------------------------------------------------------- -# SwinIR: Image Restoration Using Swin Transformer, https://arxiv.org/abs/2108.10257 -# Originally Written by Ze Liu, Modified by Jingyun Liang. -# ----------------------------------------------------------------------------------- - -import math -import torch -import torch.nn as nn -import torch.nn.functional as F -import torch.utils.checkpoint as checkpoint -from timm.models.layers import DropPath, to_2tuple, trunc_normal_ - - -class Mlp(nn.Module): - def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, drop=0.): - super().__init__() - out_features = out_features or in_features - hidden_features = hidden_features or in_features - self.fc1 = nn.Linear(in_features, hidden_features) - self.act = act_layer() - self.fc2 = nn.Linear(hidden_features, out_features) - self.drop = nn.Dropout(drop) - - def forward(self, x): - x = self.fc1(x) - x = self.act(x) - x = self.drop(x) - x = self.fc2(x) - x = self.drop(x) - return x - - -def window_partition(x, window_size): - """ - Args: - x: (B, H, W, C) - window_size (int): window size - - Returns: - windows: (num_windows*B, window_size, window_size, C) - """ - B, H, W, C = x.shape - x = x.view(B, H // window_size, window_size, W // window_size, window_size, C) - windows = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(-1, window_size, window_size, C) - return windows - - -def window_reverse(windows, window_size, H, W): - """ - Args: - windows: (num_windows*B, window_size, window_size, C) - window_size (int): Window size - H (int): Height of image - W (int): Width of image - - Returns: - x: (B, H, W, C) - """ - B = int(windows.shape[0] / (H * W / window_size / window_size)) - x = windows.view(B, H // window_size, W // window_size, window_size, window_size, -1) - x = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(B, H, W, -1) - return x - - -class WindowAttention(nn.Module): - r""" Window based multi-head self attention (W-MSA) module with relative position bias. - It supports both of shifted and non-shifted window. - - Args: - dim (int): Number of input channels. - window_size (tuple[int]): The height and width of the window. - num_heads (int): Number of attention heads. - qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True - qk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if set - attn_drop (float, optional): Dropout ratio of attention weight. Default: 0.0 - proj_drop (float, optional): Dropout ratio of output. Default: 0.0 - """ - - def __init__(self, dim, window_size, num_heads, qkv_bias=True, qk_scale=None, attn_drop=0., proj_drop=0.): - - super().__init__() - self.dim = dim - self.window_size = window_size # Wh, Ww - self.num_heads = num_heads - head_dim = dim // num_heads - self.scale = qk_scale or head_dim ** -0.5 - - # define a parameter table of relative position bias - self.relative_position_bias_table = nn.Parameter( - torch.zeros((2 * window_size[0] - 1) * (2 * window_size[1] - 1), num_heads)) # 2*Wh-1 * 2*Ww-1, nH - - # get pair-wise relative position index for each token inside the window - coords_h = torch.arange(self.window_size[0]) - coords_w = torch.arange(self.window_size[1]) - coords = torch.stack(torch.meshgrid([coords_h, coords_w])) # 2, Wh, Ww - coords_flatten = torch.flatten(coords, 1) # 2, Wh*Ww - relative_coords = coords_flatten[:, :, None] - coords_flatten[:, None, :] # 2, Wh*Ww, Wh*Ww - relative_coords = relative_coords.permute(1, 2, 0).contiguous() # Wh*Ww, Wh*Ww, 2 - relative_coords[:, :, 0] += self.window_size[0] - 1 # shift to start from 0 - relative_coords[:, :, 1] += self.window_size[1] - 1 - relative_coords[:, :, 0] *= 2 * self.window_size[1] - 1 - relative_position_index = relative_coords.sum(-1) # Wh*Ww, Wh*Ww - self.register_buffer("relative_position_index", relative_position_index) - - self.qkv = nn.Linear(dim, dim * 3, bias=qkv_bias) - self.attn_drop = nn.Dropout(attn_drop) - self.proj = nn.Linear(dim, dim) - - self.proj_drop = nn.Dropout(proj_drop) - - trunc_normal_(self.relative_position_bias_table, std=.02) - self.softmax = nn.Softmax(dim=-1) - - def forward(self, x, mask=None): - """ - Args: - x: input features with shape of (num_windows*B, N, C) - mask: (0/-inf) mask with shape of (num_windows, Wh*Ww, Wh*Ww) or None - """ - B_, N, C = x.shape - qkv = self.qkv(x).reshape(B_, N, 3, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4) - q, k, v = qkv[0], qkv[1], qkv[2] # make torchscript happy (cannot use tensor as tuple) - - q = q * self.scale - attn = (q @ k.transpose(-2, -1)) - - relative_position_bias = self.relative_position_bias_table[self.relative_position_index.view(-1)].view( - self.window_size[0] * self.window_size[1], self.window_size[0] * self.window_size[1], -1) # Wh*Ww,Wh*Ww,nH - relative_position_bias = relative_position_bias.permute(2, 0, 1).contiguous() # nH, Wh*Ww, Wh*Ww - attn = attn + relative_position_bias.unsqueeze(0) - - if mask is not None: - nW = mask.shape[0] - attn = attn.view(B_ // nW, nW, self.num_heads, N, N) + mask.unsqueeze(1).unsqueeze(0) - attn = attn.view(-1, self.num_heads, N, N) - attn = self.softmax(attn) - else: - attn = self.softmax(attn) - - attn = self.attn_drop(attn) - - x = (attn @ v).transpose(1, 2).reshape(B_, N, C) - x = self.proj(x) - x = self.proj_drop(x) - return x - - def extra_repr(self) -> str: - return f'dim={self.dim}, window_size={self.window_size}, num_heads={self.num_heads}' - - def flops(self, N): - # calculate flops for 1 window with token length of N - flops = 0 - # qkv = self.qkv(x) - flops += N * self.dim * 3 * self.dim - # attn = (q @ k.transpose(-2, -1)) - flops += self.num_heads * N * (self.dim // self.num_heads) * N - # x = (attn @ v) - flops += self.num_heads * N * N * (self.dim // self.num_heads) - # x = self.proj(x) - flops += N * self.dim * self.dim - return flops - - -class SwinTransformerBlock(nn.Module): - r""" Swin Transformer Block. - - Args: - dim (int): Number of input channels. - input_resolution (tuple[int]): Input resolution. - num_heads (int): Number of attention heads. - window_size (int): Window size. - shift_size (int): Shift size for SW-MSA. - mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. - qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True - qk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if set. - drop (float, optional): Dropout rate. Default: 0.0 - attn_drop (float, optional): Attention dropout rate. Default: 0.0 - drop_path (float, optional): Stochastic depth rate. Default: 0.0 - act_layer (nn.Module, optional): Activation layer. Default: nn.GELU - norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm - """ - - def __init__(self, dim, input_resolution, num_heads, window_size=7, shift_size=0, - mlp_ratio=4., qkv_bias=True, qk_scale=None, drop=0., attn_drop=0., drop_path=0., - act_layer=nn.GELU, norm_layer=nn.LayerNorm): - super().__init__() - self.dim = dim - self.input_resolution = input_resolution - self.num_heads = num_heads - self.window_size = window_size - self.shift_size = shift_size - self.mlp_ratio = mlp_ratio - if min(self.input_resolution) <= self.window_size: - # if window size is larger than input resolution, we don't partition windows - self.shift_size = 0 - self.window_size = min(self.input_resolution) - assert 0 <= self.shift_size < self.window_size, "shift_size must in 0-window_size" - - self.norm1 = norm_layer(dim) - self.attn = WindowAttention( - dim, window_size=to_2tuple(self.window_size), num_heads=num_heads, - qkv_bias=qkv_bias, qk_scale=qk_scale, attn_drop=attn_drop, proj_drop=drop) - - self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity() - self.norm2 = norm_layer(dim) - mlp_hidden_dim = int(dim * mlp_ratio) - self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim, act_layer=act_layer, drop=drop) - - if self.shift_size > 0: - attn_mask = self.calculate_mask(self.input_resolution) - else: - attn_mask = None - - self.register_buffer("attn_mask", attn_mask) - - def calculate_mask(self, x_size): - # calculate attention mask for SW-MSA - H, W = x_size - img_mask = torch.zeros((1, H, W, 1)) # 1 H W 1 - h_slices = (slice(0, -self.window_size), - slice(-self.window_size, -self.shift_size), - slice(-self.shift_size, None)) - w_slices = (slice(0, -self.window_size), - slice(-self.window_size, -self.shift_size), - slice(-self.shift_size, None)) - cnt = 0 - for h in h_slices: - for w in w_slices: - img_mask[:, h, w, :] = cnt - cnt += 1 - - mask_windows = window_partition(img_mask, self.window_size) # nW, window_size, window_size, 1 - mask_windows = mask_windows.view(-1, self.window_size * self.window_size) - attn_mask = mask_windows.unsqueeze(1) - mask_windows.unsqueeze(2) - attn_mask = attn_mask.masked_fill(attn_mask != 0, float(-100.0)).masked_fill(attn_mask == 0, float(0.0)) - - return attn_mask - - def forward(self, x, x_size): - H, W = x_size - B, L, C = x.shape - # assert L == H * W, "input feature has wrong size" - - shortcut = x - x = self.norm1(x) - x = x.view(B, H, W, C) - - # cyclic shift - if self.shift_size > 0: - shifted_x = torch.roll(x, shifts=(-self.shift_size, -self.shift_size), dims=(1, 2)) - else: - shifted_x = x - - # partition windows - x_windows = window_partition(shifted_x, self.window_size) # nW*B, window_size, window_size, C - x_windows = x_windows.view(-1, self.window_size * self.window_size, C) # nW*B, window_size*window_size, C - - # W-MSA/SW-MSA (to be compatible for testing on images whose shapes are the multiple of window size - if self.input_resolution == x_size: - attn_windows = self.attn(x_windows, mask=self.attn_mask) # nW*B, window_size*window_size, C - else: - attn_windows = self.attn(x_windows, mask=self.calculate_mask(x_size).to(x.device)) - - # merge windows - attn_windows = attn_windows.view(-1, self.window_size, self.window_size, C) - shifted_x = window_reverse(attn_windows, self.window_size, H, W) # B H' W' C - - # reverse cyclic shift - if self.shift_size > 0: - x = torch.roll(shifted_x, shifts=(self.shift_size, self.shift_size), dims=(1, 2)) - else: - x = shifted_x - x = x.view(B, H * W, C) - - # FFN - x = shortcut + self.drop_path(x) - x = x + self.drop_path(self.mlp(self.norm2(x))) - - return x - - def extra_repr(self) -> str: - return f"dim={self.dim}, input_resolution={self.input_resolution}, num_heads={self.num_heads}, " \ - f"window_size={self.window_size}, shift_size={self.shift_size}, mlp_ratio={self.mlp_ratio}" - - def flops(self): - flops = 0 - H, W = self.input_resolution - # norm1 - flops += self.dim * H * W - # W-MSA/SW-MSA - nW = H * W / self.window_size / self.window_size - flops += nW * self.attn.flops(self.window_size * self.window_size) - # mlp - flops += 2 * H * W * self.dim * self.dim * self.mlp_ratio - # norm2 - flops += self.dim * H * W - return flops - - -class PatchMerging(nn.Module): - r""" Patch Merging Layer. - - Args: - input_resolution (tuple[int]): Resolution of input feature. - dim (int): Number of input channels. - norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm - """ - - def __init__(self, input_resolution, dim, norm_layer=nn.LayerNorm): - super().__init__() - self.input_resolution = input_resolution - self.dim = dim - self.reduction = nn.Linear(4 * dim, 2 * dim, bias=False) - self.norm = norm_layer(4 * dim) - - def forward(self, x): - """ - x: B, H*W, C - """ - H, W = self.input_resolution - B, L, C = x.shape - assert L == H * W, "input feature has wrong size" - assert H % 2 == 0 and W % 2 == 0, f"x size ({H}*{W}) are not even." - - x = x.view(B, H, W, C) - - x0 = x[:, 0::2, 0::2, :] # B H/2 W/2 C - x1 = x[:, 1::2, 0::2, :] # B H/2 W/2 C - x2 = x[:, 0::2, 1::2, :] # B H/2 W/2 C - x3 = x[:, 1::2, 1::2, :] # B H/2 W/2 C - x = torch.cat([x0, x1, x2, x3], -1) # B H/2 W/2 4*C - x = x.view(B, -1, 4 * C) # B H/2*W/2 4*C - - x = self.norm(x) - x = self.reduction(x) - - return x - - def extra_repr(self) -> str: - return f"input_resolution={self.input_resolution}, dim={self.dim}" - - def flops(self): - H, W = self.input_resolution - flops = H * W * self.dim - flops += (H // 2) * (W // 2) * 4 * self.dim * 2 * self.dim - return flops - - -class BasicLayer(nn.Module): - """ A basic Swin Transformer layer for one stage. - - Args: - dim (int): Number of input channels. - input_resolution (tuple[int]): Input resolution. - depth (int): Number of blocks. - num_heads (int): Number of attention heads. - window_size (int): Local window size. - mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. - qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True - qk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if set. - drop (float, optional): Dropout rate. Default: 0.0 - attn_drop (float, optional): Attention dropout rate. Default: 0.0 - drop_path (float | tuple[float], optional): Stochastic depth rate. Default: 0.0 - norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm - downsample (nn.Module | None, optional): Downsample layer at the end of the layer. Default: None - use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False. - """ - - def __init__(self, dim, input_resolution, depth, num_heads, window_size, - mlp_ratio=4., qkv_bias=True, qk_scale=None, drop=0., attn_drop=0., - drop_path=0., norm_layer=nn.LayerNorm, downsample=None, use_checkpoint=False): - - super().__init__() - self.dim = dim - self.input_resolution = input_resolution - self.depth = depth - self.use_checkpoint = use_checkpoint - - # build blocks - self.blocks = nn.ModuleList([ - SwinTransformerBlock(dim=dim, input_resolution=input_resolution, - num_heads=num_heads, window_size=window_size, - shift_size=0 if (i % 2 == 0) else window_size // 2, - mlp_ratio=mlp_ratio, - qkv_bias=qkv_bias, qk_scale=qk_scale, - drop=drop, attn_drop=attn_drop, - drop_path=drop_path[i] if isinstance(drop_path, list) else drop_path, - norm_layer=norm_layer) - for i in range(depth)]) - - # patch merging layer - if downsample is not None: - self.downsample = downsample(input_resolution, dim=dim, norm_layer=norm_layer) - else: - self.downsample = None - - def forward(self, x, x_size): - for blk in self.blocks: - if self.use_checkpoint: - x = checkpoint.checkpoint(blk, x, x_size) - else: - x = blk(x, x_size) - if self.downsample is not None: - x = self.downsample(x) - return x - - def extra_repr(self) -> str: - return f"dim={self.dim}, input_resolution={self.input_resolution}, depth={self.depth}" - - def flops(self): - flops = 0 - for blk in self.blocks: - flops += blk.flops() - if self.downsample is not None: - flops += self.downsample.flops() - return flops - - -class RSTB(nn.Module): - """Residual Swin Transformer Block (RSTB). - - Args: - dim (int): Number of input channels. - input_resolution (tuple[int]): Input resolution. - depth (int): Number of blocks. - num_heads (int): Number of attention heads. - window_size (int): Local window size. - mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. - qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True - qk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if set. - drop (float, optional): Dropout rate. Default: 0.0 - attn_drop (float, optional): Attention dropout rate. Default: 0.0 - drop_path (float | tuple[float], optional): Stochastic depth rate. Default: 0.0 - norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm - downsample (nn.Module | None, optional): Downsample layer at the end of the layer. Default: None - use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False. - img_size: Input image size. - patch_size: Patch size. - resi_connection: The convolutional block before residual connection. - """ - - def __init__(self, dim, input_resolution, depth, num_heads, window_size, - mlp_ratio=4., qkv_bias=True, qk_scale=None, drop=0., attn_drop=0., - drop_path=0., norm_layer=nn.LayerNorm, downsample=None, use_checkpoint=False, - img_size=224, patch_size=4, resi_connection='1conv'): - super(RSTB, self).__init__() - - self.dim = dim - self.input_resolution = input_resolution - - self.residual_group = BasicLayer(dim=dim, - input_resolution=input_resolution, - depth=depth, - num_heads=num_heads, - window_size=window_size, - mlp_ratio=mlp_ratio, - qkv_bias=qkv_bias, qk_scale=qk_scale, - drop=drop, attn_drop=attn_drop, - drop_path=drop_path, - norm_layer=norm_layer, - downsample=downsample, - use_checkpoint=use_checkpoint) - - if resi_connection == '1conv': - self.conv = nn.Conv2d(dim, dim, 3, 1, 1) - elif resi_connection == '3conv': - # to save parameters and memory - self.conv = nn.Sequential(nn.Conv2d(dim, dim // 4, 3, 1, 1), nn.LeakyReLU(negative_slope=0.2, inplace=True), - nn.Conv2d(dim // 4, dim // 4, 1, 1, 0), - nn.LeakyReLU(negative_slope=0.2, inplace=True), - nn.Conv2d(dim // 4, dim, 3, 1, 1)) - - self.patch_embed = PatchEmbed( - img_size=img_size, patch_size=patch_size, in_chans=0, embed_dim=dim, - norm_layer=None) - - self.patch_unembed = PatchUnEmbed( - img_size=img_size, patch_size=patch_size, in_chans=0, embed_dim=dim, - norm_layer=None) - - def forward(self, x, x_size): - return self.patch_embed(self.conv(self.patch_unembed(self.residual_group(x, x_size), x_size))) + x - - def flops(self): - flops = 0 - flops += self.residual_group.flops() - H, W = self.input_resolution - flops += H * W * self.dim * self.dim * 9 - flops += self.patch_embed.flops() - flops += self.patch_unembed.flops() - - return flops - - -class PatchEmbed(nn.Module): - r""" Image to Patch Embedding - - Args: - img_size (int): Image size. Default: 224. - patch_size (int): Patch token size. Default: 4. - in_chans (int): Number of input image channels. Default: 3. - embed_dim (int): Number of linear projection output channels. Default: 96. - norm_layer (nn.Module, optional): Normalization layer. Default: None - """ - - def __init__(self, img_size=224, patch_size=4, in_chans=3, embed_dim=96, norm_layer=None): - super().__init__() - img_size = to_2tuple(img_size) - patch_size = to_2tuple(patch_size) - patches_resolution = [img_size[0] // patch_size[0], img_size[1] // patch_size[1]] - self.img_size = img_size - self.patch_size = patch_size - self.patches_resolution = patches_resolution - self.num_patches = patches_resolution[0] * patches_resolution[1] - - self.in_chans = in_chans - self.embed_dim = embed_dim - - if norm_layer is not None: - self.norm = norm_layer(embed_dim) - else: - self.norm = None - - def forward(self, x): - x = x.flatten(2).transpose(1, 2) # B Ph*Pw C - if self.norm is not None: - x = self.norm(x) - return x - - def flops(self): - flops = 0 - H, W = self.img_size - if self.norm is not None: - flops += H * W * self.embed_dim - return flops - - -class PatchUnEmbed(nn.Module): - r""" Image to Patch Unembedding - - Args: - img_size (int): Image size. Default: 224. - patch_size (int): Patch token size. Default: 4. - in_chans (int): Number of input image channels. Default: 3. - embed_dim (int): Number of linear projection output channels. Default: 96. - norm_layer (nn.Module, optional): Normalization layer. Default: None - """ - - def __init__(self, img_size=224, patch_size=4, in_chans=3, embed_dim=96, norm_layer=None): - super().__init__() - img_size = to_2tuple(img_size) - patch_size = to_2tuple(patch_size) - patches_resolution = [img_size[0] // patch_size[0], img_size[1] // patch_size[1]] - self.img_size = img_size - self.patch_size = patch_size - self.patches_resolution = patches_resolution - self.num_patches = patches_resolution[0] * patches_resolution[1] - - self.in_chans = in_chans - self.embed_dim = embed_dim - - def forward(self, x, x_size): - B, HW, C = x.shape - x = x.transpose(1, 2).view(B, self.embed_dim, x_size[0], x_size[1]) # B Ph*Pw C - return x - - def flops(self): - flops = 0 - return flops - - -class Upsample(nn.Sequential): - """Upsample module. - - Args: - scale (int): Scale factor. Supported scales: 2^n and 3. - num_feat (int): Channel number of intermediate features. - """ - - def __init__(self, scale, num_feat): - m = [] - if (scale & (scale - 1)) == 0: # scale = 2^n - for _ in range(int(math.log(scale, 2))): - m.append(nn.Conv2d(num_feat, 4 * num_feat, 3, 1, 1)) - m.append(nn.PixelShuffle(2)) - elif scale == 3: - m.append(nn.Conv2d(num_feat, 9 * num_feat, 3, 1, 1)) - m.append(nn.PixelShuffle(3)) - else: - raise ValueError(f'scale {scale} is not supported. ' 'Supported scales: 2^n and 3.') - super(Upsample, self).__init__(*m) - - -class UpsampleOneStep(nn.Sequential): - """UpsampleOneStep module (the difference with Upsample is that it always only has 1conv + 1pixelshuffle) - Used in lightweight SR to save parameters. - - Args: - scale (int): Scale factor. Supported scales: 2^n and 3. - num_feat (int): Channel number of intermediate features. - - """ - - def __init__(self, scale, num_feat, num_out_ch, input_resolution=None): - self.num_feat = num_feat - self.input_resolution = input_resolution - m = [] - m.append(nn.Conv2d(num_feat, (scale ** 2) * num_out_ch, 3, 1, 1)) - m.append(nn.PixelShuffle(scale)) - super(UpsampleOneStep, self).__init__(*m) - - def flops(self): - H, W = self.input_resolution - flops = H * W * self.num_feat * 3 * 9 - return flops - - -class SwinIR(nn.Module): - r""" SwinIR - A PyTorch impl of : `SwinIR: Image Restoration Using Swin Transformer`, based on Swin Transformer. - - Args: - img_size (int | tuple(int)): Input image size. Default 64 - patch_size (int | tuple(int)): Patch size. Default: 1 - in_chans (int): Number of input image channels. Default: 3 - embed_dim (int): Patch embedding dimension. Default: 96 - depths (tuple(int)): Depth of each Swin Transformer layer. - num_heads (tuple(int)): Number of attention heads in different layers. - window_size (int): Window size. Default: 7 - mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. Default: 4 - qkv_bias (bool): If True, add a learnable bias to query, key, value. Default: True - qk_scale (float): Override default qk scale of head_dim ** -0.5 if set. Default: None - drop_rate (float): Dropout rate. Default: 0 - attn_drop_rate (float): Attention dropout rate. Default: 0 - drop_path_rate (float): Stochastic depth rate. Default: 0.1 - norm_layer (nn.Module): Normalization layer. Default: nn.LayerNorm. - ape (bool): If True, add absolute position embedding to the patch embedding. Default: False - patch_norm (bool): If True, add normalization after patch embedding. Default: True - use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False - upscale: Upscale factor. 2/3/4/8 for image SR, 1 for denoising and compress artifact reduction - img_range: Image range. 1. or 255. - upsampler: The reconstruction reconstruction module. 'pixelshuffle'/'pixelshuffledirect'/'nearest+conv'/None - resi_connection: The convolutional block before residual connection. '1conv'/'3conv' - """ - - def __init__(self, img_size=64, patch_size=1, in_chans=3, - embed_dim=96, depths=(6, 6, 6, 6), num_heads=(6, 6, 6, 6), - window_size=7, mlp_ratio=4., qkv_bias=True, qk_scale=None, - drop_rate=0., attn_drop_rate=0., drop_path_rate=0.1, - norm_layer=nn.LayerNorm, ape=False, patch_norm=True, - use_checkpoint=False, upscale=2, img_range=1., upsampler='', resi_connection='1conv', - **kwargs): - super(SwinIR, self).__init__() - num_in_ch = in_chans - num_out_ch = in_chans - num_feat = 64 - self.img_range = img_range - if in_chans == 3: - rgb_mean = (0.4488, 0.4371, 0.4040) - self.mean = torch.Tensor(rgb_mean).view(1, 3, 1, 1) - else: - self.mean = torch.zeros(1, 1, 1, 1) - self.upscale = upscale - self.upsampler = upsampler - self.window_size = window_size - - ##################################################################################################### - ################################### 1, shallow feature extraction ################################### - self.conv_first = nn.Conv2d(num_in_ch, embed_dim, 3, 1, 1) - - ##################################################################################################### - ################################### 2, deep feature extraction ###################################### - self.num_layers = len(depths) - self.embed_dim = embed_dim - self.ape = ape - self.patch_norm = patch_norm - self.num_features = embed_dim - self.mlp_ratio = mlp_ratio - - # split image into non-overlapping patches - self.patch_embed = PatchEmbed( - img_size=img_size, patch_size=patch_size, in_chans=embed_dim, embed_dim=embed_dim, - norm_layer=norm_layer if self.patch_norm else None) - num_patches = self.patch_embed.num_patches - patches_resolution = self.patch_embed.patches_resolution - self.patches_resolution = patches_resolution - - # merge non-overlapping patches into image - self.patch_unembed = PatchUnEmbed( - img_size=img_size, patch_size=patch_size, in_chans=embed_dim, embed_dim=embed_dim, - norm_layer=norm_layer if self.patch_norm else None) - - # absolute position embedding - if self.ape: - self.absolute_pos_embed = nn.Parameter(torch.zeros(1, num_patches, embed_dim)) - trunc_normal_(self.absolute_pos_embed, std=.02) - - self.pos_drop = nn.Dropout(p=drop_rate) - - # stochastic depth - dpr = [x.item() for x in torch.linspace(0, drop_path_rate, sum(depths))] # stochastic depth decay rule - - # build Residual Swin Transformer blocks (RSTB) - self.layers = nn.ModuleList() - for i_layer in range(self.num_layers): - layer = RSTB(dim=embed_dim, - input_resolution=(patches_resolution[0], - patches_resolution[1]), - depth=depths[i_layer], - num_heads=num_heads[i_layer], - window_size=window_size, - mlp_ratio=self.mlp_ratio, - qkv_bias=qkv_bias, qk_scale=qk_scale, - drop=drop_rate, attn_drop=attn_drop_rate, - drop_path=dpr[sum(depths[:i_layer]):sum(depths[:i_layer + 1])], # no impact on SR results - norm_layer=norm_layer, - downsample=None, - use_checkpoint=use_checkpoint, - img_size=img_size, - patch_size=patch_size, - resi_connection=resi_connection - - ) - self.layers.append(layer) - self.norm = norm_layer(self.num_features) - - # build the last conv layer in deep feature extraction - if resi_connection == '1conv': - self.conv_after_body = nn.Conv2d(embed_dim, embed_dim, 3, 1, 1) - elif resi_connection == '3conv': - # to save parameters and memory - self.conv_after_body = nn.Sequential(nn.Conv2d(embed_dim, embed_dim // 4, 3, 1, 1), - nn.LeakyReLU(negative_slope=0.2, inplace=True), - nn.Conv2d(embed_dim // 4, embed_dim // 4, 1, 1, 0), - nn.LeakyReLU(negative_slope=0.2, inplace=True), - nn.Conv2d(embed_dim // 4, embed_dim, 3, 1, 1)) - - ##################################################################################################### - ################################ 3, high quality image reconstruction ################################ - if self.upsampler == 'pixelshuffle': - # for classical SR - self.conv_before_upsample = nn.Sequential(nn.Conv2d(embed_dim, num_feat, 3, 1, 1), - nn.LeakyReLU(inplace=True)) - self.upsample = Upsample(upscale, num_feat) - self.conv_last = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1) - elif self.upsampler == 'pixelshuffledirect': - # for lightweight SR (to save parameters) - self.upsample = UpsampleOneStep(upscale, embed_dim, num_out_ch, - (patches_resolution[0], patches_resolution[1])) - elif self.upsampler == 'nearest+conv': - # for real-world SR (less artifacts) - self.conv_before_upsample = nn.Sequential(nn.Conv2d(embed_dim, num_feat, 3, 1, 1), - nn.LeakyReLU(inplace=True)) - self.conv_up1 = nn.Conv2d(num_feat, num_feat, 3, 1, 1) - if self.upscale == 4: - self.conv_up2 = nn.Conv2d(num_feat, num_feat, 3, 1, 1) - self.conv_hr = nn.Conv2d(num_feat, num_feat, 3, 1, 1) - self.conv_last = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1) - self.lrelu = nn.LeakyReLU(negative_slope=0.2, inplace=True) - else: - # for image denoising and JPEG compression artifact reduction - self.conv_last = nn.Conv2d(embed_dim, num_out_ch, 3, 1, 1) - - self.apply(self._init_weights) - - def _init_weights(self, m): - if isinstance(m, nn.Linear): - trunc_normal_(m.weight, std=.02) - if isinstance(m, nn.Linear) and m.bias is not None: - nn.init.constant_(m.bias, 0) - elif isinstance(m, nn.LayerNorm): - nn.init.constant_(m.bias, 0) - nn.init.constant_(m.weight, 1.0) - - @torch.jit.ignore - def no_weight_decay(self): - return {'absolute_pos_embed'} - - @torch.jit.ignore - def no_weight_decay_keywords(self): - return {'relative_position_bias_table'} - - def check_image_size(self, x): - _, _, h, w = x.size() - mod_pad_h = (self.window_size - h % self.window_size) % self.window_size - mod_pad_w = (self.window_size - w % self.window_size) % self.window_size - x = F.pad(x, (0, mod_pad_w, 0, mod_pad_h), 'reflect') - return x - - def forward_features(self, x): - x_size = (x.shape[2], x.shape[3]) - x = self.patch_embed(x) - if self.ape: - x = x + self.absolute_pos_embed - x = self.pos_drop(x) - - for layer in self.layers: - x = layer(x, x_size) - - x = self.norm(x) # B L C - x = self.patch_unembed(x, x_size) - - return x - - def forward(self, x): - H, W = x.shape[2:] - x = self.check_image_size(x) - - self.mean = self.mean.type_as(x) - x = (x - self.mean) * self.img_range - - if self.upsampler == 'pixelshuffle': - # for classical SR - x = self.conv_first(x) - x = self.conv_after_body(self.forward_features(x)) + x - x = self.conv_before_upsample(x) - x = self.conv_last(self.upsample(x)) - elif self.upsampler == 'pixelshuffledirect': - # for lightweight SR - x = self.conv_first(x) - x = self.conv_after_body(self.forward_features(x)) + x - x = self.upsample(x) - elif self.upsampler == 'nearest+conv': - # for real-world SR - x = self.conv_first(x) - x = self.conv_after_body(self.forward_features(x)) + x - x = self.conv_before_upsample(x) - x = self.lrelu(self.conv_up1(torch.nn.functional.interpolate(x, scale_factor=2, mode='nearest'))) - if self.upscale == 4: - x = self.lrelu(self.conv_up2(torch.nn.functional.interpolate(x, scale_factor=2, mode='nearest'))) - x = self.conv_last(self.lrelu(self.conv_hr(x))) - else: - # for image denoising and JPEG compression artifact reduction - x_first = self.conv_first(x) - res = self.conv_after_body(self.forward_features(x_first)) + x_first - x = x + self.conv_last(res) - - x = x / self.img_range + self.mean - - return x[:, :, :H*self.upscale, :W*self.upscale] - - def flops(self): - flops = 0 - H, W = self.patches_resolution - flops += H * W * 3 * self.embed_dim * 9 - flops += self.patch_embed.flops() - for layer in self.layers: - flops += layer.flops() - flops += H * W * 3 * self.embed_dim * self.embed_dim - flops += self.upsample.flops() - return flops - - -if __name__ == '__main__': - upscale = 4 - window_size = 8 - height = (1024 // upscale // window_size + 1) * window_size - width = (720 // upscale // window_size + 1) * window_size - model = SwinIR(upscale=2, img_size=(height, width), - window_size=window_size, img_range=1., depths=[6, 6, 6, 6], - embed_dim=60, num_heads=[6, 6, 6, 6], mlp_ratio=2, upsampler='pixelshuffledirect') - print(model) - print(height, width, model.flops() / 1e9) - - x = torch.randn((1, 3, height, width)) - x = model(x) - print(x.shape) diff --git a/extensions-builtin/SwinIR/swinir_model_arch_v2.py b/extensions-builtin/SwinIR/swinir_model_arch_v2.py deleted file mode 100644 index dad22cca2..000000000 --- a/extensions-builtin/SwinIR/swinir_model_arch_v2.py +++ /dev/null @@ -1,1017 +0,0 @@ -# ----------------------------------------------------------------------------------- -# Swin2SR: Swin2SR: SwinV2 Transformer for Compressed Image Super-Resolution and Restoration, https://arxiv.org/abs/ -# Written by Conde and Choi et al. -# ----------------------------------------------------------------------------------- - -import math -import numpy as np -import torch -import torch.nn as nn -import torch.nn.functional as F -import torch.utils.checkpoint as checkpoint -from timm.models.layers import DropPath, to_2tuple, trunc_normal_ - - -class Mlp(nn.Module): - def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, drop=0.): - super().__init__() - out_features = out_features or in_features - hidden_features = hidden_features or in_features - self.fc1 = nn.Linear(in_features, hidden_features) - self.act = act_layer() - self.fc2 = nn.Linear(hidden_features, out_features) - self.drop = nn.Dropout(drop) - - def forward(self, x): - x = self.fc1(x) - x = self.act(x) - x = self.drop(x) - x = self.fc2(x) - x = self.drop(x) - return x - - -def window_partition(x, window_size): - """ - Args: - x: (B, H, W, C) - window_size (int): window size - Returns: - windows: (num_windows*B, window_size, window_size, C) - """ - B, H, W, C = x.shape - x = x.view(B, H // window_size, window_size, W // window_size, window_size, C) - windows = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(-1, window_size, window_size, C) - return windows - - -def window_reverse(windows, window_size, H, W): - """ - Args: - windows: (num_windows*B, window_size, window_size, C) - window_size (int): Window size - H (int): Height of image - W (int): Width of image - Returns: - x: (B, H, W, C) - """ - B = int(windows.shape[0] / (H * W / window_size / window_size)) - x = windows.view(B, H // window_size, W // window_size, window_size, window_size, -1) - x = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(B, H, W, -1) - return x - -class WindowAttention(nn.Module): - r""" Window based multi-head self attention (W-MSA) module with relative position bias. - It supports both of shifted and non-shifted window. - Args: - dim (int): Number of input channels. - window_size (tuple[int]): The height and width of the window. - num_heads (int): Number of attention heads. - qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True - attn_drop (float, optional): Dropout ratio of attention weight. Default: 0.0 - proj_drop (float, optional): Dropout ratio of output. Default: 0.0 - pretrained_window_size (tuple[int]): The height and width of the window in pre-training. - """ - - def __init__(self, dim, window_size, num_heads, qkv_bias=True, attn_drop=0., proj_drop=0., - pretrained_window_size=(0, 0)): - - super().__init__() - self.dim = dim - self.window_size = window_size # Wh, Ww - self.pretrained_window_size = pretrained_window_size - self.num_heads = num_heads - - self.logit_scale = nn.Parameter(torch.log(10 * torch.ones((num_heads, 1, 1))), requires_grad=True) - - # mlp to generate continuous relative position bias - self.cpb_mlp = nn.Sequential(nn.Linear(2, 512, bias=True), - nn.ReLU(inplace=True), - nn.Linear(512, num_heads, bias=False)) - - # get relative_coords_table - relative_coords_h = torch.arange(-(self.window_size[0] - 1), self.window_size[0], dtype=torch.float32) - relative_coords_w = torch.arange(-(self.window_size[1] - 1), self.window_size[1], dtype=torch.float32) - relative_coords_table = torch.stack( - torch.meshgrid([relative_coords_h, - relative_coords_w])).permute(1, 2, 0).contiguous().unsqueeze(0) # 1, 2*Wh-1, 2*Ww-1, 2 - if pretrained_window_size[0] > 0: - relative_coords_table[:, :, :, 0] /= (pretrained_window_size[0] - 1) - relative_coords_table[:, :, :, 1] /= (pretrained_window_size[1] - 1) - else: - relative_coords_table[:, :, :, 0] /= (self.window_size[0] - 1) - relative_coords_table[:, :, :, 1] /= (self.window_size[1] - 1) - relative_coords_table *= 8 # normalize to -8, 8 - relative_coords_table = torch.sign(relative_coords_table) * torch.log2( - torch.abs(relative_coords_table) + 1.0) / np.log2(8) - - self.register_buffer("relative_coords_table", relative_coords_table) - - # get pair-wise relative position index for each token inside the window - coords_h = torch.arange(self.window_size[0]) - coords_w = torch.arange(self.window_size[1]) - coords = torch.stack(torch.meshgrid([coords_h, coords_w])) # 2, Wh, Ww - coords_flatten = torch.flatten(coords, 1) # 2, Wh*Ww - relative_coords = coords_flatten[:, :, None] - coords_flatten[:, None, :] # 2, Wh*Ww, Wh*Ww - relative_coords = relative_coords.permute(1, 2, 0).contiguous() # Wh*Ww, Wh*Ww, 2 - relative_coords[:, :, 0] += self.window_size[0] - 1 # shift to start from 0 - relative_coords[:, :, 1] += self.window_size[1] - 1 - relative_coords[:, :, 0] *= 2 * self.window_size[1] - 1 - relative_position_index = relative_coords.sum(-1) # Wh*Ww, Wh*Ww - self.register_buffer("relative_position_index", relative_position_index) - - self.qkv = nn.Linear(dim, dim * 3, bias=False) - if qkv_bias: - self.q_bias = nn.Parameter(torch.zeros(dim)) - self.v_bias = nn.Parameter(torch.zeros(dim)) - else: - self.q_bias = None - self.v_bias = None - self.attn_drop = nn.Dropout(attn_drop) - self.proj = nn.Linear(dim, dim) - self.proj_drop = nn.Dropout(proj_drop) - self.softmax = nn.Softmax(dim=-1) - - def forward(self, x, mask=None): - """ - Args: - x: input features with shape of (num_windows*B, N, C) - mask: (0/-inf) mask with shape of (num_windows, Wh*Ww, Wh*Ww) or None - """ - B_, N, C = x.shape - qkv_bias = None - if self.q_bias is not None: - qkv_bias = torch.cat((self.q_bias, torch.zeros_like(self.v_bias, requires_grad=False), self.v_bias)) - qkv = F.linear(input=x, weight=self.qkv.weight, bias=qkv_bias) - qkv = qkv.reshape(B_, N, 3, self.num_heads, -1).permute(2, 0, 3, 1, 4) - q, k, v = qkv[0], qkv[1], qkv[2] # make torchscript happy (cannot use tensor as tuple) - - # cosine attention - attn = (F.normalize(q, dim=-1) @ F.normalize(k, dim=-1).transpose(-2, -1)) - logit_scale = torch.clamp(self.logit_scale, max=torch.log(torch.tensor(1. / 0.01)).to(self.logit_scale.device)).exp() - attn = attn * logit_scale - - relative_position_bias_table = self.cpb_mlp(self.relative_coords_table).view(-1, self.num_heads) - relative_position_bias = relative_position_bias_table[self.relative_position_index.view(-1)].view( - self.window_size[0] * self.window_size[1], self.window_size[0] * self.window_size[1], -1) # Wh*Ww,Wh*Ww,nH - relative_position_bias = relative_position_bias.permute(2, 0, 1).contiguous() # nH, Wh*Ww, Wh*Ww - relative_position_bias = 16 * torch.sigmoid(relative_position_bias) - attn = attn + relative_position_bias.unsqueeze(0) - - if mask is not None: - nW = mask.shape[0] - attn = attn.view(B_ // nW, nW, self.num_heads, N, N) + mask.unsqueeze(1).unsqueeze(0) - attn = attn.view(-1, self.num_heads, N, N) - attn = self.softmax(attn) - else: - attn = self.softmax(attn) - - attn = self.attn_drop(attn) - - x = (attn @ v).transpose(1, 2).reshape(B_, N, C) - x = self.proj(x) - x = self.proj_drop(x) - return x - - def extra_repr(self) -> str: - return f'dim={self.dim}, window_size={self.window_size}, ' \ - f'pretrained_window_size={self.pretrained_window_size}, num_heads={self.num_heads}' - - def flops(self, N): - # calculate flops for 1 window with token length of N - flops = 0 - # qkv = self.qkv(x) - flops += N * self.dim * 3 * self.dim - # attn = (q @ k.transpose(-2, -1)) - flops += self.num_heads * N * (self.dim // self.num_heads) * N - # x = (attn @ v) - flops += self.num_heads * N * N * (self.dim // self.num_heads) - # x = self.proj(x) - flops += N * self.dim * self.dim - return flops - -class SwinTransformerBlock(nn.Module): - r""" Swin Transformer Block. - Args: - dim (int): Number of input channels. - input_resolution (tuple[int]): Input resulotion. - num_heads (int): Number of attention heads. - window_size (int): Window size. - shift_size (int): Shift size for SW-MSA. - mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. - qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True - drop (float, optional): Dropout rate. Default: 0.0 - attn_drop (float, optional): Attention dropout rate. Default: 0.0 - drop_path (float, optional): Stochastic depth rate. Default: 0.0 - act_layer (nn.Module, optional): Activation layer. Default: nn.GELU - norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm - pretrained_window_size (int): Window size in pre-training. - """ - - def __init__(self, dim, input_resolution, num_heads, window_size=7, shift_size=0, - mlp_ratio=4., qkv_bias=True, drop=0., attn_drop=0., drop_path=0., - act_layer=nn.GELU, norm_layer=nn.LayerNorm, pretrained_window_size=0): - super().__init__() - self.dim = dim - self.input_resolution = input_resolution - self.num_heads = num_heads - self.window_size = window_size - self.shift_size = shift_size - self.mlp_ratio = mlp_ratio - if min(self.input_resolution) <= self.window_size: - # if window size is larger than input resolution, we don't partition windows - self.shift_size = 0 - self.window_size = min(self.input_resolution) - assert 0 <= self.shift_size < self.window_size, "shift_size must in 0-window_size" - - self.norm1 = norm_layer(dim) - self.attn = WindowAttention( - dim, window_size=to_2tuple(self.window_size), num_heads=num_heads, - qkv_bias=qkv_bias, attn_drop=attn_drop, proj_drop=drop, - pretrained_window_size=to_2tuple(pretrained_window_size)) - - self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity() - self.norm2 = norm_layer(dim) - mlp_hidden_dim = int(dim * mlp_ratio) - self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim, act_layer=act_layer, drop=drop) - - if self.shift_size > 0: - attn_mask = self.calculate_mask(self.input_resolution) - else: - attn_mask = None - - self.register_buffer("attn_mask", attn_mask) - - def calculate_mask(self, x_size): - # calculate attention mask for SW-MSA - H, W = x_size - img_mask = torch.zeros((1, H, W, 1)) # 1 H W 1 - h_slices = (slice(0, -self.window_size), - slice(-self.window_size, -self.shift_size), - slice(-self.shift_size, None)) - w_slices = (slice(0, -self.window_size), - slice(-self.window_size, -self.shift_size), - slice(-self.shift_size, None)) - cnt = 0 - for h in h_slices: - for w in w_slices: - img_mask[:, h, w, :] = cnt - cnt += 1 - - mask_windows = window_partition(img_mask, self.window_size) # nW, window_size, window_size, 1 - mask_windows = mask_windows.view(-1, self.window_size * self.window_size) - attn_mask = mask_windows.unsqueeze(1) - mask_windows.unsqueeze(2) - attn_mask = attn_mask.masked_fill(attn_mask != 0, float(-100.0)).masked_fill(attn_mask == 0, float(0.0)) - - return attn_mask - - def forward(self, x, x_size): - H, W = x_size - B, L, C = x.shape - #assert L == H * W, "input feature has wrong size" - - shortcut = x - x = x.view(B, H, W, C) - - # cyclic shift - if self.shift_size > 0: - shifted_x = torch.roll(x, shifts=(-self.shift_size, -self.shift_size), dims=(1, 2)) - else: - shifted_x = x - - # partition windows - x_windows = window_partition(shifted_x, self.window_size) # nW*B, window_size, window_size, C - x_windows = x_windows.view(-1, self.window_size * self.window_size, C) # nW*B, window_size*window_size, C - - # W-MSA/SW-MSA (to be compatible for testing on images whose shapes are the multiple of window size - if self.input_resolution == x_size: - attn_windows = self.attn(x_windows, mask=self.attn_mask) # nW*B, window_size*window_size, C - else: - attn_windows = self.attn(x_windows, mask=self.calculate_mask(x_size).to(x.device)) - - # merge windows - attn_windows = attn_windows.view(-1, self.window_size, self.window_size, C) - shifted_x = window_reverse(attn_windows, self.window_size, H, W) # B H' W' C - - # reverse cyclic shift - if self.shift_size > 0: - x = torch.roll(shifted_x, shifts=(self.shift_size, self.shift_size), dims=(1, 2)) - else: - x = shifted_x - x = x.view(B, H * W, C) - x = shortcut + self.drop_path(self.norm1(x)) - - # FFN - x = x + self.drop_path(self.norm2(self.mlp(x))) - - return x - - def extra_repr(self) -> str: - return f"dim={self.dim}, input_resolution={self.input_resolution}, num_heads={self.num_heads}, " \ - f"window_size={self.window_size}, shift_size={self.shift_size}, mlp_ratio={self.mlp_ratio}" - - def flops(self): - flops = 0 - H, W = self.input_resolution - # norm1 - flops += self.dim * H * W - # W-MSA/SW-MSA - nW = H * W / self.window_size / self.window_size - flops += nW * self.attn.flops(self.window_size * self.window_size) - # mlp - flops += 2 * H * W * self.dim * self.dim * self.mlp_ratio - # norm2 - flops += self.dim * H * W - return flops - -class PatchMerging(nn.Module): - r""" Patch Merging Layer. - Args: - input_resolution (tuple[int]): Resolution of input feature. - dim (int): Number of input channels. - norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm - """ - - def __init__(self, input_resolution, dim, norm_layer=nn.LayerNorm): - super().__init__() - self.input_resolution = input_resolution - self.dim = dim - self.reduction = nn.Linear(4 * dim, 2 * dim, bias=False) - self.norm = norm_layer(2 * dim) - - def forward(self, x): - """ - x: B, H*W, C - """ - H, W = self.input_resolution - B, L, C = x.shape - assert L == H * W, "input feature has wrong size" - assert H % 2 == 0 and W % 2 == 0, f"x size ({H}*{W}) are not even." - - x = x.view(B, H, W, C) - - x0 = x[:, 0::2, 0::2, :] # B H/2 W/2 C - x1 = x[:, 1::2, 0::2, :] # B H/2 W/2 C - x2 = x[:, 0::2, 1::2, :] # B H/2 W/2 C - x3 = x[:, 1::2, 1::2, :] # B H/2 W/2 C - x = torch.cat([x0, x1, x2, x3], -1) # B H/2 W/2 4*C - x = x.view(B, -1, 4 * C) # B H/2*W/2 4*C - - x = self.reduction(x) - x = self.norm(x) - - return x - - def extra_repr(self) -> str: - return f"input_resolution={self.input_resolution}, dim={self.dim}" - - def flops(self): - H, W = self.input_resolution - flops = (H // 2) * (W // 2) * 4 * self.dim * 2 * self.dim - flops += H * W * self.dim // 2 - return flops - -class BasicLayer(nn.Module): - """ A basic Swin Transformer layer for one stage. - Args: - dim (int): Number of input channels. - input_resolution (tuple[int]): Input resolution. - depth (int): Number of blocks. - num_heads (int): Number of attention heads. - window_size (int): Local window size. - mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. - qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True - drop (float, optional): Dropout rate. Default: 0.0 - attn_drop (float, optional): Attention dropout rate. Default: 0.0 - drop_path (float | tuple[float], optional): Stochastic depth rate. Default: 0.0 - norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm - downsample (nn.Module | None, optional): Downsample layer at the end of the layer. Default: None - use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False. - pretrained_window_size (int): Local window size in pre-training. - """ - - def __init__(self, dim, input_resolution, depth, num_heads, window_size, - mlp_ratio=4., qkv_bias=True, drop=0., attn_drop=0., - drop_path=0., norm_layer=nn.LayerNorm, downsample=None, use_checkpoint=False, - pretrained_window_size=0): - - super().__init__() - self.dim = dim - self.input_resolution = input_resolution - self.depth = depth - self.use_checkpoint = use_checkpoint - - # build blocks - self.blocks = nn.ModuleList([ - SwinTransformerBlock(dim=dim, input_resolution=input_resolution, - num_heads=num_heads, window_size=window_size, - shift_size=0 if (i % 2 == 0) else window_size // 2, - mlp_ratio=mlp_ratio, - qkv_bias=qkv_bias, - drop=drop, attn_drop=attn_drop, - drop_path=drop_path[i] if isinstance(drop_path, list) else drop_path, - norm_layer=norm_layer, - pretrained_window_size=pretrained_window_size) - for i in range(depth)]) - - # patch merging layer - if downsample is not None: - self.downsample = downsample(input_resolution, dim=dim, norm_layer=norm_layer) - else: - self.downsample = None - - def forward(self, x, x_size): - for blk in self.blocks: - if self.use_checkpoint: - x = checkpoint.checkpoint(blk, x, x_size) - else: - x = blk(x, x_size) - if self.downsample is not None: - x = self.downsample(x) - return x - - def extra_repr(self) -> str: - return f"dim={self.dim}, input_resolution={self.input_resolution}, depth={self.depth}" - - def flops(self): - flops = 0 - for blk in self.blocks: - flops += blk.flops() - if self.downsample is not None: - flops += self.downsample.flops() - return flops - - def _init_respostnorm(self): - for blk in self.blocks: - nn.init.constant_(blk.norm1.bias, 0) - nn.init.constant_(blk.norm1.weight, 0) - nn.init.constant_(blk.norm2.bias, 0) - nn.init.constant_(blk.norm2.weight, 0) - -class PatchEmbed(nn.Module): - r""" Image to Patch Embedding - Args: - img_size (int): Image size. Default: 224. - patch_size (int): Patch token size. Default: 4. - in_chans (int): Number of input image channels. Default: 3. - embed_dim (int): Number of linear projection output channels. Default: 96. - norm_layer (nn.Module, optional): Normalization layer. Default: None - """ - - def __init__(self, img_size=224, patch_size=4, in_chans=3, embed_dim=96, norm_layer=None): - super().__init__() - img_size = to_2tuple(img_size) - patch_size = to_2tuple(patch_size) - patches_resolution = [img_size[0] // patch_size[0], img_size[1] // patch_size[1]] - self.img_size = img_size - self.patch_size = patch_size - self.patches_resolution = patches_resolution - self.num_patches = patches_resolution[0] * patches_resolution[1] - - self.in_chans = in_chans - self.embed_dim = embed_dim - - self.proj = nn.Conv2d(in_chans, embed_dim, kernel_size=patch_size, stride=patch_size) - if norm_layer is not None: - self.norm = norm_layer(embed_dim) - else: - self.norm = None - - def forward(self, x): - B, C, H, W = x.shape - # FIXME look at relaxing size constraints - # assert H == self.img_size[0] and W == self.img_size[1], - # f"Input image size ({H}*{W}) doesn't match model ({self.img_size[0]}*{self.img_size[1]})." - x = self.proj(x).flatten(2).transpose(1, 2) # B Ph*Pw C - if self.norm is not None: - x = self.norm(x) - return x - - def flops(self): - Ho, Wo = self.patches_resolution - flops = Ho * Wo * self.embed_dim * self.in_chans * (self.patch_size[0] * self.patch_size[1]) - if self.norm is not None: - flops += Ho * Wo * self.embed_dim - return flops - -class RSTB(nn.Module): - """Residual Swin Transformer Block (RSTB). - - Args: - dim (int): Number of input channels. - input_resolution (tuple[int]): Input resolution. - depth (int): Number of blocks. - num_heads (int): Number of attention heads. - window_size (int): Local window size. - mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. - qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True - drop (float, optional): Dropout rate. Default: 0.0 - attn_drop (float, optional): Attention dropout rate. Default: 0.0 - drop_path (float | tuple[float], optional): Stochastic depth rate. Default: 0.0 - norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm - downsample (nn.Module | None, optional): Downsample layer at the end of the layer. Default: None - use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False. - img_size: Input image size. - patch_size: Patch size. - resi_connection: The convolutional block before residual connection. - """ - - def __init__(self, dim, input_resolution, depth, num_heads, window_size, - mlp_ratio=4., qkv_bias=True, drop=0., attn_drop=0., - drop_path=0., norm_layer=nn.LayerNorm, downsample=None, use_checkpoint=False, - img_size=224, patch_size=4, resi_connection='1conv'): - super(RSTB, self).__init__() - - self.dim = dim - self.input_resolution = input_resolution - - self.residual_group = BasicLayer(dim=dim, - input_resolution=input_resolution, - depth=depth, - num_heads=num_heads, - window_size=window_size, - mlp_ratio=mlp_ratio, - qkv_bias=qkv_bias, - drop=drop, attn_drop=attn_drop, - drop_path=drop_path, - norm_layer=norm_layer, - downsample=downsample, - use_checkpoint=use_checkpoint) - - if resi_connection == '1conv': - self.conv = nn.Conv2d(dim, dim, 3, 1, 1) - elif resi_connection == '3conv': - # to save parameters and memory - self.conv = nn.Sequential(nn.Conv2d(dim, dim // 4, 3, 1, 1), nn.LeakyReLU(negative_slope=0.2, inplace=True), - nn.Conv2d(dim // 4, dim // 4, 1, 1, 0), - nn.LeakyReLU(negative_slope=0.2, inplace=True), - nn.Conv2d(dim // 4, dim, 3, 1, 1)) - - self.patch_embed = PatchEmbed( - img_size=img_size, patch_size=patch_size, in_chans=dim, embed_dim=dim, - norm_layer=None) - - self.patch_unembed = PatchUnEmbed( - img_size=img_size, patch_size=patch_size, in_chans=dim, embed_dim=dim, - norm_layer=None) - - def forward(self, x, x_size): - return self.patch_embed(self.conv(self.patch_unembed(self.residual_group(x, x_size), x_size))) + x - - def flops(self): - flops = 0 - flops += self.residual_group.flops() - H, W = self.input_resolution - flops += H * W * self.dim * self.dim * 9 - flops += self.patch_embed.flops() - flops += self.patch_unembed.flops() - - return flops - -class PatchUnEmbed(nn.Module): - r""" Image to Patch Unembedding - - Args: - img_size (int): Image size. Default: 224. - patch_size (int): Patch token size. Default: 4. - in_chans (int): Number of input image channels. Default: 3. - embed_dim (int): Number of linear projection output channels. Default: 96. - norm_layer (nn.Module, optional): Normalization layer. Default: None - """ - - def __init__(self, img_size=224, patch_size=4, in_chans=3, embed_dim=96, norm_layer=None): - super().__init__() - img_size = to_2tuple(img_size) - patch_size = to_2tuple(patch_size) - patches_resolution = [img_size[0] // patch_size[0], img_size[1] // patch_size[1]] - self.img_size = img_size - self.patch_size = patch_size - self.patches_resolution = patches_resolution - self.num_patches = patches_resolution[0] * patches_resolution[1] - - self.in_chans = in_chans - self.embed_dim = embed_dim - - def forward(self, x, x_size): - B, HW, C = x.shape - x = x.transpose(1, 2).view(B, self.embed_dim, x_size[0], x_size[1]) # B Ph*Pw C - return x - - def flops(self): - flops = 0 - return flops - - -class Upsample(nn.Sequential): - """Upsample module. - - Args: - scale (int): Scale factor. Supported scales: 2^n and 3. - num_feat (int): Channel number of intermediate features. - """ - - def __init__(self, scale, num_feat): - m = [] - if (scale & (scale - 1)) == 0: # scale = 2^n - for _ in range(int(math.log(scale, 2))): - m.append(nn.Conv2d(num_feat, 4 * num_feat, 3, 1, 1)) - m.append(nn.PixelShuffle(2)) - elif scale == 3: - m.append(nn.Conv2d(num_feat, 9 * num_feat, 3, 1, 1)) - m.append(nn.PixelShuffle(3)) - else: - raise ValueError(f'scale {scale} is not supported. ' 'Supported scales: 2^n and 3.') - super(Upsample, self).__init__(*m) - -class Upsample_hf(nn.Sequential): - """Upsample module. - - Args: - scale (int): Scale factor. Supported scales: 2^n and 3. - num_feat (int): Channel number of intermediate features. - """ - - def __init__(self, scale, num_feat): - m = [] - if (scale & (scale - 1)) == 0: # scale = 2^n - for _ in range(int(math.log(scale, 2))): - m.append(nn.Conv2d(num_feat, 4 * num_feat, 3, 1, 1)) - m.append(nn.PixelShuffle(2)) - elif scale == 3: - m.append(nn.Conv2d(num_feat, 9 * num_feat, 3, 1, 1)) - m.append(nn.PixelShuffle(3)) - else: - raise ValueError(f'scale {scale} is not supported. ' 'Supported scales: 2^n and 3.') - super(Upsample_hf, self).__init__(*m) - - -class UpsampleOneStep(nn.Sequential): - """UpsampleOneStep module (the difference with Upsample is that it always only has 1conv + 1pixelshuffle) - Used in lightweight SR to save parameters. - - Args: - scale (int): Scale factor. Supported scales: 2^n and 3. - num_feat (int): Channel number of intermediate features. - - """ - - def __init__(self, scale, num_feat, num_out_ch, input_resolution=None): - self.num_feat = num_feat - self.input_resolution = input_resolution - m = [] - m.append(nn.Conv2d(num_feat, (scale ** 2) * num_out_ch, 3, 1, 1)) - m.append(nn.PixelShuffle(scale)) - super(UpsampleOneStep, self).__init__(*m) - - def flops(self): - H, W = self.input_resolution - flops = H * W * self.num_feat * 3 * 9 - return flops - - - -class Swin2SR(nn.Module): - r""" Swin2SR - A PyTorch impl of : `Swin2SR: SwinV2 Transformer for Compressed Image Super-Resolution and Restoration`. - - Args: - img_size (int | tuple(int)): Input image size. Default 64 - patch_size (int | tuple(int)): Patch size. Default: 1 - in_chans (int): Number of input image channels. Default: 3 - embed_dim (int): Patch embedding dimension. Default: 96 - depths (tuple(int)): Depth of each Swin Transformer layer. - num_heads (tuple(int)): Number of attention heads in different layers. - window_size (int): Window size. Default: 7 - mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. Default: 4 - qkv_bias (bool): If True, add a learnable bias to query, key, value. Default: True - drop_rate (float): Dropout rate. Default: 0 - attn_drop_rate (float): Attention dropout rate. Default: 0 - drop_path_rate (float): Stochastic depth rate. Default: 0.1 - norm_layer (nn.Module): Normalization layer. Default: nn.LayerNorm. - ape (bool): If True, add absolute position embedding to the patch embedding. Default: False - patch_norm (bool): If True, add normalization after patch embedding. Default: True - use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False - upscale: Upscale factor. 2/3/4/8 for image SR, 1 for denoising and compress artifact reduction - img_range: Image range. 1. or 255. - upsampler: The reconstruction reconstruction module. 'pixelshuffle'/'pixelshuffledirect'/'nearest+conv'/None - resi_connection: The convolutional block before residual connection. '1conv'/'3conv' - """ - - def __init__(self, img_size=64, patch_size=1, in_chans=3, - embed_dim=96, depths=(6, 6, 6, 6), num_heads=(6, 6, 6, 6), - window_size=7, mlp_ratio=4., qkv_bias=True, - drop_rate=0., attn_drop_rate=0., drop_path_rate=0.1, - norm_layer=nn.LayerNorm, ape=False, patch_norm=True, - use_checkpoint=False, upscale=2, img_range=1., upsampler='', resi_connection='1conv', - **kwargs): - super(Swin2SR, self).__init__() - num_in_ch = in_chans - num_out_ch = in_chans - num_feat = 64 - self.img_range = img_range - if in_chans == 3: - rgb_mean = (0.4488, 0.4371, 0.4040) - self.mean = torch.Tensor(rgb_mean).view(1, 3, 1, 1) - else: - self.mean = torch.zeros(1, 1, 1, 1) - self.upscale = upscale - self.upsampler = upsampler - self.window_size = window_size - - ##################################################################################################### - ################################### 1, shallow feature extraction ################################### - self.conv_first = nn.Conv2d(num_in_ch, embed_dim, 3, 1, 1) - - ##################################################################################################### - ################################### 2, deep feature extraction ###################################### - self.num_layers = len(depths) - self.embed_dim = embed_dim - self.ape = ape - self.patch_norm = patch_norm - self.num_features = embed_dim - self.mlp_ratio = mlp_ratio - - # split image into non-overlapping patches - self.patch_embed = PatchEmbed( - img_size=img_size, patch_size=patch_size, in_chans=embed_dim, embed_dim=embed_dim, - norm_layer=norm_layer if self.patch_norm else None) - num_patches = self.patch_embed.num_patches - patches_resolution = self.patch_embed.patches_resolution - self.patches_resolution = patches_resolution - - # merge non-overlapping patches into image - self.patch_unembed = PatchUnEmbed( - img_size=img_size, patch_size=patch_size, in_chans=embed_dim, embed_dim=embed_dim, - norm_layer=norm_layer if self.patch_norm else None) - - # absolute position embedding - if self.ape: - self.absolute_pos_embed = nn.Parameter(torch.zeros(1, num_patches, embed_dim)) - trunc_normal_(self.absolute_pos_embed, std=.02) - - self.pos_drop = nn.Dropout(p=drop_rate) - - # stochastic depth - dpr = [x.item() for x in torch.linspace(0, drop_path_rate, sum(depths))] # stochastic depth decay rule - - # build Residual Swin Transformer blocks (RSTB) - self.layers = nn.ModuleList() - for i_layer in range(self.num_layers): - layer = RSTB(dim=embed_dim, - input_resolution=(patches_resolution[0], - patches_resolution[1]), - depth=depths[i_layer], - num_heads=num_heads[i_layer], - window_size=window_size, - mlp_ratio=self.mlp_ratio, - qkv_bias=qkv_bias, - drop=drop_rate, attn_drop=attn_drop_rate, - drop_path=dpr[sum(depths[:i_layer]):sum(depths[:i_layer + 1])], # no impact on SR results - norm_layer=norm_layer, - downsample=None, - use_checkpoint=use_checkpoint, - img_size=img_size, - patch_size=patch_size, - resi_connection=resi_connection - - ) - self.layers.append(layer) - - if self.upsampler == 'pixelshuffle_hf': - self.layers_hf = nn.ModuleList() - for i_layer in range(self.num_layers): - layer = RSTB(dim=embed_dim, - input_resolution=(patches_resolution[0], - patches_resolution[1]), - depth=depths[i_layer], - num_heads=num_heads[i_layer], - window_size=window_size, - mlp_ratio=self.mlp_ratio, - qkv_bias=qkv_bias, - drop=drop_rate, attn_drop=attn_drop_rate, - drop_path=dpr[sum(depths[:i_layer]):sum(depths[:i_layer + 1])], # no impact on SR results - norm_layer=norm_layer, - downsample=None, - use_checkpoint=use_checkpoint, - img_size=img_size, - patch_size=patch_size, - resi_connection=resi_connection - - ) - self.layers_hf.append(layer) - - self.norm = norm_layer(self.num_features) - - # build the last conv layer in deep feature extraction - if resi_connection == '1conv': - self.conv_after_body = nn.Conv2d(embed_dim, embed_dim, 3, 1, 1) - elif resi_connection == '3conv': - # to save parameters and memory - self.conv_after_body = nn.Sequential(nn.Conv2d(embed_dim, embed_dim // 4, 3, 1, 1), - nn.LeakyReLU(negative_slope=0.2, inplace=True), - nn.Conv2d(embed_dim // 4, embed_dim // 4, 1, 1, 0), - nn.LeakyReLU(negative_slope=0.2, inplace=True), - nn.Conv2d(embed_dim // 4, embed_dim, 3, 1, 1)) - - ##################################################################################################### - ################################ 3, high quality image reconstruction ################################ - if self.upsampler == 'pixelshuffle': - # for classical SR - self.conv_before_upsample = nn.Sequential(nn.Conv2d(embed_dim, num_feat, 3, 1, 1), - nn.LeakyReLU(inplace=True)) - self.upsample = Upsample(upscale, num_feat) - self.conv_last = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1) - elif self.upsampler == 'pixelshuffle_aux': - self.conv_bicubic = nn.Conv2d(num_in_ch, num_feat, 3, 1, 1) - self.conv_before_upsample = nn.Sequential( - nn.Conv2d(embed_dim, num_feat, 3, 1, 1), - nn.LeakyReLU(inplace=True)) - self.conv_aux = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1) - self.conv_after_aux = nn.Sequential( - nn.Conv2d(3, num_feat, 3, 1, 1), - nn.LeakyReLU(inplace=True)) - self.upsample = Upsample(upscale, num_feat) - self.conv_last = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1) - - elif self.upsampler == 'pixelshuffle_hf': - self.conv_before_upsample = nn.Sequential(nn.Conv2d(embed_dim, num_feat, 3, 1, 1), - nn.LeakyReLU(inplace=True)) - self.upsample = Upsample(upscale, num_feat) - self.upsample_hf = Upsample_hf(upscale, num_feat) - self.conv_last = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1) - self.conv_first_hf = nn.Sequential(nn.Conv2d(num_feat, embed_dim, 3, 1, 1), - nn.LeakyReLU(inplace=True)) - self.conv_after_body_hf = nn.Conv2d(embed_dim, embed_dim, 3, 1, 1) - self.conv_before_upsample_hf = nn.Sequential( - nn.Conv2d(embed_dim, num_feat, 3, 1, 1), - nn.LeakyReLU(inplace=True)) - self.conv_last_hf = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1) - - elif self.upsampler == 'pixelshuffledirect': - # for lightweight SR (to save parameters) - self.upsample = UpsampleOneStep(upscale, embed_dim, num_out_ch, - (patches_resolution[0], patches_resolution[1])) - elif self.upsampler == 'nearest+conv': - # for real-world SR (less artifacts) - assert self.upscale == 4, 'only support x4 now.' - self.conv_before_upsample = nn.Sequential(nn.Conv2d(embed_dim, num_feat, 3, 1, 1), - nn.LeakyReLU(inplace=True)) - self.conv_up1 = nn.Conv2d(num_feat, num_feat, 3, 1, 1) - self.conv_up2 = nn.Conv2d(num_feat, num_feat, 3, 1, 1) - self.conv_hr = nn.Conv2d(num_feat, num_feat, 3, 1, 1) - self.conv_last = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1) - self.lrelu = nn.LeakyReLU(negative_slope=0.2, inplace=True) - else: - # for image denoising and JPEG compression artifact reduction - self.conv_last = nn.Conv2d(embed_dim, num_out_ch, 3, 1, 1) - - self.apply(self._init_weights) - - def _init_weights(self, m): - if isinstance(m, nn.Linear): - trunc_normal_(m.weight, std=.02) - if isinstance(m, nn.Linear) and m.bias is not None: - nn.init.constant_(m.bias, 0) - elif isinstance(m, nn.LayerNorm): - nn.init.constant_(m.bias, 0) - nn.init.constant_(m.weight, 1.0) - - @torch.jit.ignore - def no_weight_decay(self): - return {'absolute_pos_embed'} - - @torch.jit.ignore - def no_weight_decay_keywords(self): - return {'relative_position_bias_table'} - - def check_image_size(self, x): - _, _, h, w = x.size() - mod_pad_h = (self.window_size - h % self.window_size) % self.window_size - mod_pad_w = (self.window_size - w % self.window_size) % self.window_size - x = F.pad(x, (0, mod_pad_w, 0, mod_pad_h), 'reflect') - return x - - def forward_features(self, x): - x_size = (x.shape[2], x.shape[3]) - x = self.patch_embed(x) - if self.ape: - x = x + self.absolute_pos_embed - x = self.pos_drop(x) - - for layer in self.layers: - x = layer(x, x_size) - - x = self.norm(x) # B L C - x = self.patch_unembed(x, x_size) - - return x - - def forward_features_hf(self, x): - x_size = (x.shape[2], x.shape[3]) - x = self.patch_embed(x) - if self.ape: - x = x + self.absolute_pos_embed - x = self.pos_drop(x) - - for layer in self.layers_hf: - x = layer(x, x_size) - - x = self.norm(x) # B L C - x = self.patch_unembed(x, x_size) - - return x - - def forward(self, x): - H, W = x.shape[2:] - x = self.check_image_size(x) - - self.mean = self.mean.type_as(x) - x = (x - self.mean) * self.img_range - - if self.upsampler == 'pixelshuffle': - # for classical SR - x = self.conv_first(x) - x = self.conv_after_body(self.forward_features(x)) + x - x = self.conv_before_upsample(x) - x = self.conv_last(self.upsample(x)) - elif self.upsampler == 'pixelshuffle_aux': - bicubic = F.interpolate(x, size=(H * self.upscale, W * self.upscale), mode='bicubic', align_corners=False) - bicubic = self.conv_bicubic(bicubic) - x = self.conv_first(x) - x = self.conv_after_body(self.forward_features(x)) + x - x = self.conv_before_upsample(x) - aux = self.conv_aux(x) # b, 3, LR_H, LR_W - x = self.conv_after_aux(aux) - x = self.upsample(x)[:, :, :H * self.upscale, :W * self.upscale] + bicubic[:, :, :H * self.upscale, :W * self.upscale] - x = self.conv_last(x) - aux = aux / self.img_range + self.mean - elif self.upsampler == 'pixelshuffle_hf': - # for classical SR with HF - x = self.conv_first(x) - x = self.conv_after_body(self.forward_features(x)) + x - x_before = self.conv_before_upsample(x) - x_out = self.conv_last(self.upsample(x_before)) - - x_hf = self.conv_first_hf(x_before) - x_hf = self.conv_after_body_hf(self.forward_features_hf(x_hf)) + x_hf - x_hf = self.conv_before_upsample_hf(x_hf) - x_hf = self.conv_last_hf(self.upsample_hf(x_hf)) - x = x_out + x_hf - x_hf = x_hf / self.img_range + self.mean - - elif self.upsampler == 'pixelshuffledirect': - # for lightweight SR - x = self.conv_first(x) - x = self.conv_after_body(self.forward_features(x)) + x - x = self.upsample(x) - elif self.upsampler == 'nearest+conv': - # for real-world SR - x = self.conv_first(x) - x = self.conv_after_body(self.forward_features(x)) + x - x = self.conv_before_upsample(x) - x = self.lrelu(self.conv_up1(torch.nn.functional.interpolate(x, scale_factor=2, mode='nearest'))) - x = self.lrelu(self.conv_up2(torch.nn.functional.interpolate(x, scale_factor=2, mode='nearest'))) - x = self.conv_last(self.lrelu(self.conv_hr(x))) - else: - # for image denoising and JPEG compression artifact reduction - x_first = self.conv_first(x) - res = self.conv_after_body(self.forward_features(x_first)) + x_first - x = x + self.conv_last(res) - - x = x / self.img_range + self.mean - if self.upsampler == "pixelshuffle_aux": - return x[:, :, :H*self.upscale, :W*self.upscale], aux - - elif self.upsampler == "pixelshuffle_hf": - x_out = x_out / self.img_range + self.mean - return x_out[:, :, :H*self.upscale, :W*self.upscale], x[:, :, :H*self.upscale, :W*self.upscale], x_hf[:, :, :H*self.upscale, :W*self.upscale] - - else: - return x[:, :, :H*self.upscale, :W*self.upscale] - - def flops(self): - flops = 0 - H, W = self.patches_resolution - flops += H * W * 3 * self.embed_dim * 9 - flops += self.patch_embed.flops() - for layer in self.layers: - flops += layer.flops() - flops += H * W * 3 * self.embed_dim * self.embed_dim - flops += self.upsample.flops() - return flops - - -if __name__ == '__main__': - upscale = 4 - window_size = 8 - height = (1024 // upscale // window_size + 1) * window_size - width = (720 // upscale // window_size + 1) * window_size - model = Swin2SR(upscale=2, img_size=(height, width), - window_size=window_size, img_range=1., depths=[6, 6, 6, 6], - embed_dim=60, num_heads=[6, 6, 6, 6], mlp_ratio=2, upsampler='pixelshuffledirect') - print(model) - print(height, width, model.flops() / 1e9) - - x = torch.randn((1, 3, height, width)) - x = model(x) - print(x.shape) diff --git a/modules/codeformer/codeformer_arch.py b/modules/codeformer/codeformer_arch.py deleted file mode 100644 index 12db68142..000000000 --- a/modules/codeformer/codeformer_arch.py +++ /dev/null @@ -1,276 +0,0 @@ -# this file is copied from CodeFormer repository. Please see comment in modules/codeformer_model.py - -import math -import torch -from torch import nn, Tensor -import torch.nn.functional as F -from typing import Optional - -from modules.codeformer.vqgan_arch import VQAutoEncoder, ResBlock -from basicsr.utils.registry import ARCH_REGISTRY - -def calc_mean_std(feat, eps=1e-5): - """Calculate mean and std for adaptive_instance_normalization. - - Args: - feat (Tensor): 4D tensor. - eps (float): A small value added to the variance to avoid - divide-by-zero. Default: 1e-5. - """ - size = feat.size() - assert len(size) == 4, 'The input feature should be 4D tensor.' - b, c = size[:2] - feat_var = feat.view(b, c, -1).var(dim=2) + eps - feat_std = feat_var.sqrt().view(b, c, 1, 1) - feat_mean = feat.view(b, c, -1).mean(dim=2).view(b, c, 1, 1) - return feat_mean, feat_std - - -def adaptive_instance_normalization(content_feat, style_feat): - """Adaptive instance normalization. - - Adjust the reference features to have the similar color and illuminations - as those in the degradate features. - - Args: - content_feat (Tensor): The reference feature. - style_feat (Tensor): The degradate features. - """ - size = content_feat.size() - style_mean, style_std = calc_mean_std(style_feat) - content_mean, content_std = calc_mean_std(content_feat) - normalized_feat = (content_feat - content_mean.expand(size)) / content_std.expand(size) - return normalized_feat * style_std.expand(size) + style_mean.expand(size) - - -class PositionEmbeddingSine(nn.Module): - """ - This is a more standard version of the position embedding, very similar to the one - used by the Attention is all you need paper, generalized to work on images. - """ - - def __init__(self, num_pos_feats=64, temperature=10000, normalize=False, scale=None): - super().__init__() - self.num_pos_feats = num_pos_feats - self.temperature = temperature - self.normalize = normalize - if scale is not None and normalize is False: - raise ValueError("normalize should be True if scale is passed") - if scale is None: - scale = 2 * math.pi - self.scale = scale - - def forward(self, x, mask=None): - if mask is None: - mask = torch.zeros((x.size(0), x.size(2), x.size(3)), device=x.device, dtype=torch.bool) - not_mask = ~mask - y_embed = not_mask.cumsum(1, dtype=torch.float32) - x_embed = not_mask.cumsum(2, dtype=torch.float32) - if self.normalize: - eps = 1e-6 - y_embed = y_embed / (y_embed[:, -1:, :] + eps) * self.scale - x_embed = x_embed / (x_embed[:, :, -1:] + eps) * self.scale - - dim_t = torch.arange(self.num_pos_feats, dtype=torch.float32, device=x.device) - dim_t = self.temperature ** (2 * (dim_t // 2) / self.num_pos_feats) - - pos_x = x_embed[:, :, :, None] / dim_t - pos_y = y_embed[:, :, :, None] / dim_t - pos_x = torch.stack( - (pos_x[:, :, :, 0::2].sin(), pos_x[:, :, :, 1::2].cos()), dim=4 - ).flatten(3) - pos_y = torch.stack( - (pos_y[:, :, :, 0::2].sin(), pos_y[:, :, :, 1::2].cos()), dim=4 - ).flatten(3) - pos = torch.cat((pos_y, pos_x), dim=3).permute(0, 3, 1, 2) - return pos - -def _get_activation_fn(activation): - """Return an activation function given a string""" - if activation == "relu": - return F.relu - if activation == "gelu": - return F.gelu - if activation == "glu": - return F.glu - raise RuntimeError(F"activation should be relu/gelu, not {activation}.") - - -class TransformerSALayer(nn.Module): - def __init__(self, embed_dim, nhead=8, dim_mlp=2048, dropout=0.0, activation="gelu"): - super().__init__() - self.self_attn = nn.MultiheadAttention(embed_dim, nhead, dropout=dropout) - # Implementation of Feedforward model - MLP - self.linear1 = nn.Linear(embed_dim, dim_mlp) - self.dropout = nn.Dropout(dropout) - self.linear2 = nn.Linear(dim_mlp, embed_dim) - - self.norm1 = nn.LayerNorm(embed_dim) - self.norm2 = nn.LayerNorm(embed_dim) - self.dropout1 = nn.Dropout(dropout) - self.dropout2 = nn.Dropout(dropout) - - self.activation = _get_activation_fn(activation) - - def with_pos_embed(self, tensor, pos: Optional[Tensor]): - return tensor if pos is None else tensor + pos - - def forward(self, tgt, - tgt_mask: Optional[Tensor] = None, - tgt_key_padding_mask: Optional[Tensor] = None, - query_pos: Optional[Tensor] = None): - - # self attention - tgt2 = self.norm1(tgt) - q = k = self.with_pos_embed(tgt2, query_pos) - tgt2 = self.self_attn(q, k, value=tgt2, attn_mask=tgt_mask, - key_padding_mask=tgt_key_padding_mask)[0] - tgt = tgt + self.dropout1(tgt2) - - # ffn - tgt2 = self.norm2(tgt) - tgt2 = self.linear2(self.dropout(self.activation(self.linear1(tgt2)))) - tgt = tgt + self.dropout2(tgt2) - return tgt - -class Fuse_sft_block(nn.Module): - def __init__(self, in_ch, out_ch): - super().__init__() - self.encode_enc = ResBlock(2*in_ch, out_ch) - - self.scale = nn.Sequential( - nn.Conv2d(in_ch, out_ch, kernel_size=3, padding=1), - nn.LeakyReLU(0.2, True), - nn.Conv2d(out_ch, out_ch, kernel_size=3, padding=1)) - - self.shift = nn.Sequential( - nn.Conv2d(in_ch, out_ch, kernel_size=3, padding=1), - nn.LeakyReLU(0.2, True), - nn.Conv2d(out_ch, out_ch, kernel_size=3, padding=1)) - - def forward(self, enc_feat, dec_feat, w=1): - enc_feat = self.encode_enc(torch.cat([enc_feat, dec_feat], dim=1)) - scale = self.scale(enc_feat) - shift = self.shift(enc_feat) - residual = w * (dec_feat * scale + shift) - out = dec_feat + residual - return out - - -@ARCH_REGISTRY.register() -class CodeFormer(VQAutoEncoder): - def __init__(self, dim_embd=512, n_head=8, n_layers=9, - codebook_size=1024, latent_size=256, - connect_list=('32', '64', '128', '256'), - fix_modules=('quantize', 'generator')): - super(CodeFormer, self).__init__(512, 64, [1, 2, 2, 4, 4, 8], 'nearest',2, [16], codebook_size) - - if fix_modules is not None: - for module in fix_modules: - for param in getattr(self, module).parameters(): - param.requires_grad = False - - self.connect_list = connect_list - self.n_layers = n_layers - self.dim_embd = dim_embd - self.dim_mlp = dim_embd*2 - - self.position_emb = nn.Parameter(torch.zeros(latent_size, self.dim_embd)) - self.feat_emb = nn.Linear(256, self.dim_embd) - - # transformer - self.ft_layers = nn.Sequential(*[TransformerSALayer(embed_dim=dim_embd, nhead=n_head, dim_mlp=self.dim_mlp, dropout=0.0) - for _ in range(self.n_layers)]) - - # logits_predict head - self.idx_pred_layer = nn.Sequential( - nn.LayerNorm(dim_embd), - nn.Linear(dim_embd, codebook_size, bias=False)) - - self.channels = { - '16': 512, - '32': 256, - '64': 256, - '128': 128, - '256': 128, - '512': 64, - } - - # after second residual block for > 16, before attn layer for ==16 - self.fuse_encoder_block = {'512':2, '256':5, '128':8, '64':11, '32':14, '16':18} - # after first residual block for > 16, before attn layer for ==16 - self.fuse_generator_block = {'16':6, '32': 9, '64':12, '128':15, '256':18, '512':21} - - # fuse_convs_dict - self.fuse_convs_dict = nn.ModuleDict() - for f_size in self.connect_list: - in_ch = self.channels[f_size] - self.fuse_convs_dict[f_size] = Fuse_sft_block(in_ch, in_ch) - - def _init_weights(self, module): - if isinstance(module, (nn.Linear, nn.Embedding)): - module.weight.data.normal_(mean=0.0, std=0.02) - if isinstance(module, nn.Linear) and module.bias is not None: - module.bias.data.zero_() - elif isinstance(module, nn.LayerNorm): - module.bias.data.zero_() - module.weight.data.fill_(1.0) - - def forward(self, x, w=0, detach_16=True, code_only=False, adain=False): - # ################### Encoder ##################### - enc_feat_dict = {} - out_list = [self.fuse_encoder_block[f_size] for f_size in self.connect_list] - for i, block in enumerate(self.encoder.blocks): - x = block(x) - if i in out_list: - enc_feat_dict[str(x.shape[-1])] = x.clone() - - lq_feat = x - # ################# Transformer ################### - # quant_feat, codebook_loss, quant_stats = self.quantize(lq_feat) - pos_emb = self.position_emb.unsqueeze(1).repeat(1,x.shape[0],1) - # BCHW -> BC(HW) -> (HW)BC - feat_emb = self.feat_emb(lq_feat.flatten(2).permute(2,0,1)) - query_emb = feat_emb - # Transformer encoder - for layer in self.ft_layers: - query_emb = layer(query_emb, query_pos=pos_emb) - - # output logits - logits = self.idx_pred_layer(query_emb) # (hw)bn - logits = logits.permute(1,0,2) # (hw)bn -> b(hw)n - - if code_only: # for training stage II - # logits doesn't need softmax before cross_entropy loss - return logits, lq_feat - - # ################# Quantization ################### - # if self.training: - # quant_feat = torch.einsum('btn,nc->btc', [soft_one_hot, self.quantize.embedding.weight]) - # # b(hw)c -> bc(hw) -> bchw - # quant_feat = quant_feat.permute(0,2,1).view(lq_feat.shape) - # ------------ - soft_one_hot = F.softmax(logits, dim=2) - _, top_idx = torch.topk(soft_one_hot, 1, dim=2) - quant_feat = self.quantize.get_codebook_feat(top_idx, shape=[x.shape[0],16,16,256]) - # preserve gradients - # quant_feat = lq_feat + (quant_feat - lq_feat).detach() - - if detach_16: - quant_feat = quant_feat.detach() # for training stage III - if adain: - quant_feat = adaptive_instance_normalization(quant_feat, lq_feat) - - # ################## Generator #################### - x = quant_feat - fuse_list = [self.fuse_generator_block[f_size] for f_size in self.connect_list] - - for i, block in enumerate(self.generator.blocks): - x = block(x) - if i in fuse_list: # fuse after i-th block - f_size = str(x.shape[-1]) - if w>0: - x = self.fuse_convs_dict[f_size](enc_feat_dict[f_size].detach(), x, w) - out = x - # logits doesn't need softmax before cross_entropy loss - return out, logits, lq_feat diff --git a/modules/codeformer/vqgan_arch.py b/modules/codeformer/vqgan_arch.py deleted file mode 100644 index 09ee6660d..000000000 --- a/modules/codeformer/vqgan_arch.py +++ /dev/null @@ -1,435 +0,0 @@ -# this file is copied from CodeFormer repository. Please see comment in modules/codeformer_model.py - -''' -VQGAN code, adapted from the original created by the Unleashing Transformers authors: -https://github.com/samb-t/unleashing-transformers/blob/master/models/vqgan.py - -''' -import torch -import torch.nn as nn -import torch.nn.functional as F -from basicsr.utils import get_root_logger -from basicsr.utils.registry import ARCH_REGISTRY - -def normalize(in_channels): - return torch.nn.GroupNorm(num_groups=32, num_channels=in_channels, eps=1e-6, affine=True) - - -@torch.jit.script -def swish(x): - return x*torch.sigmoid(x) - - -# Define VQVAE classes -class VectorQuantizer(nn.Module): - def __init__(self, codebook_size, emb_dim, beta): - super(VectorQuantizer, self).__init__() - self.codebook_size = codebook_size # number of embeddings - self.emb_dim = emb_dim # dimension of embedding - self.beta = beta # commitment cost used in loss term, beta * ||z_e(x)-sg[e]||^2 - self.embedding = nn.Embedding(self.codebook_size, self.emb_dim) - self.embedding.weight.data.uniform_(-1.0 / self.codebook_size, 1.0 / self.codebook_size) - - def forward(self, z): - # reshape z -> (batch, height, width, channel) and flatten - z = z.permute(0, 2, 3, 1).contiguous() - z_flattened = z.view(-1, self.emb_dim) - - # distances from z to embeddings e_j (z - e)^2 = z^2 + e^2 - 2 e * z - d = (z_flattened ** 2).sum(dim=1, keepdim=True) + (self.embedding.weight**2).sum(1) - \ - 2 * torch.matmul(z_flattened, self.embedding.weight.t()) - - mean_distance = torch.mean(d) - # find closest encodings - # min_encoding_indices = torch.argmin(d, dim=1).unsqueeze(1) - min_encoding_scores, min_encoding_indices = torch.topk(d, 1, dim=1, largest=False) - # [0-1], higher score, higher confidence - min_encoding_scores = torch.exp(-min_encoding_scores/10) - - min_encodings = torch.zeros(min_encoding_indices.shape[0], self.codebook_size).to(z) - min_encodings.scatter_(1, min_encoding_indices, 1) - - # get quantized latent vectors - z_q = torch.matmul(min_encodings, self.embedding.weight).view(z.shape) - # compute loss for embedding - loss = torch.mean((z_q.detach()-z)**2) + self.beta * torch.mean((z_q - z.detach()) ** 2) - # preserve gradients - z_q = z + (z_q - z).detach() - - # perplexity - e_mean = torch.mean(min_encodings, dim=0) - perplexity = torch.exp(-torch.sum(e_mean * torch.log(e_mean + 1e-10))) - # reshape back to match original input shape - z_q = z_q.permute(0, 3, 1, 2).contiguous() - - return z_q, loss, { - "perplexity": perplexity, - "min_encodings": min_encodings, - "min_encoding_indices": min_encoding_indices, - "min_encoding_scores": min_encoding_scores, - "mean_distance": mean_distance - } - - def get_codebook_feat(self, indices, shape): - # input indices: batch*token_num -> (batch*token_num)*1 - # shape: batch, height, width, channel - indices = indices.view(-1,1) - min_encodings = torch.zeros(indices.shape[0], self.codebook_size).to(indices) - min_encodings.scatter_(1, indices, 1) - # get quantized latent vectors - z_q = torch.matmul(min_encodings.float(), self.embedding.weight) - - if shape is not None: # reshape back to match original input shape - z_q = z_q.view(shape).permute(0, 3, 1, 2).contiguous() - - return z_q - - -class GumbelQuantizer(nn.Module): - def __init__(self, codebook_size, emb_dim, num_hiddens, straight_through=False, kl_weight=5e-4, temp_init=1.0): - super().__init__() - self.codebook_size = codebook_size # number of embeddings - self.emb_dim = emb_dim # dimension of embedding - self.straight_through = straight_through - self.temperature = temp_init - self.kl_weight = kl_weight - self.proj = nn.Conv2d(num_hiddens, codebook_size, 1) # projects last encoder layer to quantized logits - self.embed = nn.Embedding(codebook_size, emb_dim) - - def forward(self, z): - hard = self.straight_through if self.training else True - - logits = self.proj(z) - - soft_one_hot = F.gumbel_softmax(logits, tau=self.temperature, dim=1, hard=hard) - - z_q = torch.einsum("b n h w, n d -> b d h w", soft_one_hot, self.embed.weight) - - # + kl divergence to the prior loss - qy = F.softmax(logits, dim=1) - diff = self.kl_weight * torch.sum(qy * torch.log(qy * self.codebook_size + 1e-10), dim=1).mean() - min_encoding_indices = soft_one_hot.argmax(dim=1) - - return z_q, diff, { - "min_encoding_indices": min_encoding_indices - } - - -class Downsample(nn.Module): - def __init__(self, in_channels): - super().__init__() - self.conv = torch.nn.Conv2d(in_channels, in_channels, kernel_size=3, stride=2, padding=0) - - def forward(self, x): - pad = (0, 1, 0, 1) - x = torch.nn.functional.pad(x, pad, mode="constant", value=0) - x = self.conv(x) - return x - - -class Upsample(nn.Module): - def __init__(self, in_channels): - super().__init__() - self.conv = nn.Conv2d(in_channels, in_channels, kernel_size=3, stride=1, padding=1) - - def forward(self, x): - x = F.interpolate(x, scale_factor=2.0, mode="nearest") - x = self.conv(x) - - return x - - -class ResBlock(nn.Module): - def __init__(self, in_channels, out_channels=None): - super(ResBlock, self).__init__() - self.in_channels = in_channels - self.out_channels = in_channels if out_channels is None else out_channels - self.norm1 = normalize(in_channels) - self.conv1 = nn.Conv2d(in_channels, out_channels, kernel_size=3, stride=1, padding=1) - self.norm2 = normalize(out_channels) - self.conv2 = nn.Conv2d(out_channels, out_channels, kernel_size=3, stride=1, padding=1) - if self.in_channels != self.out_channels: - self.conv_out = nn.Conv2d(in_channels, out_channels, kernel_size=1, stride=1, padding=0) - - def forward(self, x_in): - x = x_in - x = self.norm1(x) - x = swish(x) - x = self.conv1(x) - x = self.norm2(x) - x = swish(x) - x = self.conv2(x) - if self.in_channels != self.out_channels: - x_in = self.conv_out(x_in) - - return x + x_in - - -class AttnBlock(nn.Module): - def __init__(self, in_channels): - super().__init__() - self.in_channels = in_channels - - self.norm = normalize(in_channels) - self.q = torch.nn.Conv2d( - in_channels, - in_channels, - kernel_size=1, - stride=1, - padding=0 - ) - self.k = torch.nn.Conv2d( - in_channels, - in_channels, - kernel_size=1, - stride=1, - padding=0 - ) - self.v = torch.nn.Conv2d( - in_channels, - in_channels, - kernel_size=1, - stride=1, - padding=0 - ) - self.proj_out = torch.nn.Conv2d( - in_channels, - in_channels, - kernel_size=1, - stride=1, - padding=0 - ) - - def forward(self, x): - h_ = x - h_ = self.norm(h_) - q = self.q(h_) - k = self.k(h_) - v = self.v(h_) - - # compute attention - b, c, h, w = q.shape - q = q.reshape(b, c, h*w) - q = q.permute(0, 2, 1) - k = k.reshape(b, c, h*w) - w_ = torch.bmm(q, k) - w_ = w_ * (int(c)**(-0.5)) - w_ = F.softmax(w_, dim=2) - - # attend to values - v = v.reshape(b, c, h*w) - w_ = w_.permute(0, 2, 1) - h_ = torch.bmm(v, w_) - h_ = h_.reshape(b, c, h, w) - - h_ = self.proj_out(h_) - - return x+h_ - - -class Encoder(nn.Module): - def __init__(self, in_channels, nf, emb_dim, ch_mult, num_res_blocks, resolution, attn_resolutions): - super().__init__() - self.nf = nf - self.num_resolutions = len(ch_mult) - self.num_res_blocks = num_res_blocks - self.resolution = resolution - self.attn_resolutions = attn_resolutions - - curr_res = self.resolution - in_ch_mult = (1,)+tuple(ch_mult) - - blocks = [] - # initial convultion - blocks.append(nn.Conv2d(in_channels, nf, kernel_size=3, stride=1, padding=1)) - - # residual and downsampling blocks, with attention on smaller res (16x16) - for i in range(self.num_resolutions): - block_in_ch = nf * in_ch_mult[i] - block_out_ch = nf * ch_mult[i] - for _ in range(self.num_res_blocks): - blocks.append(ResBlock(block_in_ch, block_out_ch)) - block_in_ch = block_out_ch - if curr_res in attn_resolutions: - blocks.append(AttnBlock(block_in_ch)) - - if i != self.num_resolutions - 1: - blocks.append(Downsample(block_in_ch)) - curr_res = curr_res // 2 - - # non-local attention block - blocks.append(ResBlock(block_in_ch, block_in_ch)) - blocks.append(AttnBlock(block_in_ch)) - blocks.append(ResBlock(block_in_ch, block_in_ch)) - - # normalise and convert to latent size - blocks.append(normalize(block_in_ch)) - blocks.append(nn.Conv2d(block_in_ch, emb_dim, kernel_size=3, stride=1, padding=1)) - self.blocks = nn.ModuleList(blocks) - - def forward(self, x): - for block in self.blocks: - x = block(x) - - return x - - -class Generator(nn.Module): - def __init__(self, nf, emb_dim, ch_mult, res_blocks, img_size, attn_resolutions): - super().__init__() - self.nf = nf - self.ch_mult = ch_mult - self.num_resolutions = len(self.ch_mult) - self.num_res_blocks = res_blocks - self.resolution = img_size - self.attn_resolutions = attn_resolutions - self.in_channels = emb_dim - self.out_channels = 3 - block_in_ch = self.nf * self.ch_mult[-1] - curr_res = self.resolution // 2 ** (self.num_resolutions-1) - - blocks = [] - # initial conv - blocks.append(nn.Conv2d(self.in_channels, block_in_ch, kernel_size=3, stride=1, padding=1)) - - # non-local attention block - blocks.append(ResBlock(block_in_ch, block_in_ch)) - blocks.append(AttnBlock(block_in_ch)) - blocks.append(ResBlock(block_in_ch, block_in_ch)) - - for i in reversed(range(self.num_resolutions)): - block_out_ch = self.nf * self.ch_mult[i] - - for _ in range(self.num_res_blocks): - blocks.append(ResBlock(block_in_ch, block_out_ch)) - block_in_ch = block_out_ch - - if curr_res in self.attn_resolutions: - blocks.append(AttnBlock(block_in_ch)) - - if i != 0: - blocks.append(Upsample(block_in_ch)) - curr_res = curr_res * 2 - - blocks.append(normalize(block_in_ch)) - blocks.append(nn.Conv2d(block_in_ch, self.out_channels, kernel_size=3, stride=1, padding=1)) - - self.blocks = nn.ModuleList(blocks) - - - def forward(self, x): - for block in self.blocks: - x = block(x) - - return x - - -@ARCH_REGISTRY.register() -class VQAutoEncoder(nn.Module): - def __init__(self, img_size, nf, ch_mult, quantizer="nearest", res_blocks=2, attn_resolutions=None, codebook_size=1024, emb_dim=256, - beta=0.25, gumbel_straight_through=False, gumbel_kl_weight=1e-8, model_path=None): - super().__init__() - logger = get_root_logger() - self.in_channels = 3 - self.nf = nf - self.n_blocks = res_blocks - self.codebook_size = codebook_size - self.embed_dim = emb_dim - self.ch_mult = ch_mult - self.resolution = img_size - self.attn_resolutions = attn_resolutions or [16] - self.quantizer_type = quantizer - self.encoder = Encoder( - self.in_channels, - self.nf, - self.embed_dim, - self.ch_mult, - self.n_blocks, - self.resolution, - self.attn_resolutions - ) - if self.quantizer_type == "nearest": - self.beta = beta #0.25 - self.quantize = VectorQuantizer(self.codebook_size, self.embed_dim, self.beta) - elif self.quantizer_type == "gumbel": - self.gumbel_num_hiddens = emb_dim - self.straight_through = gumbel_straight_through - self.kl_weight = gumbel_kl_weight - self.quantize = GumbelQuantizer( - self.codebook_size, - self.embed_dim, - self.gumbel_num_hiddens, - self.straight_through, - self.kl_weight - ) - self.generator = Generator( - self.nf, - self.embed_dim, - self.ch_mult, - self.n_blocks, - self.resolution, - self.attn_resolutions - ) - - if model_path is not None: - chkpt = torch.load(model_path, map_location='cpu') - if 'params_ema' in chkpt: - self.load_state_dict(torch.load(model_path, map_location='cpu')['params_ema']) - logger.info(f'vqgan is loaded from: {model_path} [params_ema]') - elif 'params' in chkpt: - self.load_state_dict(torch.load(model_path, map_location='cpu')['params']) - logger.info(f'vqgan is loaded from: {model_path} [params]') - else: - raise ValueError('Wrong params!') - - - def forward(self, x): - x = self.encoder(x) - quant, codebook_loss, quant_stats = self.quantize(x) - x = self.generator(quant) - return x, codebook_loss, quant_stats - - - -# patch based discriminator -@ARCH_REGISTRY.register() -class VQGANDiscriminator(nn.Module): - def __init__(self, nc=3, ndf=64, n_layers=4, model_path=None): - super().__init__() - - layers = [nn.Conv2d(nc, ndf, kernel_size=4, stride=2, padding=1), nn.LeakyReLU(0.2, True)] - ndf_mult = 1 - ndf_mult_prev = 1 - for n in range(1, n_layers): # gradually increase the number of filters - ndf_mult_prev = ndf_mult - ndf_mult = min(2 ** n, 8) - layers += [ - nn.Conv2d(ndf * ndf_mult_prev, ndf * ndf_mult, kernel_size=4, stride=2, padding=1, bias=False), - nn.BatchNorm2d(ndf * ndf_mult), - nn.LeakyReLU(0.2, True) - ] - - ndf_mult_prev = ndf_mult - ndf_mult = min(2 ** n_layers, 8) - - layers += [ - nn.Conv2d(ndf * ndf_mult_prev, ndf * ndf_mult, kernel_size=4, stride=1, padding=1, bias=False), - nn.BatchNorm2d(ndf * ndf_mult), - nn.LeakyReLU(0.2, True) - ] - - layers += [ - nn.Conv2d(ndf * ndf_mult, 1, kernel_size=4, stride=1, padding=1)] # output 1 channel prediction map - self.main = nn.Sequential(*layers) - - if model_path is not None: - chkpt = torch.load(model_path, map_location='cpu') - if 'params_d' in chkpt: - self.load_state_dict(torch.load(model_path, map_location='cpu')['params_d']) - elif 'params' in chkpt: - self.load_state_dict(torch.load(model_path, map_location='cpu')['params']) - else: - raise ValueError('Wrong params!') - - def forward(self, x): - return self.main(x) diff --git a/modules/codeformer_model.py b/modules/codeformer_model.py index da42b5e99..517eadfd8 100644 --- a/modules/codeformer_model.py +++ b/modules/codeformer_model.py @@ -8,9 +8,6 @@ import modules.shared from modules import shared, devices, modelloader, errors from modules.paths import models_path -# codeformer people made a choice to include modified basicsr library to their project which makes -# it utterly impossible to use it alongside with other libraries that also use basicsr, like GFPGAN. -# I am making a choice to include some files from codeformer to work around this issue. model_dir = "Codeformer" model_path = os.path.join(models_path, model_dir) model_url = 'https://github.com/sczhou/CodeFormer/releases/download/v0.1.0/codeformer.pth' @@ -18,115 +15,127 @@ model_url = 'https://github.com/sczhou/CodeFormer/releases/download/v0.1.0/codef codeformer = None +class FaceRestorerCodeFormer(modules.face_restoration.FaceRestoration): + def name(self): + return "CodeFormer" + + def __init__(self, dirname): + self.net = None + self.face_helper = None + self.cmd_dir = dirname + + def create_models(self): + from facexlib.detection import retinaface + from facexlib.utils.face_restoration_helper import FaceRestoreHelper + + if self.net is not None and self.face_helper is not None: + self.net.to(devices.device_codeformer) + return self.net, self.face_helper + model_paths = modelloader.load_models( + model_path, + model_url, + self.cmd_dir, + download_name='codeformer-v0.1.0.pth', + ext_filter=['.pth'], + ) + + if len(model_paths) != 0: + ckpt_path = model_paths[0] + else: + print("Unable to load codeformer model.") + return None, None + net = modelloader.load_spandrel_model(ckpt_path, device=devices.device_codeformer) + + if hasattr(retinaface, 'device'): + retinaface.device = devices.device_codeformer + + face_helper = FaceRestoreHelper( + upscale_factor=1, + face_size=512, + crop_ratio=(1, 1), + det_model='retinaface_resnet50', + save_ext='png', + use_parse=True, + device=devices.device_codeformer, + ) + + self.net = net + self.face_helper = face_helper + + def send_model_to(self, device): + self.net.to(device) + self.face_helper.face_det.to(device) + self.face_helper.face_parse.to(device) + + def restore(self, np_image, w=None): + from torchvision.transforms.functional import normalize + from basicsr.utils import img2tensor, tensor2img + np_image = np_image[:, :, ::-1] + + original_resolution = np_image.shape[0:2] + + self.create_models() + if self.net is None or self.face_helper is None: + return np_image + + self.send_model_to(devices.device_codeformer) + + self.face_helper.clean_all() + self.face_helper.read_image(np_image) + self.face_helper.get_face_landmarks_5(only_center_face=False, resize=640, eye_dist_threshold=5) + self.face_helper.align_warp_face() + + for cropped_face in self.face_helper.cropped_faces: + cropped_face_t = img2tensor(cropped_face / 255., bgr2rgb=True, float32=True) + normalize(cropped_face_t, (0.5, 0.5, 0.5), (0.5, 0.5, 0.5), inplace=True) + cropped_face_t = cropped_face_t.unsqueeze(0).to(devices.device_codeformer) + + try: + with torch.no_grad(): + res = self.net(cropped_face_t, w=w if w is not None else shared.opts.code_former_weight, adain=True) + if isinstance(res, tuple): + output = res[0] + else: + output = res + if not isinstance(res, torch.Tensor): + raise TypeError(f"Expected torch.Tensor, got {type(res)}") + restored_face = tensor2img(output, rgb2bgr=True, min_max=(-1, 1)) + del output + devices.torch_gc() + except Exception: + errors.report('Failed inference for CodeFormer', exc_info=True) + restored_face = tensor2img(cropped_face_t, rgb2bgr=True, min_max=(-1, 1)) + + restored_face = restored_face.astype('uint8') + self.face_helper.add_restored_face(restored_face) + + self.face_helper.get_inverse_affine(None) + + restored_img = self.face_helper.paste_faces_to_input_image() + restored_img = restored_img[:, :, ::-1] + + if original_resolution != restored_img.shape[0:2]: + restored_img = cv2.resize( + restored_img, + (0, 0), + fx=original_resolution[1]/restored_img.shape[1], + fy=original_resolution[0]/restored_img.shape[0], + interpolation=cv2.INTER_LINEAR, + ) + + self.face_helper.clean_all() + + if shared.opts.face_restoration_unload: + self.send_model_to(devices.cpu) + + return restored_img + + def setup_model(dirname): os.makedirs(model_path, exist_ok=True) - - path = modules.paths.paths.get("CodeFormer", None) - if path is None: - return - try: - from torchvision.transforms.functional import normalize - from modules.codeformer.codeformer_arch import CodeFormer - from basicsr.utils import img2tensor, tensor2img - from facelib.utils.face_restoration_helper import FaceRestoreHelper - from facelib.detection.retinaface import retinaface - - net_class = CodeFormer - - class FaceRestorerCodeFormer(modules.face_restoration.FaceRestoration): - def name(self): - return "CodeFormer" - - def __init__(self, dirname): - self.net = None - self.face_helper = None - self.cmd_dir = dirname - - def create_models(self): - - if self.net is not None and self.face_helper is not None: - self.net.to(devices.device_codeformer) - return self.net, self.face_helper - model_paths = modelloader.load_models(model_path, model_url, self.cmd_dir, download_name='codeformer-v0.1.0.pth', ext_filter=['.pth']) - if len(model_paths) != 0: - ckpt_path = model_paths[0] - else: - print("Unable to load codeformer model.") - return None, None - net = net_class(dim_embd=512, codebook_size=1024, n_head=8, n_layers=9, connect_list=['32', '64', '128', '256']).to(devices.device_codeformer) - checkpoint = torch.load(ckpt_path)['params_ema'] - net.load_state_dict(checkpoint) - net.eval() - - if hasattr(retinaface, 'device'): - retinaface.device = devices.device_codeformer - face_helper = FaceRestoreHelper(1, face_size=512, crop_ratio=(1, 1), det_model='retinaface_resnet50', save_ext='png', use_parse=True, device=devices.device_codeformer) - - self.net = net - self.face_helper = face_helper - - return net, face_helper - - def send_model_to(self, device): - self.net.to(device) - self.face_helper.face_det.to(device) - self.face_helper.face_parse.to(device) - - def restore(self, np_image, w=None): - np_image = np_image[:, :, ::-1] - - original_resolution = np_image.shape[0:2] - - self.create_models() - if self.net is None or self.face_helper is None: - return np_image - - self.send_model_to(devices.device_codeformer) - - self.face_helper.clean_all() - self.face_helper.read_image(np_image) - self.face_helper.get_face_landmarks_5(only_center_face=False, resize=640, eye_dist_threshold=5) - self.face_helper.align_warp_face() - - for cropped_face in self.face_helper.cropped_faces: - cropped_face_t = img2tensor(cropped_face / 255., bgr2rgb=True, float32=True) - normalize(cropped_face_t, (0.5, 0.5, 0.5), (0.5, 0.5, 0.5), inplace=True) - cropped_face_t = cropped_face_t.unsqueeze(0).to(devices.device_codeformer) - - try: - with torch.no_grad(): - output = self.net(cropped_face_t, w=w if w is not None else shared.opts.code_former_weight, adain=True)[0] - restored_face = tensor2img(output, rgb2bgr=True, min_max=(-1, 1)) - del output - devices.torch_gc() - except Exception: - errors.report('Failed inference for CodeFormer', exc_info=True) - restored_face = tensor2img(cropped_face_t, rgb2bgr=True, min_max=(-1, 1)) - - restored_face = restored_face.astype('uint8') - self.face_helper.add_restored_face(restored_face) - - self.face_helper.get_inverse_affine(None) - - restored_img = self.face_helper.paste_faces_to_input_image() - restored_img = restored_img[:, :, ::-1] - - if original_resolution != restored_img.shape[0:2]: - restored_img = cv2.resize(restored_img, (0, 0), fx=original_resolution[1]/restored_img.shape[1], fy=original_resolution[0]/restored_img.shape[0], interpolation=cv2.INTER_LINEAR) - - self.face_helper.clean_all() - - if shared.opts.face_restoration_unload: - self.send_model_to(devices.cpu) - - return restored_img - global codeformer codeformer = FaceRestorerCodeFormer(dirname) shared.face_restorers.append(codeformer) - except Exception: errors.report("Error setting up CodeFormer", exc_info=True) - - # sys.path = stored_sys_path diff --git a/modules/esrgan_model.py b/modules/esrgan_model.py index c0d22a992..a7c7c9e30 100644 --- a/modules/esrgan_model.py +++ b/modules/esrgan_model.py @@ -1,122 +1,9 @@ -import sys - -import torch - -import modules.esrgan_model_arch as arch -from modules import modelloader, devices +from modules import modelloader, devices, errors from modules.shared import opts from modules.upscaler import Upscaler, UpscalerData from modules.upscaler_utils import upscale_with_model -def mod2normal(state_dict): - # this code is copied from https://github.com/victorca25/iNNfer - if 'conv_first.weight' in state_dict: - crt_net = {} - items = list(state_dict) - - crt_net['model.0.weight'] = state_dict['conv_first.weight'] - crt_net['model.0.bias'] = state_dict['conv_first.bias'] - - for k in items.copy(): - if 'RDB' in k: - ori_k = k.replace('RRDB_trunk.', 'model.1.sub.') - if '.weight' in k: - ori_k = ori_k.replace('.weight', '.0.weight') - elif '.bias' in k: - ori_k = ori_k.replace('.bias', '.0.bias') - crt_net[ori_k] = state_dict[k] - items.remove(k) - - crt_net['model.1.sub.23.weight'] = state_dict['trunk_conv.weight'] - crt_net['model.1.sub.23.bias'] = state_dict['trunk_conv.bias'] - crt_net['model.3.weight'] = state_dict['upconv1.weight'] - crt_net['model.3.bias'] = state_dict['upconv1.bias'] - crt_net['model.6.weight'] = state_dict['upconv2.weight'] - crt_net['model.6.bias'] = state_dict['upconv2.bias'] - crt_net['model.8.weight'] = state_dict['HRconv.weight'] - crt_net['model.8.bias'] = state_dict['HRconv.bias'] - crt_net['model.10.weight'] = state_dict['conv_last.weight'] - crt_net['model.10.bias'] = state_dict['conv_last.bias'] - state_dict = crt_net - return state_dict - - -def resrgan2normal(state_dict, nb=23): - # this code is copied from https://github.com/victorca25/iNNfer - if "conv_first.weight" in state_dict and "body.0.rdb1.conv1.weight" in state_dict: - re8x = 0 - crt_net = {} - items = list(state_dict) - - crt_net['model.0.weight'] = state_dict['conv_first.weight'] - crt_net['model.0.bias'] = state_dict['conv_first.bias'] - - for k in items.copy(): - if "rdb" in k: - ori_k = k.replace('body.', 'model.1.sub.') - ori_k = ori_k.replace('.rdb', '.RDB') - if '.weight' in k: - ori_k = ori_k.replace('.weight', '.0.weight') - elif '.bias' in k: - ori_k = ori_k.replace('.bias', '.0.bias') - crt_net[ori_k] = state_dict[k] - items.remove(k) - - crt_net[f'model.1.sub.{nb}.weight'] = state_dict['conv_body.weight'] - crt_net[f'model.1.sub.{nb}.bias'] = state_dict['conv_body.bias'] - crt_net['model.3.weight'] = state_dict['conv_up1.weight'] - crt_net['model.3.bias'] = state_dict['conv_up1.bias'] - crt_net['model.6.weight'] = state_dict['conv_up2.weight'] - crt_net['model.6.bias'] = state_dict['conv_up2.bias'] - - if 'conv_up3.weight' in state_dict: - # modification supporting: https://github.com/ai-forever/Real-ESRGAN/blob/main/RealESRGAN/rrdbnet_arch.py - re8x = 3 - crt_net['model.9.weight'] = state_dict['conv_up3.weight'] - crt_net['model.9.bias'] = state_dict['conv_up3.bias'] - - crt_net[f'model.{8+re8x}.weight'] = state_dict['conv_hr.weight'] - crt_net[f'model.{8+re8x}.bias'] = state_dict['conv_hr.bias'] - crt_net[f'model.{10+re8x}.weight'] = state_dict['conv_last.weight'] - crt_net[f'model.{10+re8x}.bias'] = state_dict['conv_last.bias'] - - state_dict = crt_net - return state_dict - - -def infer_params(state_dict): - # this code is copied from https://github.com/victorca25/iNNfer - scale2x = 0 - scalemin = 6 - n_uplayer = 0 - plus = False - - for block in list(state_dict): - parts = block.split(".") - n_parts = len(parts) - if n_parts == 5 and parts[2] == "sub": - nb = int(parts[3]) - elif n_parts == 3: - part_num = int(parts[1]) - if (part_num > scalemin - and parts[0] == "model" - and parts[2] == "weight"): - scale2x += 1 - if part_num > n_uplayer: - n_uplayer = part_num - out_nc = state_dict[block].shape[0] - if not plus and "conv1x1" in block: - plus = True - - nf = state_dict["model.0.weight"].shape[0] - in_nc = state_dict["model.0.weight"].shape[1] - out_nc = out_nc - scale = 2 ** scale2x - - return in_nc, out_nc, nf, nb, plus, scale - - class UpscalerESRGAN(Upscaler): def __init__(self, dirname): self.name = "ESRGAN" @@ -142,12 +29,11 @@ class UpscalerESRGAN(Upscaler): def do_upscale(self, img, selected_model): try: model = self.load_model(selected_model) - except Exception as e: - print(f"Unable to load ESRGAN model {selected_model}: {e}", file=sys.stderr) + except Exception: + errors.report(f"Unable to load ESRGAN model {selected_model}", exc_info=True) return img model.to(devices.device_esrgan) - img = esrgan_upscale(model, img) - return img + return esrgan_upscale(model, img) def load_model(self, path: str): if path.startswith("http"): @@ -160,33 +46,10 @@ class UpscalerESRGAN(Upscaler): else: filename = path - state_dict = torch.load(filename, map_location='cpu' if devices.device_esrgan.type == 'mps' else None) - - if "params_ema" in state_dict: - state_dict = state_dict["params_ema"] - elif "params" in state_dict: - state_dict = state_dict["params"] - num_conv = 16 if "realesr-animevideov3" in filename else 32 - model = arch.SRVGGNetCompact(num_in_ch=3, num_out_ch=3, num_feat=64, num_conv=num_conv, upscale=4, act_type='prelu') - model.load_state_dict(state_dict) - model.eval() - return model - - if "body.0.rdb1.conv1.weight" in state_dict and "conv_first.weight" in state_dict: - nb = 6 if "RealESRGAN_x4plus_anime_6B" in filename else 23 - state_dict = resrgan2normal(state_dict, nb) - elif "conv_first.weight" in state_dict: - state_dict = mod2normal(state_dict) - elif "model.0.weight" not in state_dict: - raise Exception("The file is not a recognized ESRGAN model.") - - in_nc, out_nc, nf, nb, plus, mscale = infer_params(state_dict) - - model = arch.RRDBNet(in_nc=in_nc, out_nc=out_nc, nf=nf, nb=nb, upscale=mscale, plus=plus) - model.load_state_dict(state_dict) - model.eval() - - return model + return modelloader.load_spandrel_model( + filename, + device=('cpu' if devices.device_esrgan.type == 'mps' else None), + ) def esrgan_upscale(model, img): diff --git a/modules/esrgan_model_arch.py b/modules/esrgan_model_arch.py deleted file mode 100644 index 2b9888baf..000000000 --- a/modules/esrgan_model_arch.py +++ /dev/null @@ -1,465 +0,0 @@ -# this file is adapted from https://github.com/victorca25/iNNfer - -from collections import OrderedDict -import math -import torch -import torch.nn as nn -import torch.nn.functional as F - - -#################### -# RRDBNet Generator -#################### - -class RRDBNet(nn.Module): - def __init__(self, in_nc, out_nc, nf, nb, nr=3, gc=32, upscale=4, norm_type=None, - act_type='leakyrelu', mode='CNA', upsample_mode='upconv', convtype='Conv2D', - finalact=None, gaussian_noise=False, plus=False): - super(RRDBNet, self).__init__() - n_upscale = int(math.log(upscale, 2)) - if upscale == 3: - n_upscale = 1 - - self.resrgan_scale = 0 - if in_nc % 16 == 0: - self.resrgan_scale = 1 - elif in_nc != 4 and in_nc % 4 == 0: - self.resrgan_scale = 2 - - fea_conv = conv_block(in_nc, nf, kernel_size=3, norm_type=None, act_type=None, convtype=convtype) - rb_blocks = [RRDB(nf, nr, kernel_size=3, gc=32, stride=1, bias=1, pad_type='zero', - norm_type=norm_type, act_type=act_type, mode='CNA', convtype=convtype, - gaussian_noise=gaussian_noise, plus=plus) for _ in range(nb)] - LR_conv = conv_block(nf, nf, kernel_size=3, norm_type=norm_type, act_type=None, mode=mode, convtype=convtype) - - if upsample_mode == 'upconv': - upsample_block = upconv_block - elif upsample_mode == 'pixelshuffle': - upsample_block = pixelshuffle_block - else: - raise NotImplementedError(f'upsample mode [{upsample_mode}] is not found') - if upscale == 3: - upsampler = upsample_block(nf, nf, 3, act_type=act_type, convtype=convtype) - else: - upsampler = [upsample_block(nf, nf, act_type=act_type, convtype=convtype) for _ in range(n_upscale)] - HR_conv0 = conv_block(nf, nf, kernel_size=3, norm_type=None, act_type=act_type, convtype=convtype) - HR_conv1 = conv_block(nf, out_nc, kernel_size=3, norm_type=None, act_type=None, convtype=convtype) - - outact = act(finalact) if finalact else None - - self.model = sequential(fea_conv, ShortcutBlock(sequential(*rb_blocks, LR_conv)), - *upsampler, HR_conv0, HR_conv1, outact) - - def forward(self, x, outm=None): - if self.resrgan_scale == 1: - feat = pixel_unshuffle(x, scale=4) - elif self.resrgan_scale == 2: - feat = pixel_unshuffle(x, scale=2) - else: - feat = x - - return self.model(feat) - - -class RRDB(nn.Module): - """ - Residual in Residual Dense Block - (ESRGAN: Enhanced Super-Resolution Generative Adversarial Networks) - """ - - def __init__(self, nf, nr=3, kernel_size=3, gc=32, stride=1, bias=1, pad_type='zero', - norm_type=None, act_type='leakyrelu', mode='CNA', convtype='Conv2D', - spectral_norm=False, gaussian_noise=False, plus=False): - super(RRDB, self).__init__() - # This is for backwards compatibility with existing models - if nr == 3: - self.RDB1 = ResidualDenseBlock_5C(nf, kernel_size, gc, stride, bias, pad_type, - norm_type, act_type, mode, convtype, spectral_norm=spectral_norm, - gaussian_noise=gaussian_noise, plus=plus) - self.RDB2 = ResidualDenseBlock_5C(nf, kernel_size, gc, stride, bias, pad_type, - norm_type, act_type, mode, convtype, spectral_norm=spectral_norm, - gaussian_noise=gaussian_noise, plus=plus) - self.RDB3 = ResidualDenseBlock_5C(nf, kernel_size, gc, stride, bias, pad_type, - norm_type, act_type, mode, convtype, spectral_norm=spectral_norm, - gaussian_noise=gaussian_noise, plus=plus) - else: - RDB_list = [ResidualDenseBlock_5C(nf, kernel_size, gc, stride, bias, pad_type, - norm_type, act_type, mode, convtype, spectral_norm=spectral_norm, - gaussian_noise=gaussian_noise, plus=plus) for _ in range(nr)] - self.RDBs = nn.Sequential(*RDB_list) - - def forward(self, x): - if hasattr(self, 'RDB1'): - out = self.RDB1(x) - out = self.RDB2(out) - out = self.RDB3(out) - else: - out = self.RDBs(x) - return out * 0.2 + x - - -class ResidualDenseBlock_5C(nn.Module): - """ - Residual Dense Block - The core module of paper: (Residual Dense Network for Image Super-Resolution, CVPR 18) - Modified options that can be used: - - "Partial Convolution based Padding" arXiv:1811.11718 - - "Spectral normalization" arXiv:1802.05957 - - "ICASSP 2020 - ESRGAN+ : Further Improving ESRGAN" N. C. - {Rakotonirina} and A. {Rasoanaivo} - """ - - def __init__(self, nf=64, kernel_size=3, gc=32, stride=1, bias=1, pad_type='zero', - norm_type=None, act_type='leakyrelu', mode='CNA', convtype='Conv2D', - spectral_norm=False, gaussian_noise=False, plus=False): - super(ResidualDenseBlock_5C, self).__init__() - - self.noise = GaussianNoise() if gaussian_noise else None - self.conv1x1 = conv1x1(nf, gc) if plus else None - - self.conv1 = conv_block(nf, gc, kernel_size, stride, bias=bias, pad_type=pad_type, - norm_type=norm_type, act_type=act_type, mode=mode, convtype=convtype, - spectral_norm=spectral_norm) - self.conv2 = conv_block(nf+gc, gc, kernel_size, stride, bias=bias, pad_type=pad_type, - norm_type=norm_type, act_type=act_type, mode=mode, convtype=convtype, - spectral_norm=spectral_norm) - self.conv3 = conv_block(nf+2*gc, gc, kernel_size, stride, bias=bias, pad_type=pad_type, - norm_type=norm_type, act_type=act_type, mode=mode, convtype=convtype, - spectral_norm=spectral_norm) - self.conv4 = conv_block(nf+3*gc, gc, kernel_size, stride, bias=bias, pad_type=pad_type, - norm_type=norm_type, act_type=act_type, mode=mode, convtype=convtype, - spectral_norm=spectral_norm) - if mode == 'CNA': - last_act = None - else: - last_act = act_type - self.conv5 = conv_block(nf+4*gc, nf, 3, stride, bias=bias, pad_type=pad_type, - norm_type=norm_type, act_type=last_act, mode=mode, convtype=convtype, - spectral_norm=spectral_norm) - - def forward(self, x): - x1 = self.conv1(x) - x2 = self.conv2(torch.cat((x, x1), 1)) - if self.conv1x1: - x2 = x2 + self.conv1x1(x) - x3 = self.conv3(torch.cat((x, x1, x2), 1)) - x4 = self.conv4(torch.cat((x, x1, x2, x3), 1)) - if self.conv1x1: - x4 = x4 + x2 - x5 = self.conv5(torch.cat((x, x1, x2, x3, x4), 1)) - if self.noise: - return self.noise(x5.mul(0.2) + x) - else: - return x5 * 0.2 + x - - -#################### -# ESRGANplus -#################### - -class GaussianNoise(nn.Module): - def __init__(self, sigma=0.1, is_relative_detach=False): - super().__init__() - self.sigma = sigma - self.is_relative_detach = is_relative_detach - self.noise = torch.tensor(0, dtype=torch.float) - - def forward(self, x): - if self.training and self.sigma != 0: - self.noise = self.noise.to(x.device) - scale = self.sigma * x.detach() if self.is_relative_detach else self.sigma * x - sampled_noise = self.noise.repeat(*x.size()).normal_() * scale - x = x + sampled_noise - return x - -def conv1x1(in_planes, out_planes, stride=1): - return nn.Conv2d(in_planes, out_planes, kernel_size=1, stride=stride, bias=False) - - -#################### -# SRVGGNetCompact -#################### - -class SRVGGNetCompact(nn.Module): - """A compact VGG-style network structure for super-resolution. - This class is copied from https://github.com/xinntao/Real-ESRGAN - """ - - def __init__(self, num_in_ch=3, num_out_ch=3, num_feat=64, num_conv=16, upscale=4, act_type='prelu'): - super(SRVGGNetCompact, self).__init__() - self.num_in_ch = num_in_ch - self.num_out_ch = num_out_ch - self.num_feat = num_feat - self.num_conv = num_conv - self.upscale = upscale - self.act_type = act_type - - self.body = nn.ModuleList() - # the first conv - self.body.append(nn.Conv2d(num_in_ch, num_feat, 3, 1, 1)) - # the first activation - if act_type == 'relu': - activation = nn.ReLU(inplace=True) - elif act_type == 'prelu': - activation = nn.PReLU(num_parameters=num_feat) - elif act_type == 'leakyrelu': - activation = nn.LeakyReLU(negative_slope=0.1, inplace=True) - self.body.append(activation) - - # the body structure - for _ in range(num_conv): - self.body.append(nn.Conv2d(num_feat, num_feat, 3, 1, 1)) - # activation - if act_type == 'relu': - activation = nn.ReLU(inplace=True) - elif act_type == 'prelu': - activation = nn.PReLU(num_parameters=num_feat) - elif act_type == 'leakyrelu': - activation = nn.LeakyReLU(negative_slope=0.1, inplace=True) - self.body.append(activation) - - # the last conv - self.body.append(nn.Conv2d(num_feat, num_out_ch * upscale * upscale, 3, 1, 1)) - # upsample - self.upsampler = nn.PixelShuffle(upscale) - - def forward(self, x): - out = x - for i in range(0, len(self.body)): - out = self.body[i](out) - - out = self.upsampler(out) - # add the nearest upsampled image, so that the network learns the residual - base = F.interpolate(x, scale_factor=self.upscale, mode='nearest') - out += base - return out - - -#################### -# Upsampler -#################### - -class Upsample(nn.Module): - r"""Upsamples a given multi-channel 1D (temporal), 2D (spatial) or 3D (volumetric) data. - The input data is assumed to be of the form - `minibatch x channels x [optional depth] x [optional height] x width`. - """ - - def __init__(self, size=None, scale_factor=None, mode="nearest", align_corners=None): - super(Upsample, self).__init__() - if isinstance(scale_factor, tuple): - self.scale_factor = tuple(float(factor) for factor in scale_factor) - else: - self.scale_factor = float(scale_factor) if scale_factor else None - self.mode = mode - self.size = size - self.align_corners = align_corners - - def forward(self, x): - return nn.functional.interpolate(x, size=self.size, scale_factor=self.scale_factor, mode=self.mode, align_corners=self.align_corners) - - def extra_repr(self): - if self.scale_factor is not None: - info = f'scale_factor={self.scale_factor}' - else: - info = f'size={self.size}' - info += f', mode={self.mode}' - return info - - -def pixel_unshuffle(x, scale): - """ Pixel unshuffle. - Args: - x (Tensor): Input feature with shape (b, c, hh, hw). - scale (int): Downsample ratio. - Returns: - Tensor: the pixel unshuffled feature. - """ - b, c, hh, hw = x.size() - out_channel = c * (scale**2) - assert hh % scale == 0 and hw % scale == 0 - h = hh // scale - w = hw // scale - x_view = x.view(b, c, h, scale, w, scale) - return x_view.permute(0, 1, 3, 5, 2, 4).reshape(b, out_channel, h, w) - - -def pixelshuffle_block(in_nc, out_nc, upscale_factor=2, kernel_size=3, stride=1, bias=True, - pad_type='zero', norm_type=None, act_type='relu', convtype='Conv2D'): - """ - Pixel shuffle layer - (Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional - Neural Network, CVPR17) - """ - conv = conv_block(in_nc, out_nc * (upscale_factor ** 2), kernel_size, stride, bias=bias, - pad_type=pad_type, norm_type=None, act_type=None, convtype=convtype) - pixel_shuffle = nn.PixelShuffle(upscale_factor) - - n = norm(norm_type, out_nc) if norm_type else None - a = act(act_type) if act_type else None - return sequential(conv, pixel_shuffle, n, a) - - -def upconv_block(in_nc, out_nc, upscale_factor=2, kernel_size=3, stride=1, bias=True, - pad_type='zero', norm_type=None, act_type='relu', mode='nearest', convtype='Conv2D'): - """ Upconv layer """ - upscale_factor = (1, upscale_factor, upscale_factor) if convtype == 'Conv3D' else upscale_factor - upsample = Upsample(scale_factor=upscale_factor, mode=mode) - conv = conv_block(in_nc, out_nc, kernel_size, stride, bias=bias, - pad_type=pad_type, norm_type=norm_type, act_type=act_type, convtype=convtype) - return sequential(upsample, conv) - - - - - - - - -#################### -# Basic blocks -#################### - - -def make_layer(basic_block, num_basic_block, **kwarg): - """Make layers by stacking the same blocks. - Args: - basic_block (nn.module): nn.module class for basic block. (block) - num_basic_block (int): number of blocks. (n_layers) - Returns: - nn.Sequential: Stacked blocks in nn.Sequential. - """ - layers = [] - for _ in range(num_basic_block): - layers.append(basic_block(**kwarg)) - return nn.Sequential(*layers) - - -def act(act_type, inplace=True, neg_slope=0.2, n_prelu=1, beta=1.0): - """ activation helper """ - act_type = act_type.lower() - if act_type == 'relu': - layer = nn.ReLU(inplace) - elif act_type in ('leakyrelu', 'lrelu'): - layer = nn.LeakyReLU(neg_slope, inplace) - elif act_type == 'prelu': - layer = nn.PReLU(num_parameters=n_prelu, init=neg_slope) - elif act_type == 'tanh': # [-1, 1] range output - layer = nn.Tanh() - elif act_type == 'sigmoid': # [0, 1] range output - layer = nn.Sigmoid() - else: - raise NotImplementedError(f'activation layer [{act_type}] is not found') - return layer - - -class Identity(nn.Module): - def __init__(self, *kwargs): - super(Identity, self).__init__() - - def forward(self, x, *kwargs): - return x - - -def norm(norm_type, nc): - """ Return a normalization layer """ - norm_type = norm_type.lower() - if norm_type == 'batch': - layer = nn.BatchNorm2d(nc, affine=True) - elif norm_type == 'instance': - layer = nn.InstanceNorm2d(nc, affine=False) - elif norm_type == 'none': - def norm_layer(x): return Identity() - else: - raise NotImplementedError(f'normalization layer [{norm_type}] is not found') - return layer - - -def pad(pad_type, padding): - """ padding layer helper """ - pad_type = pad_type.lower() - if padding == 0: - return None - if pad_type == 'reflect': - layer = nn.ReflectionPad2d(padding) - elif pad_type == 'replicate': - layer = nn.ReplicationPad2d(padding) - elif pad_type == 'zero': - layer = nn.ZeroPad2d(padding) - else: - raise NotImplementedError(f'padding layer [{pad_type}] is not implemented') - return layer - - -def get_valid_padding(kernel_size, dilation): - kernel_size = kernel_size + (kernel_size - 1) * (dilation - 1) - padding = (kernel_size - 1) // 2 - return padding - - -class ShortcutBlock(nn.Module): - """ Elementwise sum the output of a submodule to its input """ - def __init__(self, submodule): - super(ShortcutBlock, self).__init__() - self.sub = submodule - - def forward(self, x): - output = x + self.sub(x) - return output - - def __repr__(self): - return 'Identity + \n|' + self.sub.__repr__().replace('\n', '\n|') - - -def sequential(*args): - """ Flatten Sequential. It unwraps nn.Sequential. """ - if len(args) == 1: - if isinstance(args[0], OrderedDict): - raise NotImplementedError('sequential does not support OrderedDict input.') - return args[0] # No sequential is needed. - modules = [] - for module in args: - if isinstance(module, nn.Sequential): - for submodule in module.children(): - modules.append(submodule) - elif isinstance(module, nn.Module): - modules.append(module) - return nn.Sequential(*modules) - - -def conv_block(in_nc, out_nc, kernel_size, stride=1, dilation=1, groups=1, bias=True, - pad_type='zero', norm_type=None, act_type='relu', mode='CNA', convtype='Conv2D', - spectral_norm=False): - """ Conv layer with padding, normalization, activation """ - assert mode in ['CNA', 'NAC', 'CNAC'], f'Wrong conv mode [{mode}]' - padding = get_valid_padding(kernel_size, dilation) - p = pad(pad_type, padding) if pad_type and pad_type != 'zero' else None - padding = padding if pad_type == 'zero' else 0 - - if convtype=='PartialConv2D': - from torchvision.ops import PartialConv2d # this is definitely not going to work, but PartialConv2d doesn't work anyway and this shuts up static analyzer - c = PartialConv2d(in_nc, out_nc, kernel_size=kernel_size, stride=stride, padding=padding, - dilation=dilation, bias=bias, groups=groups) - elif convtype=='DeformConv2D': - from torchvision.ops import DeformConv2d # not tested - c = DeformConv2d(in_nc, out_nc, kernel_size=kernel_size, stride=stride, padding=padding, - dilation=dilation, bias=bias, groups=groups) - elif convtype=='Conv3D': - c = nn.Conv3d(in_nc, out_nc, kernel_size=kernel_size, stride=stride, padding=padding, - dilation=dilation, bias=bias, groups=groups) - else: - c = nn.Conv2d(in_nc, out_nc, kernel_size=kernel_size, stride=stride, padding=padding, - dilation=dilation, bias=bias, groups=groups) - - if spectral_norm: - c = nn.utils.spectral_norm(c) - - a = act(act_type) if act_type else None - if 'CNA' in mode: - n = norm(norm_type, out_nc) if norm_type else None - return sequential(p, c, n, a) - elif mode == 'NAC': - if norm_type is None and act_type is not None: - a = act(act_type, inplace=False) - n = norm(norm_type, in_nc) if norm_type else None - return sequential(n, a, p, c) diff --git a/modules/gfpgan_model.py b/modules/gfpgan_model.py index 01d668ecd..6b6f17c43 100644 --- a/modules/gfpgan_model.py +++ b/modules/gfpgan_model.py @@ -1,8 +1,5 @@ import os -import facexlib -import gfpgan - import modules.face_restoration from modules import paths, shared, devices, modelloader, errors @@ -41,6 +38,8 @@ def gfpgann(): print("Unable to load gfpgan model!") return None + import facexlib.detection.retinaface + if hasattr(facexlib.detection.retinaface, 'device'): facexlib.detection.retinaface.device = devices.device_gfpgan model_file_path = model_file @@ -81,8 +80,10 @@ gfpgan_constructor = None def setup_model(dirname): try: os.makedirs(model_path, exist_ok=True) - from gfpgan import GFPGANer - from facexlib import detection, parsing # noqa: F401 + import gfpgan + import facexlib.detection + import facexlib.parsing + global user_path global have_gfpgan global gfpgan_constructor @@ -111,7 +112,7 @@ def setup_model(dirname): facexlib.parsing.load_file_from_url = facex_load_file_from_url2 user_path = dirname have_gfpgan = True - gfpgan_constructor = GFPGANer + gfpgan_constructor = gfpgan.GFPGANer class FaceRestorerGFPGAN(modules.face_restoration.FaceRestoration): def name(self): diff --git a/modules/launch_utils.py b/modules/launch_utils.py index dabef0f53..c2cbd8ce7 100644 --- a/modules/launch_utils.py +++ b/modules/launch_utils.py @@ -345,13 +345,11 @@ def prepare_environment(): stable_diffusion_repo = os.environ.get('STABLE_DIFFUSION_REPO', "https://github.com/Stability-AI/stablediffusion.git") stable_diffusion_xl_repo = os.environ.get('STABLE_DIFFUSION_XL_REPO', "https://github.com/Stability-AI/generative-models.git") k_diffusion_repo = os.environ.get('K_DIFFUSION_REPO', 'https://github.com/crowsonkb/k-diffusion.git') - codeformer_repo = os.environ.get('CODEFORMER_REPO', 'https://github.com/sczhou/CodeFormer.git') blip_repo = os.environ.get('BLIP_REPO', 'https://github.com/salesforce/BLIP.git') stable_diffusion_commit_hash = os.environ.get('STABLE_DIFFUSION_COMMIT_HASH', "cf1d67a6fd5ea1aa600c4df58e5b47da45f6bdbf") stable_diffusion_xl_commit_hash = os.environ.get('STABLE_DIFFUSION_XL_COMMIT_HASH', "45c443b316737a4ab6e40413d7794a7f5657c19f") k_diffusion_commit_hash = os.environ.get('K_DIFFUSION_COMMIT_HASH', "ab527a9a6d347f364e3d185ba6d714e22d80cb3c") - codeformer_commit_hash = os.environ.get('CODEFORMER_COMMIT_HASH', "c5b4593074ba6214284d6acd5f1719b6c5d739af") blip_commit_hash = os.environ.get('BLIP_COMMIT_HASH', "48211a1594f1321b00f14c9f7a5b4813144b2fb9") try: @@ -408,15 +406,10 @@ def prepare_environment(): git_clone(stable_diffusion_repo, repo_dir('stable-diffusion-stability-ai'), "Stable Diffusion", stable_diffusion_commit_hash) git_clone(stable_diffusion_xl_repo, repo_dir('generative-models'), "Stable Diffusion XL", stable_diffusion_xl_commit_hash) git_clone(k_diffusion_repo, repo_dir('k-diffusion'), "K-diffusion", k_diffusion_commit_hash) - git_clone(codeformer_repo, repo_dir('CodeFormer'), "CodeFormer", codeformer_commit_hash) git_clone(blip_repo, repo_dir('BLIP'), "BLIP", blip_commit_hash) startup_timer.record("clone repositores") - if not is_installed("lpips"): - run_pip(f"install -r \"{os.path.join(repo_dir('CodeFormer'), 'requirements.txt')}\"", "requirements for CodeFormer") - startup_timer.record("install CodeFormer requirements") - if not os.path.isfile(requirements_file): requirements_file = os.path.join(script_path, requirements_file) diff --git a/modules/modelloader.py b/modules/modelloader.py index 098bcb793..30116932a 100644 --- a/modules/modelloader.py +++ b/modules/modelloader.py @@ -1,5 +1,6 @@ from __future__ import annotations +import logging import os import shutil import importlib @@ -10,6 +11,9 @@ from modules.upscaler import Upscaler, UpscalerLanczos, UpscalerNearest, Upscale from modules.paths import script_path, models_path +logger = logging.getLogger(__name__) + + def load_file_from_url( url: str, *, @@ -177,3 +181,15 @@ def load_upscalers(): # Special case for UpscalerNone keeps it at the beginning of the list. key=lambda x: x.name.lower() if not isinstance(x.scaler, (UpscalerNone, UpscalerLanczos, UpscalerNearest)) else "" ) + + +def load_spandrel_model(path, *, device, half: bool = False, dtype=None): + import spandrel + model = spandrel.ModelLoader(device=device).load_from_file(path) + if half: + model = model.model.half() + if dtype: + model = model.model.to(dtype=dtype) + model.eval() + logger.debug("Loaded %s from %s (device=%s, half=%s, dtype=%s)", model, path, device, half, dtype) + return model diff --git a/modules/paths.py b/modules/paths.py index 187b94961..030646519 100644 --- a/modules/paths.py +++ b/modules/paths.py @@ -38,7 +38,6 @@ mute_sdxl_imports() path_dirs = [ (sd_path, 'ldm', 'Stable Diffusion', []), (os.path.join(sd_path, '../generative-models'), 'sgm', 'Stable Diffusion XL', ["sgm"]), - (os.path.join(sd_path, '../CodeFormer'), 'inference_codeformer.py', 'CodeFormer', []), (os.path.join(sd_path, '../BLIP'), 'models/blip.py', 'BLIP', []), (os.path.join(sd_path, '../k-diffusion'), 'k_diffusion/sampling.py', 'k_diffusion', ["atstart"]), ] diff --git a/modules/realesrgan_model.py b/modules/realesrgan_model.py index 02841c302..332d8f4b1 100644 --- a/modules/realesrgan_model.py +++ b/modules/realesrgan_model.py @@ -1,9 +1,6 @@ import os -import numpy as np -from PIL import Image -from realesrgan import RealESRGANer - +from modules.upscaler_utils import upscale_with_model from modules.upscaler import Upscaler, UpscalerData from modules.shared import cmd_opts, opts from modules import modelloader, errors @@ -14,29 +11,20 @@ class UpscalerRealESRGAN(Upscaler): self.name = "RealESRGAN" self.user_path = path super().__init__() - try: - from basicsr.archs.rrdbnet_arch import RRDBNet # noqa: F401 - from realesrgan import RealESRGANer # noqa: F401 - from realesrgan.archs.srvgg_arch import SRVGGNetCompact # noqa: F401 - self.enable = True - self.scalers = [] - scalers = self.load_models(path) + self.enable = True + self.scalers = [] + scalers = get_realesrgan_models(self) - local_model_paths = self.find_models(ext_filter=[".pth"]) - for scaler in scalers: - if scaler.local_data_path.startswith("http"): - filename = modelloader.friendly_name(scaler.local_data_path) - local_model_candidates = [local_model for local_model in local_model_paths if local_model.endswith(f"{filename}.pth")] - if local_model_candidates: - scaler.local_data_path = local_model_candidates[0] + local_model_paths = self.find_models(ext_filter=[".pth"]) + for scaler in scalers: + if scaler.local_data_path.startswith("http"): + filename = modelloader.friendly_name(scaler.local_data_path) + local_model_candidates = [local_model for local_model in local_model_paths if local_model.endswith(f"{filename}.pth")] + if local_model_candidates: + scaler.local_data_path = local_model_candidates[0] - if scaler.name in opts.realesrgan_enabled_models: - self.scalers.append(scaler) - - except Exception: - errors.report("Error importing Real-ESRGAN", exc_info=True) - self.enable = False - self.scalers = [] + if scaler.name in opts.realesrgan_enabled_models: + self.scalers.append(scaler) def do_upscale(self, img, path): if not self.enable: @@ -48,20 +36,18 @@ class UpscalerRealESRGAN(Upscaler): errors.report(f"Unable to load RealESRGAN model {path}", exc_info=True) return img - upsampler = RealESRGANer( - scale=info.scale, - model_path=info.local_data_path, - model=info.model(), - half=not cmd_opts.no_half and not cmd_opts.upcast_sampling, - tile=opts.ESRGAN_tile, - tile_pad=opts.ESRGAN_tile_overlap, + mod = modelloader.load_spandrel_model( + info.local_data_path, device=self.device, + half=(not cmd_opts.no_half and not cmd_opts.upcast_sampling), + ) + return upscale_with_model( + mod, + img, + tile_size=opts.ESRGAN_tile, + tile_overlap=opts.ESRGAN_tile_overlap, + # TODO: `outscale`? ) - - upsampled = upsampler.enhance(np.array(img), outscale=info.scale)[0] - - image = Image.fromarray(upsampled) - return image def load_model(self, path): for scaler in self.scalers: @@ -76,58 +62,43 @@ class UpscalerRealESRGAN(Upscaler): return scaler raise ValueError(f"Unable to find model info: {path}") - def load_models(self, _): - return get_realesrgan_models(self) - -def get_realesrgan_models(scaler): - try: - from basicsr.archs.rrdbnet_arch import RRDBNet - from realesrgan.archs.srvgg_arch import SRVGGNetCompact - models = [ - UpscalerData( - name="R-ESRGAN General 4xV3", - path="https://github.com/xinntao/Real-ESRGAN/releases/download/v0.2.5.0/realesr-general-x4v3.pth", - scale=4, - upscaler=scaler, - model=lambda: SRVGGNetCompact(num_in_ch=3, num_out_ch=3, num_feat=64, num_conv=32, upscale=4, act_type='prelu') - ), - UpscalerData( - name="R-ESRGAN General WDN 4xV3", - path="https://github.com/xinntao/Real-ESRGAN/releases/download/v0.2.5.0/realesr-general-wdn-x4v3.pth", - scale=4, - upscaler=scaler, - model=lambda: SRVGGNetCompact(num_in_ch=3, num_out_ch=3, num_feat=64, num_conv=32, upscale=4, act_type='prelu') - ), - UpscalerData( - name="R-ESRGAN AnimeVideo", - path="https://github.com/xinntao/Real-ESRGAN/releases/download/v0.2.5.0/realesr-animevideov3.pth", - scale=4, - upscaler=scaler, - model=lambda: SRVGGNetCompact(num_in_ch=3, num_out_ch=3, num_feat=64, num_conv=16, upscale=4, act_type='prelu') - ), - UpscalerData( - name="R-ESRGAN 4x+", - path="https://github.com/xinntao/Real-ESRGAN/releases/download/v0.1.0/RealESRGAN_x4plus.pth", - scale=4, - upscaler=scaler, - model=lambda: RRDBNet(num_in_ch=3, num_out_ch=3, num_feat=64, num_block=23, num_grow_ch=32, scale=4) - ), - UpscalerData( - name="R-ESRGAN 4x+ Anime6B", - path="https://github.com/xinntao/Real-ESRGAN/releases/download/v0.2.2.4/RealESRGAN_x4plus_anime_6B.pth", - scale=4, - upscaler=scaler, - model=lambda: RRDBNet(num_in_ch=3, num_out_ch=3, num_feat=64, num_block=6, num_grow_ch=32, scale=4) - ), - UpscalerData( - name="R-ESRGAN 2x+", - path="https://github.com/xinntao/Real-ESRGAN/releases/download/v0.2.1/RealESRGAN_x2plus.pth", - scale=2, - upscaler=scaler, - model=lambda: RRDBNet(num_in_ch=3, num_out_ch=3, num_feat=64, num_block=23, num_grow_ch=32, scale=2) - ), - ] - return models - except Exception: - errors.report("Error making Real-ESRGAN models list", exc_info=True) +def get_realesrgan_models(scaler: UpscalerRealESRGAN): + return [ + UpscalerData( + name="R-ESRGAN General 4xV3", + path="https://github.com/xinntao/Real-ESRGAN/releases/download/v0.2.5.0/realesr-general-x4v3.pth", + scale=4, + upscaler=scaler, + ), + UpscalerData( + name="R-ESRGAN General WDN 4xV3", + path="https://github.com/xinntao/Real-ESRGAN/releases/download/v0.2.5.0/realesr-general-wdn-x4v3.pth", + scale=4, + upscaler=scaler, + ), + UpscalerData( + name="R-ESRGAN AnimeVideo", + path="https://github.com/xinntao/Real-ESRGAN/releases/download/v0.2.5.0/realesr-animevideov3.pth", + scale=4, + upscaler=scaler, + ), + UpscalerData( + name="R-ESRGAN 4x+", + path="https://github.com/xinntao/Real-ESRGAN/releases/download/v0.1.0/RealESRGAN_x4plus.pth", + scale=4, + upscaler=scaler, + ), + UpscalerData( + name="R-ESRGAN 4x+ Anime6B", + path="https://github.com/xinntao/Real-ESRGAN/releases/download/v0.2.2.4/RealESRGAN_x4plus_anime_6B.pth", + scale=4, + upscaler=scaler, + ), + UpscalerData( + name="R-ESRGAN 2x+", + path="https://github.com/xinntao/Real-ESRGAN/releases/download/v0.2.1/RealESRGAN_x2plus.pth", + scale=2, + upscaler=scaler, + ), + ] diff --git a/modules/sysinfo.py b/modules/sysinfo.py index b669edd0c..5abf616b7 100644 --- a/modules/sysinfo.py +++ b/modules/sysinfo.py @@ -26,11 +26,9 @@ environment_whitelist = { "OPENCLIP_PACKAGE", "STABLE_DIFFUSION_REPO", "K_DIFFUSION_REPO", - "CODEFORMER_REPO", "BLIP_REPO", "STABLE_DIFFUSION_COMMIT_HASH", "K_DIFFUSION_COMMIT_HASH", - "CODEFORMER_COMMIT_HASH", "BLIP_COMMIT_HASH", "COMMANDLINE_ARGS", "IGNORE_CMD_ARGS_ERRORS", diff --git a/modules/upscaler.py b/modules/upscaler.py index b256e085b..3aee69db8 100644 --- a/modules/upscaler.py +++ b/modules/upscaler.py @@ -98,6 +98,9 @@ class UpscalerData: self.scale = scale self.model = model + def __repr__(self): + return f"" + class UpscalerNone(Upscaler): name = "None" diff --git a/requirements.txt b/requirements.txt index 80b438455..36f5674ad 100644 --- a/requirements.txt +++ b/requirements.txt @@ -6,6 +6,7 @@ basicsr blendmodes clean-fid einops +facexlib fastapi>=0.90.1 gfpgan gradio==3.41.2 @@ -20,13 +21,11 @@ open-clip-torch piexif psutil pytorch_lightning -realesrgan requests resize-right safetensors scikit-image>=0.19 -timm tomesd torch torchdiffeq diff --git a/requirements_versions.txt b/requirements_versions.txt index cb7403a9d..042fa708c 100644 --- a/requirements_versions.txt +++ b/requirements_versions.txt @@ -5,6 +5,7 @@ basicsr==1.4.2 blendmodes==2022 clean-fid==0.1.35 einops==0.4.1 +facexlib==0.3.0 fastapi==0.94.0 gfpgan==1.3.8 gradio==3.41.2 @@ -19,11 +20,10 @@ open-clip-torch==2.20.0 piexif==1.1.3 psutil==5.9.5 pytorch_lightning==1.9.4 -realesrgan==0.3.0 resize-right==0.0.2 safetensors==0.3.1 scikit-image==0.21.0 -timm==0.9.2 +spandrel==0.1.6 tomesd==0.1.3 torch torchdiffeq==0.2.3 From b621a63cf68c788487684250856707cb352b82d0 Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Mon, 25 Dec 2023 23:01:02 +0200 Subject: [PATCH 307/342] Unify CodeFormer and GFPGAN restoration backends, use Spandrel for GFPGAN --- .github/workflows/run_tests.yaml | 8 ++ .gitignore | 1 + modules/codeformer_model.py | 158 ++++++++--------------------- modules/face_restoration_utils.py | 163 ++++++++++++++++++++++++++++++ modules/gfpgan_model.py | 154 +++++++++------------------- requirements.txt | 1 - requirements_versions.txt | 1 - test/conftest.py | 15 ++- test/test_face_restorers.py | 29 ++++++ test/test_files/two-faces.jpg | Bin 0 -> 14768 bytes test/test_outputs/.gitkeep | 0 11 files changed, 302 insertions(+), 228 deletions(-) create mode 100644 modules/face_restoration_utils.py create mode 100644 test/test_face_restorers.py create mode 100644 test/test_files/two-faces.jpg create mode 100644 test/test_outputs/.gitkeep diff --git a/.github/workflows/run_tests.yaml b/.github/workflows/run_tests.yaml index 3dafaf8dc..cd5c3f868 100644 --- a/.github/workflows/run_tests.yaml +++ b/.github/workflows/run_tests.yaml @@ -20,6 +20,12 @@ jobs: cache-dependency-path: | **/requirements*txt launch.py + - name: Cache models + id: cache-models + uses: actions/cache@v3 + with: + path: models + key: "2023-12-30" - name: Install test dependencies run: pip install wait-for-it -r requirements-test.txt env: @@ -33,6 +39,8 @@ jobs: TORCH_INDEX_URL: https://download.pytorch.org/whl/cpu WEBUI_LAUNCH_LIVE_OUTPUT: "1" PYTHONUNBUFFERED: "1" + - name: Print installed packages + run: pip freeze - name: Start test server run: > python -m coverage run diff --git a/.gitignore b/.gitignore index 09734267f..6790e9ee7 100644 --- a/.gitignore +++ b/.gitignore @@ -37,3 +37,4 @@ notification.mp3 /node_modules /package-lock.json /.coverage* +/test/test_outputs diff --git a/modules/codeformer_model.py b/modules/codeformer_model.py index 517eadfd8..ceda4bab9 100644 --- a/modules/codeformer_model.py +++ b/modules/codeformer_model.py @@ -1,140 +1,62 @@ -import os +from __future__ import annotations + +import logging -import cv2 import torch -import modules.face_restoration -import modules.shared -from modules import shared, devices, modelloader, errors -from modules.paths import models_path +from modules import ( + devices, + errors, + face_restoration, + face_restoration_utils, + modelloader, + shared, +) + +logger = logging.getLogger(__name__) -model_dir = "Codeformer" -model_path = os.path.join(models_path, model_dir) model_url = 'https://github.com/sczhou/CodeFormer/releases/download/v0.1.0/codeformer.pth' +model_download_name = 'codeformer-v0.1.0.pth' -codeformer = None +# used by e.g. postprocessing_codeformer.py +codeformer: face_restoration.FaceRestoration | None = None -class FaceRestorerCodeFormer(modules.face_restoration.FaceRestoration): +class FaceRestorerCodeFormer(face_restoration_utils.CommonFaceRestoration): def name(self): return "CodeFormer" - def __init__(self, dirname): - self.net = None - self.face_helper = None - self.cmd_dir = dirname - - def create_models(self): - from facexlib.detection import retinaface - from facexlib.utils.face_restoration_helper import FaceRestoreHelper - - if self.net is not None and self.face_helper is not None: - self.net.to(devices.device_codeformer) - return self.net, self.face_helper - model_paths = modelloader.load_models( - model_path, - model_url, - self.cmd_dir, - download_name='codeformer-v0.1.0.pth', + def load_net(self) -> torch.Module: + for model_path in modelloader.load_models( + model_path=self.model_path, + model_url=model_url, + command_path=self.model_path, + download_name=model_download_name, ext_filter=['.pth'], - ) + ): + return modelloader.load_spandrel_model( + model_path, + device=devices.device_codeformer, + ).model + raise ValueError("No codeformer model found") - if len(model_paths) != 0: - ckpt_path = model_paths[0] - else: - print("Unable to load codeformer model.") - return None, None - net = modelloader.load_spandrel_model(ckpt_path, device=devices.device_codeformer) + def get_device(self): + return devices.device_codeformer - if hasattr(retinaface, 'device'): - retinaface.device = devices.device_codeformer + def restore(self, np_image, w: float | None = None): + if w is None: + w = getattr(shared.opts, "code_former_weight", 0.5) - face_helper = FaceRestoreHelper( - upscale_factor=1, - face_size=512, - crop_ratio=(1, 1), - det_model='retinaface_resnet50', - save_ext='png', - use_parse=True, - device=devices.device_codeformer, - ) + def restore_face(cropped_face_t): + assert self.net is not None + return self.net(cropped_face_t, w=w, adain=True)[0] - self.net = net - self.face_helper = face_helper - - def send_model_to(self, device): - self.net.to(device) - self.face_helper.face_det.to(device) - self.face_helper.face_parse.to(device) - - def restore(self, np_image, w=None): - from torchvision.transforms.functional import normalize - from basicsr.utils import img2tensor, tensor2img - np_image = np_image[:, :, ::-1] - - original_resolution = np_image.shape[0:2] - - self.create_models() - if self.net is None or self.face_helper is None: - return np_image - - self.send_model_to(devices.device_codeformer) - - self.face_helper.clean_all() - self.face_helper.read_image(np_image) - self.face_helper.get_face_landmarks_5(only_center_face=False, resize=640, eye_dist_threshold=5) - self.face_helper.align_warp_face() - - for cropped_face in self.face_helper.cropped_faces: - cropped_face_t = img2tensor(cropped_face / 255., bgr2rgb=True, float32=True) - normalize(cropped_face_t, (0.5, 0.5, 0.5), (0.5, 0.5, 0.5), inplace=True) - cropped_face_t = cropped_face_t.unsqueeze(0).to(devices.device_codeformer) - - try: - with torch.no_grad(): - res = self.net(cropped_face_t, w=w if w is not None else shared.opts.code_former_weight, adain=True) - if isinstance(res, tuple): - output = res[0] - else: - output = res - if not isinstance(res, torch.Tensor): - raise TypeError(f"Expected torch.Tensor, got {type(res)}") - restored_face = tensor2img(output, rgb2bgr=True, min_max=(-1, 1)) - del output - devices.torch_gc() - except Exception: - errors.report('Failed inference for CodeFormer', exc_info=True) - restored_face = tensor2img(cropped_face_t, rgb2bgr=True, min_max=(-1, 1)) - - restored_face = restored_face.astype('uint8') - self.face_helper.add_restored_face(restored_face) - - self.face_helper.get_inverse_affine(None) - - restored_img = self.face_helper.paste_faces_to_input_image() - restored_img = restored_img[:, :, ::-1] - - if original_resolution != restored_img.shape[0:2]: - restored_img = cv2.resize( - restored_img, - (0, 0), - fx=original_resolution[1]/restored_img.shape[1], - fy=original_resolution[0]/restored_img.shape[0], - interpolation=cv2.INTER_LINEAR, - ) - - self.face_helper.clean_all() - - if shared.opts.face_restoration_unload: - self.send_model_to(devices.cpu) - - return restored_img + return self.restore_with_helper(np_image, restore_face) -def setup_model(dirname): - os.makedirs(model_path, exist_ok=True) +def setup_model(dirname: str) -> None: + global codeformer try: - global codeformer codeformer = FaceRestorerCodeFormer(dirname) shared.face_restorers.append(codeformer) except Exception: diff --git a/modules/face_restoration_utils.py b/modules/face_restoration_utils.py new file mode 100644 index 000000000..c65c85ef8 --- /dev/null +++ b/modules/face_restoration_utils.py @@ -0,0 +1,163 @@ +from __future__ import annotations + +import logging +import os +from functools import cached_property +from typing import TYPE_CHECKING, Callable + +import cv2 +import numpy as np +import torch + +from modules import devices, errors, face_restoration, shared + +if TYPE_CHECKING: + from facexlib.utils.face_restoration_helper import FaceRestoreHelper + +logger = logging.getLogger(__name__) + + +def create_face_helper(device) -> FaceRestoreHelper: + from facexlib.detection import retinaface + from facexlib.utils.face_restoration_helper import FaceRestoreHelper + if hasattr(retinaface, 'device'): + retinaface.device = device + return FaceRestoreHelper( + upscale_factor=1, + face_size=512, + crop_ratio=(1, 1), + det_model='retinaface_resnet50', + save_ext='png', + use_parse=True, + device=device, + ) + + +def restore_with_face_helper( + np_image: np.ndarray, + face_helper: FaceRestoreHelper, + restore_face: Callable[[np.ndarray], np.ndarray], +) -> np.ndarray: + """ + Find faces in the image using face_helper, restore them using restore_face, and paste them back into the image. + + `restore_face` should take a cropped face image and return a restored face image. + """ + from basicsr.utils import img2tensor, tensor2img + from torchvision.transforms.functional import normalize + np_image = np_image[:, :, ::-1] + original_resolution = np_image.shape[0:2] + + try: + logger.debug("Detecting faces...") + face_helper.clean_all() + face_helper.read_image(np_image) + face_helper.get_face_landmarks_5(only_center_face=False, resize=640, eye_dist_threshold=5) + face_helper.align_warp_face() + logger.debug("Found %d faces, restoring", len(face_helper.cropped_faces)) + for cropped_face in face_helper.cropped_faces: + cropped_face_t = img2tensor(cropped_face / 255., bgr2rgb=True, float32=True) + normalize(cropped_face_t, (0.5, 0.5, 0.5), (0.5, 0.5, 0.5), inplace=True) + cropped_face_t = cropped_face_t.unsqueeze(0).to(devices.device_codeformer) + + try: + with torch.no_grad(): + restored_face = tensor2img( + restore_face(cropped_face_t), + rgb2bgr=True, + min_max=(-1, 1), + ) + devices.torch_gc() + except Exception: + errors.report('Failed face-restoration inference', exc_info=True) + restored_face = tensor2img(cropped_face_t, rgb2bgr=True, min_max=(-1, 1)) + + restored_face = restored_face.astype('uint8') + face_helper.add_restored_face(restored_face) + + logger.debug("Merging restored faces into image") + face_helper.get_inverse_affine(None) + img = face_helper.paste_faces_to_input_image() + img = img[:, :, ::-1] + if original_resolution != img.shape[0:2]: + img = cv2.resize( + img, + (0, 0), + fx=original_resolution[1] / img.shape[1], + fy=original_resolution[0] / img.shape[0], + interpolation=cv2.INTER_LINEAR, + ) + logger.debug("Face restoration complete") + finally: + face_helper.clean_all() + return img + + +class CommonFaceRestoration(face_restoration.FaceRestoration): + net: torch.Module | None + model_url: str + model_download_name: str + + def __init__(self, model_path: str): + super().__init__() + self.net = None + self.model_path = model_path + os.makedirs(model_path, exist_ok=True) + + @cached_property + def face_helper(self) -> FaceRestoreHelper: + return create_face_helper(self.get_device()) + + def send_model_to(self, device): + if self.net: + logger.debug("Sending %s to %s", self.net, device) + self.net.to(device) + if self.face_helper: + logger.debug("Sending face helper to %s", device) + self.face_helper.face_det.to(device) + self.face_helper.face_parse.to(device) + + def get_device(self): + raise NotImplementedError("get_device must be implemented by subclasses") + + def load_net(self) -> torch.Module: + raise NotImplementedError("load_net must be implemented by subclasses") + + def restore_with_helper( + self, + np_image: np.ndarray, + restore_face: Callable[[np.ndarray], np.ndarray], + ) -> np.ndarray: + try: + if self.net is None: + self.net = self.load_net() + except Exception: + logger.warning("Unable to load face-restoration model", exc_info=True) + return np_image + + try: + self.send_model_to(self.get_device()) + return restore_with_face_helper(np_image, self.face_helper, restore_face) + finally: + if shared.opts.face_restoration_unload: + self.send_model_to(devices.cpu) + + +def patch_facexlib(dirname: str) -> None: + import facexlib.detection + import facexlib.parsing + + det_facex_load_file_from_url = facexlib.detection.load_file_from_url + par_facex_load_file_from_url = facexlib.parsing.load_file_from_url + + def update_kwargs(kwargs): + return dict(kwargs, save_dir=dirname, model_dir=None) + + def facex_load_file_from_url(**kwargs): + return det_facex_load_file_from_url(**update_kwargs(kwargs)) + + def facex_load_file_from_url2(**kwargs): + return par_facex_load_file_from_url(**update_kwargs(kwargs)) + + facexlib.detection.load_file_from_url = facex_load_file_from_url + facexlib.parsing.load_file_from_url = facex_load_file_from_url2 diff --git a/modules/gfpgan_model.py b/modules/gfpgan_model.py index 6b6f17c43..a356b56fe 100644 --- a/modules/gfpgan_model.py +++ b/modules/gfpgan_model.py @@ -1,126 +1,68 @@ +from __future__ import annotations + +import logging import os -import modules.face_restoration -from modules import paths, shared, devices, modelloader, errors +from modules import ( + devices, + errors, + face_restoration, + face_restoration_utils, + modelloader, + shared, +) -model_dir = "GFPGAN" -user_path = None -model_path = os.path.join(paths.models_path, model_dir) -model_file_path = None +logger = logging.getLogger(__name__) model_url = "https://github.com/TencentARC/GFPGAN/releases/download/v1.3.0/GFPGANv1.4.pth" -have_gfpgan = False -loaded_gfpgan_model = None +model_download_name = "GFPGANv1.4.pth" +gfpgan_face_restorer: face_restoration.FaceRestoration | None = None -def gfpgann(): - global loaded_gfpgan_model - global model_path - global model_file_path - if loaded_gfpgan_model is not None: - loaded_gfpgan_model.gfpgan.to(devices.device_gfpgan) - return loaded_gfpgan_model +class FaceRestorerGFPGAN(face_restoration_utils.CommonFaceRestoration): + def name(self): + return "GFPGAN" - if gfpgan_constructor is None: - return None + def get_device(self): + return devices.device_gfpgan - models = modelloader.load_models(model_path, model_url, user_path, ext_filter=['.pth']) + def load_net(self) -> None: + for model_path in modelloader.load_models( + model_path=self.model_path, + model_url=model_url, + command_path=self.model_path, + download_name=model_download_name, + ext_filter=['.pth'], + ): + if 'GFPGAN' in os.path.basename(model_path): + net = modelloader.load_spandrel_model( + model_path, + device=self.get_device(), + ).model + net.different_w = True # see https://github.com/chaiNNer-org/spandrel/pull/81 + return net + raise ValueError("No GFPGAN model found") - if len(models) == 1 and models[0].startswith("http"): - model_file = models[0] - elif len(models) != 0: - gfp_models = [] - for item in models: - if 'GFPGAN' in os.path.basename(item): - gfp_models.append(item) - latest_file = max(gfp_models, key=os.path.getctime) - model_file = latest_file - else: - print("Unable to load gfpgan model!") - return None + def restore(self, np_image): + def restore_face(cropped_face_t): + assert self.net is not None + return self.net(cropped_face_t, return_rgb=False)[0] - import facexlib.detection.retinaface - - if hasattr(facexlib.detection.retinaface, 'device'): - facexlib.detection.retinaface.device = devices.device_gfpgan - model_file_path = model_file - model = gfpgan_constructor(model_path=model_file, upscale=1, arch='clean', channel_multiplier=2, bg_upsampler=None, device=devices.device_gfpgan) - loaded_gfpgan_model = model - - return model - - -def send_model_to(model, device): - model.gfpgan.to(device) - model.face_helper.face_det.to(device) - model.face_helper.face_parse.to(device) + return self.restore_with_helper(np_image, restore_face) def gfpgan_fix_faces(np_image): - model = gfpgann() - if model is None: - return np_image - - send_model_to(model, devices.device_gfpgan) - - np_image_bgr = np_image[:, :, ::-1] - cropped_faces, restored_faces, gfpgan_output_bgr = model.enhance(np_image_bgr, has_aligned=False, only_center_face=False, paste_back=True) - np_image = gfpgan_output_bgr[:, :, ::-1] - - model.face_helper.clean_all() - - if shared.opts.face_restoration_unload: - send_model_to(model, devices.cpu) - + if gfpgan_face_restorer: + return gfpgan_face_restorer.restore(np_image) + logger.warning("GFPGAN face restorer not set up") return np_image -gfpgan_constructor = None +def setup_model(dirname: str) -> None: + global gfpgan_face_restorer - -def setup_model(dirname): try: - os.makedirs(model_path, exist_ok=True) - import gfpgan - import facexlib.detection - import facexlib.parsing - - global user_path - global have_gfpgan - global gfpgan_constructor - global model_file_path - - facexlib_path = model_path - - if dirname is not None: - facexlib_path = dirname - - load_file_from_url_orig = gfpgan.utils.load_file_from_url - facex_load_file_from_url_orig = facexlib.detection.load_file_from_url - facex_load_file_from_url_orig2 = facexlib.parsing.load_file_from_url - - def my_load_file_from_url(**kwargs): - return load_file_from_url_orig(**dict(kwargs, model_dir=model_file_path)) - - def facex_load_file_from_url(**kwargs): - return facex_load_file_from_url_orig(**dict(kwargs, save_dir=facexlib_path, model_dir=None)) - - def facex_load_file_from_url2(**kwargs): - return facex_load_file_from_url_orig2(**dict(kwargs, save_dir=facexlib_path, model_dir=None)) - - gfpgan.utils.load_file_from_url = my_load_file_from_url - facexlib.detection.load_file_from_url = facex_load_file_from_url - facexlib.parsing.load_file_from_url = facex_load_file_from_url2 - user_path = dirname - have_gfpgan = True - gfpgan_constructor = gfpgan.GFPGANer - - class FaceRestorerGFPGAN(modules.face_restoration.FaceRestoration): - def name(self): - return "GFPGAN" - - def restore(self, np_image): - return gfpgan_fix_faces(np_image) - - shared.face_restorers.append(FaceRestorerGFPGAN()) + face_restoration_utils.patch_facexlib(dirname) + gfpgan_face_restorer = FaceRestorerGFPGAN(model_path=dirname) + shared.face_restorers.append(gfpgan_face_restorer) except Exception: errors.report("Error setting up GFPGAN", exc_info=True) diff --git a/requirements.txt b/requirements.txt index 36f5674ad..b1329c9e3 100644 --- a/requirements.txt +++ b/requirements.txt @@ -8,7 +8,6 @@ clean-fid einops facexlib fastapi>=0.90.1 -gfpgan gradio==3.41.2 inflection jsonmerge diff --git a/requirements_versions.txt b/requirements_versions.txt index 042fa708c..edbb6db9e 100644 --- a/requirements_versions.txt +++ b/requirements_versions.txt @@ -7,7 +7,6 @@ clean-fid==0.1.35 einops==0.4.1 facexlib==0.3.0 fastapi==0.94.0 -gfpgan==1.3.8 gradio==3.41.2 httpcore==0.15 inflection==0.5.1 diff --git a/test/conftest.py b/test/conftest.py index 31a5d9eaf..e4fc56785 100644 --- a/test/conftest.py +++ b/test/conftest.py @@ -1,10 +1,16 @@ +import base64 import os import pytest -import base64 - test_files_path = os.path.dirname(__file__) + "/test_files" +test_outputs_path = os.path.dirname(__file__) + "/test_outputs" + + +def pytest_configure(config): + # We don't want to fail on Py.test command line arguments being + # parsed by webui: + os.environ.setdefault("IGNORE_CMD_ARGS_ERRORS", "1") def file_to_base64(filename): @@ -23,3 +29,8 @@ def img2img_basic_image_base64() -> str: @pytest.fixture(scope="session") # session so we don't read this over and over def mask_basic_image_base64() -> str: return file_to_base64(os.path.join(test_files_path, "mask_basic.png")) + + +@pytest.fixture(scope="session") +def initialize() -> None: + import webui # noqa: F401 diff --git a/test/test_face_restorers.py b/test/test_face_restorers.py new file mode 100644 index 000000000..7760d51bf --- /dev/null +++ b/test/test_face_restorers.py @@ -0,0 +1,29 @@ +import os +from test.conftest import test_files_path, test_outputs_path + +import numpy as np +import pytest +from PIL import Image + + +@pytest.mark.usefixtures("initialize") +@pytest.mark.parametrize("restorer_name", ["gfpgan", "codeformer"]) +def test_face_restorers(restorer_name): + from modules import shared + + if restorer_name == "gfpgan": + from modules import gfpgan_model + gfpgan_model.setup_model(shared.cmd_opts.gfpgan_models_path) + restorer = gfpgan_model.gfpgan_fix_faces + elif restorer_name == "codeformer": + from modules import codeformer_model + codeformer_model.setup_model(shared.cmd_opts.codeformer_models_path) + restorer = codeformer_model.codeformer.restore + else: + raise NotImplementedError("...") + img = Image.open(os.path.join(test_files_path, "two-faces.jpg")) + np_img = np.array(img, dtype=np.uint8) + fixed_image = restorer(np_img) + assert fixed_image.shape == np_img.shape + assert not np.allclose(fixed_image, np_img) # should have visibly changed + Image.fromarray(fixed_image).save(os.path.join(test_outputs_path, f"{restorer_name}.png")) diff --git a/test/test_files/two-faces.jpg b/test/test_files/two-faces.jpg new file mode 100644 index 0000000000000000000000000000000000000000..c9d1b01032a7298d76608c8b65cbb243463491c5 GIT binary patch literal 14768 zcmbVz^LHju)9n-6wr$(Cp4hfEu`#i2+t$QRCUz#~#7-u-dEfhe_Yb(${adZmeb(t! zyQ=oC{#yIm1t7~x%18o0KmY)c?+fs?2?zy%{yzf)0|f&E2m3Z~Nbvs{Bn$*3#PQS_fQ*KMf`*NSO+-Ws^8eOpBU(=V^Q~-H4M7#g@dO9zX4+!d2gNfw(CEqw0*5$;BaF${~6H~tc~o# z_=CaCf^y3T4B`tXfI!ujEhIeVg!1U-Bn4;cI5D`Ju6(vFJnmH}txN}~biq~yee(OMc zX_Jzrvr-8Bw`n3>f?;Cgig5eMy#^l4FnC!wf9R7c|hQBhx+$hVSM_`$oH z3~XPl^$2<}H+=zm8571hkCPGKCRb?{E$vDwNlQL-5iK&L{w#Z#3EA3TYvG*EMkU1A zk*f@MRjxZwZPcwl6=$wCx~_7voQc{@VAPv_1Mcv z`?M)V>uVmKNSu~-A1(SP)dkwC{A$JfZ7G=XUeT`6z|i2-qhh%}ISi2tHw>$lOqOhj z80$M#{939h#mRz#1mq9@yzVS2@>OrDJNzd{hNN*{6dH@~z%; z?_|t|JKV$;ai&gSEjH!3z?qngyNPx;f$?n*?QUUSDi&9I^k8C~cmHpV1`+%KDCB>J6^5GDC*R!s<)wn-EW>O=K2v@>)eR0r(CaP{?EQG%`~9QN0k3O&OD5n zv+O>Q0h3CtJ%_WCHSASGj^tjKYdaX8(@a6GE0!9gP@_-W9V_|D<>N*&PpqLjpp zWEAv+GrSV**8AFNbkaO0^FH6{68!3zx~{P%{TILa$ym*;B3zP?WGxg*{gtk*mtmmp-mix7hY4GYsC|j-3b9cuH>`3G?5;DxwaK1qAg2p zk8a@rb3dz0fy*sq(x?ljVJ>OT7eL@mCbhr9u`$wC$Yw~s@IsxUMdfN?u3VG<;aQ0G zS2l*kKwNjpZ9X#|st`#V)cI3d!V_$b$vs{X@I;`tK18k2>ylT;IBD23zamiA{*(U5l0p3NCg<+fZlkL+ghzHTA$I+6~3etKo*LhX2k28WLQ zPtwLRF`0J6?w5a?0(Q{OvN5?CDGR|&h@|VxmCFk)3;m-}F(#HV&AG?4gO2=E5sA#P zsPc4U82o9>P`Cyi_TgI&;PKHnhNmA-R2(9qQGfONaKDvbHV2~}ug|=&wH2$hD6`$3 zFQjzIxq8<{JXwmkNrf_#XA)JDz(fy`1Uk8YAN^9#E={NSy#8-l0)*qj0G3+aRf->6 zUUc^S7eL07CZ64MZ5^&YOG+f#6v90PSA8S5cSi=rHAlIo&JJp>ckZ?a!IIIyw&Y&U zLBVz-FS=6h@?RJ!Mw=(~>DE(cpsS3kXfXbb*HgO39vvZ7tZe+PJ3m9OU$TMPZ<1fj z&yMmWXa2r;988u&QX#9T2Ketp?@aH8FUrNST=ne6e2IsuuqJ9p(OT|XE9>a0jP|4uJ+r?97atpquh)gM_oaT9Ft zcPG=KTYk`BE0!xc{HJSk;2PBppVp+CcJix>YQ9Aud*-p-2$iGhs_dG)_E{WB@e9!Y z7wJS(YA?JLK}u@)9^W2A+_j+M!E0gdqlt2=b5gtToMp+i%i5VX9c03-hns&fj%Md>yZE&1nx4l*d{?ORjPx|pXltABl>?3+$hB!l$(G2hl<(` zZBlKse0pW#g-`bg#(pz->grd0))GLrOn5g$2(e&2c`-Pk#puFQ_go5^_ncMp5sC#J zs}F`RL(Wwkb^WP4X+cHMfepct2*70F~zf38+^r1FWXYI;duNz-}dbH%dF=X$Yx zqDsf1A-~)%c1lp2nkAP7@sqt-FkolJt^5-Isyao9V7HEMgR%HU+0POILSrL2zqFW} zaLdH0)c*2m8>)nNRD;Qw%_{Ma3)Rc%5|~zXTF-@C^L>`$&ZC1X{dzL0-^sRx#CdXU z>vmXjS}m#vY!s?sG(W#g)xHLn4H*Ujm^$a?&{UZSU8wi+b)O3vxFt=le2mPpf{BmR zdO=OMY9FP5&0YAu^m4TvL%+F1=os$#@E3rob#7&}YGw+(qMKqrqrj8Q1;*q}eZADq@js(w)d+ zhu#TcSAp=J-BFGdMA5)ow1tP>=X4223u+f#KHGsB6?0L_xO>^T-IH`O4%uc46ZZLQ z(MToV8@NdJ{F??#C$ossHx!B>13gZ~4SU_u~HBsK$?Iuz56+i*5&dP`br zS;yQ2>gs%Wcy+?^WPb!l)QI-uyUKpXwN&{^95wbFUFFEhGL_K`sQi3(A(HUJ>V@fHUg)r%wQAIEp%wy~kU;62 z)nY-IzuO7{p-yLmY$lre!Yq~%D{~j+VDG7LT*An-mUNd{-q2TU&=A{}&D^`lh%puI zS2QoHTyxNx^g@4h%JPVgi&|UB`_)8yC?RYUxEoY2^UUHF=rD#r1S9bkaEL&jx6WI} zXL799Gy`{*Ik}~LDAOI9vL8OM&a>Uv;W*}I;W)MNmvPElyO=gwaP_&!)I@aQ3hdoj zw>e@-;Br-HC$C^^PDgLt~at{&lZA4JAmCqC4++}HIh z+o9z!!p}t*hJXptjHd2w)dH!-^9Fl7n;2?FC+vqp6!fZQ2`J||^ma8ELE^nwc4&V- zCDVx{egZk)5d>x1Vl0`n!A-5L=iaKfcGNww=nEin*4bm#z-0X{Fh?eu@GBto&F5_} z0u@r*oAwL_xI0ecoU%5VJd2WwY&qXj4b@XGpH=erdmKN=bgQ7DF1dRKHajdF5C2w4 z<}q-nSsDo%6at4311$UvkHE|O_s0zWa!FhRNS{wJDCHl?^@l3v=wXvAnv$T8Gr%&p zL10U#5kkm~9F1MdY=2{1y7DG%2ADMPTkru#2WURYedE+b0>U7%sBz~DRsKgd z{2{?kDRVJ^m7M~nGC~#xB$pDgo0vdsY#PHJZlx2yq#`APvUIpM#EZj)ANoTotLSPP z6doT9C$SBe{3zWQKzoI4i)AIYTwh9E0h~TI5CDp~=+5?DWqQ1>w$#w1mG$=-_{1Gf zI13@?zRxD1E~jFXeq)>YZV@LFao_$4RLo!d%VTBB?iV{m@v?aeZ02YZpK6G19jZ_# ztn!EzQRB{qt?V07M+^hD7pDqoKqx{b&^iXv1_VjmKoIm)VPl^W8Lb5G`^#_9Oo;SFq9`L@%b%&+Rww%V@4?GZRS@cRevcwZ+FiD zuI?Gtj)yOmk|_wE4ejUrehRqw%Q22&a_NAwFNoL;wLb*NJBZpBAPpT;@R9=@SwTT% z_r>nqEA8Dij4WfOFDXq&x-fux_xU&Et9aw(`9%AH38<{MkVNTjc3 z%iKj7d&eaq5iEML1RO)QI=@t>B>OD~T*u{V8qS6K+`5!m_7UH2CmPOs*|^>g`BikY zf;cK6hrzD2nWxb@cIlF7$Q_4m?gZt;*8DM`X4(Nn5r!IK| z=f*>nX^Kr1pRFA~zKYuk!IrlIJz$nIe=3N+03rg8EaF`cM8FOng9py5qbG@buGw2p6ZJbxK56z!CUs z=<0Z8ZGbdp@paj=FlGU{yjmk#wqZv&q3&&KXCjDCx0mp=&=9@Z^-e8N2@mNwVW!siGT zDqC@l15j%eV3YOf+3c=QkUg5eK9yr&V!rJ#bJ6fZ8(?yCDIuZF`&;`6s;{alDUXHD zlwcDxkP#2IwrYe0w}y+j@;lm?pSG|N{LkB(2UEBoO~rqNVfccvxy^0|kZ(-02A2^1 z5gqdzg!T^%=A$A-kt%XW3cuyixm&%#pYt)4;7~%OwYo4{{>{JgNUh4t%Pq4nGwygy z{`{z7{Nk&~f=JfSL~Ap5AZwf$VSpe8mi>HMrRAa|B8AQ~oOk@(@v1E*cZuq>F()~N zoR3^r`SXoabWKYk+j#52fNg;4$A1Ro?PFjR$aaRGZ?y1?IAD00myXv&mH!G=qW4j`9Oxo8#i`%L z@D<&&`Q!w^76Q>=;cI!KitXVZE@+hA*bL63hntH5&Yz`$d80D+umrSGWUxI z8{(}iTqP|#j$#AfQq@(^8mSd^R+GtMd_dis>Iy2&F-ccBm0cp-Xf0->@2*`bqfv@) z`~)dN)dgYGG$e2mvya@D6^YNzonbSna&LGw1*I$~NB1mm9tPQ?pt<0+ zNw8E}XQL$#$OUk|3D($@J~U}@4qR`wce7*LMy;gO1BE;-iO}j-!53g2xT@CHmS*wu z;K;)vq-%t^9vS$1-;UK+HgkxbGCxDlpTOtAfZCm3sPdggD!e%)ahW@1NaO>>!M`%2 zIs5GyGas|QTRM(qpe%NJZ9TVFokD=lnr7FOthK2@3!H{*k{w``Z@6cZ*iDic{|}E5 z&%|i^dRnSL5(#eM_?~B0XdASfE>`&>WAXf_eaf4}ZWB zSC(PDaDyJdPZlN57XUx!>hl~rJ3SE0Ux%I!vp5O2>c{q^eGbTzvT^HNYo@*eo!)6A zQI_p34xNk3sFz!lUEFTR`OX9grdYvMgBZ=z7FFKiKLsYT*1ttVhp`EB2iV;J%<**s zpIq32vXm8B@}#nz-l_nUOR%LXd|oV+f01>hj@tNDEZ!5>I`j<)_a^8WEw>!1j&8}AJ%-q=E){WY1Do8U1;QdaGz0$zhdjW*y$TkTaUg3`&tyH@>xl~y| z49<`8ogW_004KKdt`|Aqx6^F!c~v>03g_}GM8kt}+FZjXvtVb*{e-$oYD4~Pe8(Z+tp;yde1 zp4y?KCMh|M)cz(qiR5xLtt8Ia!QD*8Ckq#F1RP{roIHMmP~)p*k5d^T z)9+66_E^_mp_yyr!K?m0I+V!8mBkT$Mod=5z{*SAjsswbe1+)trb;tV7=xamj#O2@uSwjm*JV%tjM<66>fsR%TMOG7`G!mC zxp{Qaoaqc;S|qM;ng6T$qjv{2%7t`ZNjv1eH-5>n{L4S;x#>}_oO7&RoYrZgz@t+# zpZq+Bw=7BQ+6XE@{D(4M1*79y*#vrGK3!Pf*#qn3sGCpr)4UAI^2F%zxy=~z=8+-? zo8r^m2wdcAw{Vk{H^e`c+@aK*i3Q%Pt=#l3pU9<}E7$OAVx$RBbN3de#lt1J583_c zf%6y@mO5Mn>CskFEa#gk9(Iy^lyCdb$e$Uyf-o(7BKe`L8eWJVhn*w5R8YJ*!x?Vc`K7cRUS zm+0EH_ace>k6p|Q@=vcnCwDUSC?kH2&%2!w>hV_N;RR1H<(`OPyJ0{IH>( zU_XCDysr=$HI3^?mgOCsD%(r2X)h)Dm*PcZ3*+hBuD!mGXKx>Xt!{&dz8Qq-?9nVIj6ylw@`i z*$@XjkWFAvW8 zv^*9f{-w3TZ%K(gn_IJ-M>>C-tNbJk`P}|5As;zU>Xq&9oy~KKDvI6jt|@J$vUpUa zLTJG*q}+(fbI$!F75+`Gp6E}}$r(&lgyc8$t7;%GeRut`7Jb9;KcKo#nm3eJ?vuFr zYmp1lkwyx;n~~M`r@SVMi~eRwDUiQe(f_FK|CQJOqppJhP)JxrRE)ow(ZWXLft&fg z{}qjdko75Ir;y=i6=sr|d#vs8ESIl%mC1RfK*DCnwL?A%GBi4jaQV&AfSN3+`VyrZ zg+0=eR<8~!Aqib^K}H=Yc+sjcPA!dM%!wFc~YSadN!a z?MLQuvPF}Nb;thu$jBH_>OK_Wzb=TQ{YN}lyCD%p(hwqCs;%{9JoN|I&uIW(fFNk; z&??v%RV!+f$V!7W^L@;@5k}Hsm-XNjJM7FA*~*`~6>=O-OPm|#^HympQ9q5-J4Gz$ z8AKR$Q>w058{}QnWpo;qD@Brs$qzKKyJiN)!)kXAbgWs{R8SHQNEbS9F>hQS`hNm# z2*xxclM8-|!Cx9z^n*TNp;*TRg$x(iBcl%47~z-A4_{i)g;SIxLn)dDnmZ|UrmyYE zLbj~{q;Ihy^@dfC0!uNA8D(V0M^VRK47Rp7HONwXVA|RmG!ueqDHc@PbgxQ@bNjo5 zsX7k_2q3RWgPELuZiN(=r5*U04iJz{Fi44Uol9a*sUE{)atuS$6hIs!N!zas+)-(g zbrTDRqk*UElA975vw?<7qN~bcb;wWl%BnV5vY4ss`J%}KMzAbd@*88fH=?qE8VtRo z5X@3|TqT7f1+<8Nr0ZKtup+GiLIXjUMCtX z$U>YAScf#70Ke4|QW<83Uqrk)IhYuXL$*4eH~-#ACOU8pqZaB&Ni?H~B(ZWr$5jnB zA#1BUz{LJ0b;iZEhKc8lm8uG4HGqhStf3Fhb?R0th<=KgCdhb!5nkLiM1nL8yUl@W z{TvnQP-qyBHpQ*DUC7(!?@E6;CwYTR@3g!}#KsKNDpvjibdxi!8|7VTw8vek!n6Gl zL=guVN%_+E_3vSEGDlPxQ8A7XBFPNX4Vm!7#|=wXd0>+lb3}Gl_H z5L1k$Bhx7&vAk|Bs7Kb7+8xo@cQjV2fK{lq2*DO{G*^FkSvVNDmt*}xTU%8a!b;P0 zsB0ex)EPY8=aHY(VP1wfyO2SoI5Ph|xXKt3dYZvi!lT*8EP4T=u0I^!R)b2w zYRlot*w#sT7xd8W&t#Wj)h4T#&zRGn)Hv6D>kc{4m<3-ZGA(kL`D)!skE)F{+>!^O z^nMNt?Eyqw;+KNlTk_L-)MOOqyS=!D2cjpf%kQnF+@Hf60uJm&@3^LCCUIZ^6)=dZ zB*%7A*naCI_Yq`NJI*t&29KxQ6pL>h)>BZj%F17%DuCEyeDj^miYTde-$fN9f^|$GDYntaZ@Ua+vPJIhFRtNREeZkY z$ZvRWN=hR#emy3!3b_?yb$<}@U-HIgsu^VlF48f)6-4`NB05R?gEkE3`ZTob{w700 z{-T|rov7=xAIP?{SbV|#N*z?EdSb)i0O2E;cAtVEtO(@gEu&K>&i-MLrkYqM3A6}~ zB>H_5Q*DR60fAFi%=V&P`*=u}WS|}4v|^saDD=JDj)3 zsB}2=0Tbge%FOqm>?>RlkNzY(BCPXH1TP-P4|ip-ze2SGL0Y<{`_1E-%2QN>pCQV_`!eVh53<9%kyJ);4C^z? zE)by9wlyM~8tbsWmf#;yLwWr#f+m z%Z8`&PjF>`iTr?EJGg+)q&dJG9)>Q0q}DgsZ`h;Io(HVI*;8~s$YHr8jieU&>sCas z)#PVO?rJa2Qgi&m-j_|+P?oD$1^KaLG6`TsnvX0+C7RCt9n4FM53@N=`NuloeJvNG z>t@DBOC+4lXp0ig6s%O>K8%jAP^Zul3wqHE%xqdsEoB-t(OREE2+9)Z-^S_}o}=kw zy9dE=y{Am2FSiXbAe1kSh|AnkzHM{yTX3?~GF*v!*g zy=6q?QmE845+xO4CWK$kP9ite%LHCj8Pe40Px6H`Q`GYw)nLL1IwV1(e2rKh2% z0bPn-ooV%Nhc&N{*m@t0YuKlP>IHu5U(UG`6b4tC^ zj*0-A%V;Qx80nkg_(Gjd7c=Ur4;XC}&a_^kqKm?NJXv+Aa-O9qUeVrW8?@%uMsH)x zYq~|jl;)s)uXHIQCn2XXijom+7HAe=coQE2Kf|f^^7|K_gH+;zL7}^8y+_M({LC}FInu9)Z_;a)k-?MPV%Jv-Cn!y?*!k%d%a*Z@Hfl)U z7GRM_+OWHVeRJU)#%)CA6&;X?PKc^rsu>SJ;YusniL|R1FXRa+2%f&{i-Wc@JoN}$ zTC38@){YHkh9s<_Kr(_hsV8&l5irsTVA%&P1U#gB;)r@X4T*PHNnSihhMONo%6e4C%yBoI3lQwvFBmYuv@~fJx)8Yx1aKnNIjl;=a zqL#BuxbT6aHPXyoJWwQ!Xi#TbegtU|SIJlDTamywlFfk}rD+J(Hs-=aFp@7vuVD=_ zWnZYQ856b@M7p3n;qfOk8ZfYUQjlW-IOKfG8w7r6=VIL`Y(aFPuNpH4)7&2G6#a zxQw6z=+iJhQe}L)B}Ido9x)&Q1FVu(5U&q$iP4ohhVkOGX4b~vaWFZ`tJY!BO@WH~ zCR5zXZ2fAmJOEFv_?$>m6)MIE+R7CM@BOvRgR%{&OxUqSYrLtabAsefMw3QRfK8j|s+0eWg%;-q)z|jn`RCj5Usd#Gd?=@lCgpymXMiDq?T& z^9We0P5v(a=zt!inyXEdv1@UF{Q@YDtcM?=I?FgA&OtT9Lp5OrBqNt5je)WRXX%?p zKGVg$^o}{D9tIc#=$edbryl{u9j_&$G=n1ld?J6t>VuR)G{_A`2PAtz<+0|B-1lvx6orHoD5qNWJO&#U0aqI~r;s%nHm}g`u?a2zQ8=Bh zW>GXC3{{KsDYtK^Zi!b-tF9Cu6|Wdsb`9-g)F2uvcNiTw1C8b5?2S!W=dI+A3&HhH zhs{i8NtYF`ZU$;Ff8Rf5nzey!_fIFKMmkaV(BFtyxF$fc+}iyu8sxOlXdN|mYDk|M z5L|h`!sY~MI)&n=Jpx2`YN)S2a~tJWuFhyF=H=}au=}W?oq`7k`xok=*yx-*qhT^D zJ}WSbG?Pt+`xaDSG?OAcN-|d30?IJT*VRoa53I~$*O-{;Jz273WkGdEz{mLfPd=49 zq*$TuQFI$D-lK6Cfj7=*L$EF7T|p}NnF6d$UZIZgSExmQO$aO&ZkcwAH?7|>aEFlr z$AUnxbL4eR$;l2xnOX>=mj%Z~HZo%obwr74ClY;?nIlHL2#|H%(u6re?7*j)+3~Zi zEHgQNVaQ4aFF(JEuq+NwE!)>%X5HDqVVy~kBhbB7$<;$p=ow#A6H$B-7H)IMqo9;g zx~Nye{22!kLula;jIIG>OXz}lr!7KEIF3lS=I@S;4dr_r8VnTTd$tAQKT|E=9UCYZ z02~FC1On~5ZzDB!hAbpw6%B$yCwEE2R849e_;3FP3IZYo$iz!?;Tn+wV-~cr(wFHk zSNwQe{)N}mMxx&t2?^l%EbMe26vpUztp4dx?qJ&Jn71?%e_Rs6khe5XtyeKnVP-}g z6cL&9d5cXmK0{^%@Q$*M9!5tPkzy?BimxioV3+;76xEa=eFsSZA2fN@~74Uy&{S14FfisiGbDbfiWQICe$gqp1P-)arz9~%HgIIzt znWBM=9okf!^!B(H63k(6*%Wv{<*VXr}Fe;)Wx|y>!-#7$1$h5!%$^e)~-Wj#J>uC5)zAu zRH_7*vP>-p8WW_XZd{fWFm;A*`F?^ha#9XuKGj>98`dhC>#}6xAW>^y^&CfMW~jT^ z8EJ7^rJ$@Z^tZ#9=G+Mm>ab*Z_7_3A@X2OAGGuK29+gOv$qBvzfpLl!cUW5Z}kz@JNFViIVnqxl8lqXZxM>xp?qj^e?3#nYxRwI4cn zl({*;VaV>)x0m3L{w=X2;0&Wob%JvaYeG%wZ4nxTL37Ru^HkrG>q)d&Ln@cB!SAOA z>xIwYlmd<+h`9$L;BRR4_occk;@fPd*FUI0aPGWqUfPRE6<-V1or*5)RM6$VQaFr^3bE3CGAT70rWE@BBSS4;g5*ipAvUMNYjIQ`KOY)#6Mkvfiaee+ysDG|pWuNWgLMvknHuOF(x`jrIHN zu~izt8qfjmB;mjqWS^*geCka!eySU`j;oencrZFNO;6dAbl6=@q%5obPOWrcT!%`I zVw! zn+@SOR_W&(n?IlX3leE;9UsdP;RFN45Aln>9xHS&Mz%7P!msVz%3~D^E_6S?0EjQ; zNSb+r61{&#I0Hsv}uSxXIbts;uiZ4i4Zc(m< z>QXK=N;bXu63saWc);QFCF-X;1>H?}^V1&YI+z!g&ic|)_q}ft8dnC|5R8LhHiy$6 z0!=BHmD;_IgRmQldqNa90n3QxGHP$d>mBe0e4PD+2IEHJ3?pJ!#xJ`$zp4v6Wj{`@yd}bo-njyD(j*& zBVBc_60V*OuKo+LC0xOpsG8601S3Y$>aB$Qi8RMJj-RNHuFIzyI9Jl9N*5c-IJ8#x zm|MYeum3lc^!Kpceu6pNVLY#AB`?>daKZx0l3V>Z(zE{;>Hjms2nzRqe~I~Dq-TK; zK_gXhhD2vIMh+q)XA@O*`6lZI<|+OQ_20waLI4era9R`oTIs!TItrNwjynr;eHifMZP&ZU{(%*Z%LoiQ{Mp1 zbZJpr28Fn?lpLNQRt!#lO((i6(?!35tsjF*ECtjk1MPGOgCBKss>qFYFz7DkR{t01 zmPrH^JCY5x`>g>EzpDl&3@Y5uQQcIv%|A&zRO7KYjq4I}c4w2fGwyKY?x>^e z0liwA~t@uA2bM6Iv{2}kJ)d$BV(E#u1;1b0yq+ zuYT!KW+c9w?jFxUi+iM{*qw?Q56SUAyA&%QGbADK4Thauk+8uT^QP^uVfoM=>n3{D zxM8RUKlNtGUdwV5l!cGAi+-C1G1nocw_KmFdUV# z3@YUSyQrVV;K1hn3jm?cLp&#zvtKB-U$(v@l&U>|%Zmg1646L(e=2w!J$)S(gZ-Hc zZU(%wG6XLoIe>;-f2xjL&O|#tYTvCU21mINssE`wWNlH)*lt0k!F~vGrjN>f^(G60 zGdVn66S?8KxZ4 zB%|;0nz@o{ZX$#H$Hc(RXpL&=;e1+j*%-POPtqx-@u}!jX{vIM&G2gz_9`$!VmBo= z6H;hIzt6(kl3Yb4<~YyJKKFyB+=UqrY|dUJn+M)iAD&`NDQd5ujjk!}sj_wO{8Wht z+y#M3r6r5GU7v$c-eHUw?OeiLu^yw~M{Oc_d8kO3th%LT;sm|#aNgmHHa2S#j-m+0 z0`FNy2^wlUvE4!V*b4{4$l<|LNyXI|2SO*j{w9OIjes0_ROg)MSg8WH*QN@|1Qk5= z$)8Aa8_?#q{nmW_Gef`AR$n@9PEY1MRlQGnS}Bd{zpL#(9~xIA2b;9c$9e^PP4aA88&T>*|=^ z^tYZ=JJJVT^8BSQTs5{W{z%TVW~{wk{aR_(AF(vD-NRfJxFG*fsg8H&{sOqut3o>U zR#!>w>h>i4I>SFxQ7$`s!i|-Lj~nSwwhr8UTUBhoWg-(5d_>(7ZAS-O_E#PmIkk{v zd4eNEkY7OfYE*dl40zHe2Ky<}Tt1?*TuX-8+3r>LFjk@_2N9cnzk)@&*n)c;hkRA@ zoQTf@wDBfoFW^!aCBpZ9tf7S0D6zXyop8+CYK$;jQ$gtifo?J&n8t%=;;EFr&WV$K zJaf6+JEf5qr2he9=>*G9_DZFTy@ci;<6u+_kyz4BaLlJ Date: Wed, 27 Dec 2023 10:55:01 +0200 Subject: [PATCH 308/342] Add experimental HAT model --- modules/hat_model.py | 42 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 42 insertions(+) create mode 100644 modules/hat_model.py diff --git a/modules/hat_model.py b/modules/hat_model.py new file mode 100644 index 000000000..553e19411 --- /dev/null +++ b/modules/hat_model.py @@ -0,0 +1,42 @@ +import os +import sys + +from modules import modelloader, devices +from modules.shared import opts +from modules.upscaler import Upscaler, UpscalerData +from modules.upscaler_utils import upscale_with_model + + +class UpscalerHAT(Upscaler): + def __init__(self, dirname): + self.name = "HAT" + self.scalers = [] + self.user_path = dirname + super().__init__() + for file in self.find_models(ext_filter=[".pt", ".pth"]): + name = modelloader.friendly_name(file) + scale = 4 # TODO: scale might not be 4, but we can't know without loading the model + scaler_data = UpscalerData(name, file, upscaler=self, scale=scale) + self.scalers.append(scaler_data) + + def do_upscale(self, img, selected_model): + try: + model = self.load_model(selected_model) + except Exception as e: + print(f"Unable to load HAT model {selected_model}: {e}", file=sys.stderr) + return img + model.to(devices.device_esrgan) # TODO: should probably be device_hat + return upscale_with_model( + model, + img, + tile_size=opts.ESRGAN_tile, # TODO: should probably be HAT_tile + tile_overlap=opts.ESRGAN_tile_overlap, # TODO: should probably be HAT_tile_overlap + ) + + def load_model(self, path: str): + if not os.path.isfile(path): + raise FileNotFoundError(f"Model file {path} not found") + return modelloader.load_spandrel_model( + path, + device=devices.device_esrgan, # TODO: should probably be device_hat + ) From 4ad0c0c0a805da4bac03cff86ea17c25a1291546 Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Sat, 30 Dec 2023 16:37:03 +0200 Subject: [PATCH 309/342] Verify architecture for loaded Spandrel models --- extensions-builtin/ScuNET/scripts/scunet_model.py | 2 +- extensions-builtin/SwinIR/scripts/swinir_model.py | 1 + modules/codeformer_model.py | 1 + modules/esrgan_model.py | 1 + modules/gfpgan_model.py | 1 + modules/hat_model.py | 1 + modules/modelloader.py | 13 ++++++++++++- modules/realesrgan_model.py | 7 ++++--- 8 files changed, 22 insertions(+), 5 deletions(-) diff --git a/extensions-builtin/ScuNET/scripts/scunet_model.py b/extensions-builtin/ScuNET/scripts/scunet_model.py index 18cf8e1a0..5f3dd08b3 100644 --- a/extensions-builtin/ScuNET/scripts/scunet_model.py +++ b/extensions-builtin/ScuNET/scripts/scunet_model.py @@ -121,7 +121,7 @@ class UpscalerScuNET(modules.upscaler.Upscaler): filename = modelloader.load_file_from_url(self.model_url, model_dir=self.model_download_path, file_name=f"{self.name}.pth") else: filename = path - return modelloader.load_spandrel_model(filename, device=device) + return modelloader.load_spandrel_model(filename, device=device, expected_architecture='SCUNet') def on_ui_settings(): diff --git a/extensions-builtin/SwinIR/scripts/swinir_model.py b/extensions-builtin/SwinIR/scripts/swinir_model.py index 85c18b9e9..aae159af5 100644 --- a/extensions-builtin/SwinIR/scripts/swinir_model.py +++ b/extensions-builtin/SwinIR/scripts/swinir_model.py @@ -75,6 +75,7 @@ class UpscalerSwinIR(Upscaler): filename, device=self._get_device(), dtype=devices.dtype, + expected_architecture="SwinIR", ) if getattr(opts, 'SWIN_torch_compile', False): try: diff --git a/modules/codeformer_model.py b/modules/codeformer_model.py index ceda4bab9..44b84618e 100644 --- a/modules/codeformer_model.py +++ b/modules/codeformer_model.py @@ -37,6 +37,7 @@ class FaceRestorerCodeFormer(face_restoration_utils.CommonFaceRestoration): return modelloader.load_spandrel_model( model_path, device=devices.device_codeformer, + expected_architecture='CodeFormer', ).model raise ValueError("No codeformer model found") diff --git a/modules/esrgan_model.py b/modules/esrgan_model.py index a7c7c9e30..70041ab02 100644 --- a/modules/esrgan_model.py +++ b/modules/esrgan_model.py @@ -49,6 +49,7 @@ class UpscalerESRGAN(Upscaler): return modelloader.load_spandrel_model( filename, device=('cpu' if devices.device_esrgan.type == 'mps' else None), + expected_architecture='ESRGAN', ) diff --git a/modules/gfpgan_model.py b/modules/gfpgan_model.py index a356b56fe..48f8ad5e2 100644 --- a/modules/gfpgan_model.py +++ b/modules/gfpgan_model.py @@ -37,6 +37,7 @@ class FaceRestorerGFPGAN(face_restoration_utils.CommonFaceRestoration): net = modelloader.load_spandrel_model( model_path, device=self.get_device(), + expected_architecture='GFPGAN', ).model net.different_w = True # see https://github.com/chaiNNer-org/spandrel/pull/81 return net diff --git a/modules/hat_model.py b/modules/hat_model.py index 553e19411..7f2abb416 100644 --- a/modules/hat_model.py +++ b/modules/hat_model.py @@ -39,4 +39,5 @@ class UpscalerHAT(Upscaler): return modelloader.load_spandrel_model( path, device=devices.device_esrgan, # TODO: should probably be device_hat + expected_architecture='HAT', ) diff --git a/modules/modelloader.py b/modules/modelloader.py index 30116932a..f4182559e 100644 --- a/modules/modelloader.py +++ b/modules/modelloader.py @@ -6,6 +6,8 @@ import shutil import importlib from urllib.parse import urlparse +import torch + from modules import shared from modules.upscaler import Upscaler, UpscalerLanczos, UpscalerNearest, UpscalerNone from modules.paths import script_path, models_path @@ -183,9 +185,18 @@ def load_upscalers(): ) -def load_spandrel_model(path, *, device, half: bool = False, dtype=None): +def load_spandrel_model( + path: str, + *, + device: str | torch.device | None, + half: bool = False, + dtype: str | None = None, + expected_architecture: str | None = None, +): import spandrel model = spandrel.ModelLoader(device=device).load_from_file(path) + if expected_architecture and model.architecture != expected_architecture: + raise TypeError(f"Model {path} is not a {expected_architecture} model") if half: model = model.model.half() if dtype: diff --git a/modules/realesrgan_model.py b/modules/realesrgan_model.py index 332d8f4b1..2a2be5ad7 100644 --- a/modules/realesrgan_model.py +++ b/modules/realesrgan_model.py @@ -1,9 +1,9 @@ import os -from modules.upscaler_utils import upscale_with_model -from modules.upscaler import Upscaler, UpscalerData -from modules.shared import cmd_opts, opts from modules import modelloader, errors +from modules.shared import cmd_opts, opts +from modules.upscaler import Upscaler, UpscalerData +from modules.upscaler_utils import upscale_with_model class UpscalerRealESRGAN(Upscaler): @@ -40,6 +40,7 @@ class UpscalerRealESRGAN(Upscaler): info.local_data_path, device=self.device, half=(not cmd_opts.no_half and not cmd_opts.upcast_sampling), + expected_architecture="RealESRGAN", ) return upscale_with_model( mod, From 05230c02606080527b65ace9eacb6fb835239877 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sat, 30 Dec 2023 18:02:51 +0300 Subject: [PATCH 310/342] fix img2img api that i broke when implementing infotext support --- modules/api/api.py | 1 + 1 file changed, 1 insertion(+) diff --git a/modules/api/api.py b/modules/api/api.py index 2918f7857..2e18c6b91 100644 --- a/modules/api/api.py +++ b/modules/api/api.py @@ -507,6 +507,7 @@ class Api: args.pop('script_name', None) args.pop('script_args', None) # will refeed them to the pipeline directly after initializing them args.pop('alwayson_scripts', None) + args.pop('infotext', None) script_args = self.init_script_args(img2imgreq, self.default_script_arg_img2img, selectable_scripts, selectable_script_idx, script_runner, input_script_args=infotext_script_args) From f476649c02cf3547d891fa08c50a92f92c4d73bd Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Sat, 30 Dec 2023 17:41:19 +0200 Subject: [PATCH 311/342] Correct arg type for restore_face --- modules/face_restoration_utils.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/modules/face_restoration_utils.py b/modules/face_restoration_utils.py index c65c85ef8..85cb30570 100644 --- a/modules/face_restoration_utils.py +++ b/modules/face_restoration_utils.py @@ -36,7 +36,7 @@ def create_face_helper(device) -> FaceRestoreHelper: def restore_with_face_helper( np_image: np.ndarray, face_helper: FaceRestoreHelper, - restore_face: Callable[[np.ndarray], np.ndarray], + restore_face: Callable[[torch.Tensor], torch.Tensor], ) -> np.ndarray: """ Find faces in the image using face_helper, restore them using restore_face, and paste them back into the image. @@ -126,7 +126,7 @@ class CommonFaceRestoration(face_restoration.FaceRestoration): def restore_with_helper( self, np_image: np.ndarray, - restore_face: Callable[[np.ndarray], np.ndarray], + restore_face: Callable[[torch.Tensor], torch.Tensor], ) -> np.ndarray: try: if self.net is None: From 91560e98c47f8271d444556ef4ae6505dece9aba Mon Sep 17 00:00:00 2001 From: lanyeeee <1210347077@qq.com> Date: Sat, 30 Dec 2023 23:42:10 +0800 Subject: [PATCH 312/342] fix format issue --- modules/api/api.py | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) diff --git a/modules/api/api.py b/modules/api/api.py index 2f718ec26..d202cb8d2 100644 --- a/modules/api/api.py +++ b/modules/api/api.py @@ -418,7 +418,6 @@ class Api: task_id = txt2imgreq.force_task_id or create_task_id("txt2img") script_runner = scripts.scripts_txt2img - with self.txt2img_script_arg_init_lock: if not script_runner.scripts: script_runner.initialize_scripts(False) @@ -489,14 +488,13 @@ class Api: mask = decode_base64_to_image(mask) script_runner = scripts.scripts_img2img - with self.img2img_script_arg_init_lock: if not script_runner.scripts: script_runner.initialize_scripts(True) ui.create_ui() - infotext_script_args = {} - self.apply_infotext(img2imgreq, "img2img", script_runner=script_runner, mentioned_script_args=infotext_script_args) + infotext_script_args = {} + self.apply_infotext(img2imgreq, "img2img", script_runner=script_runner, mentioned_script_args=infotext_script_args) if not self.default_script_arg_img2img: self.default_script_arg_img2img = self.init_default_script_args(script_runner) From c9174253fb603e6b2552e4c2721fd767b6ede87d Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Sat, 30 Dec 2023 17:45:26 +0200 Subject: [PATCH 313/342] Drop dependency on basicsr --- modules/face_restoration_utils.py | 35 +++++++++++++++++++++++-------- requirements.txt | 1 - requirements_versions.txt | 1 - 3 files changed, 26 insertions(+), 11 deletions(-) diff --git a/modules/face_restoration_utils.py b/modules/face_restoration_utils.py index 85cb30570..1cbac2364 100644 --- a/modules/face_restoration_utils.py +++ b/modules/face_restoration_utils.py @@ -17,6 +17,28 @@ if TYPE_CHECKING: logger = logging.getLogger(__name__) +def bgr_image_to_rgb_tensor(img: np.ndarray) -> torch.Tensor: + """Convert a BGR NumPy image in [0..1] range to a PyTorch RGB float32 tensor.""" + assert img.shape[2] == 3, "image must be RGB" + if img.dtype == "float64": + img = img.astype("float32") + img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB) + return torch.from_numpy(img.transpose(2, 0, 1)).float() + + +def rgb_tensor_to_bgr_image(tensor: torch.Tensor, *, min_max=(0.0, 1.0)) -> np.ndarray: + """ + Convert a PyTorch RGB tensor in range `min_max` to a BGR NumPy image in [0..1] range. + """ + tensor = tensor.squeeze(0).float().detach().cpu().clamp_(*min_max) + tensor = (tensor - min_max[0]) / (min_max[1] - min_max[0]) + assert tensor.dim() == 3, "tensor must be RGB" + img_np = tensor.numpy().transpose(1, 2, 0) + if img_np.shape[2] == 1: # gray image, no RGB/BGR required + return np.squeeze(img_np, axis=2) + return cv2.cvtColor(img_np, cv2.COLOR_BGR2RGB) + + def create_face_helper(device) -> FaceRestoreHelper: from facexlib.detection import retinaface from facexlib.utils.face_restoration_helper import FaceRestoreHelper @@ -43,7 +65,6 @@ def restore_with_face_helper( `restore_face` should take a cropped face image and return a restored face image. """ - from basicsr.utils import img2tensor, tensor2img from torchvision.transforms.functional import normalize np_image = np_image[:, :, ::-1] original_resolution = np_image.shape[0:2] @@ -56,23 +77,19 @@ def restore_with_face_helper( face_helper.align_warp_face() logger.debug("Found %d faces, restoring", len(face_helper.cropped_faces)) for cropped_face in face_helper.cropped_faces: - cropped_face_t = img2tensor(cropped_face / 255., bgr2rgb=True, float32=True) + cropped_face_t = bgr_image_to_rgb_tensor(cropped_face / 255.0) normalize(cropped_face_t, (0.5, 0.5, 0.5), (0.5, 0.5, 0.5), inplace=True) cropped_face_t = cropped_face_t.unsqueeze(0).to(devices.device_codeformer) try: with torch.no_grad(): - restored_face = tensor2img( - restore_face(cropped_face_t), - rgb2bgr=True, - min_max=(-1, 1), - ) + cropped_face_t = restore_face(cropped_face_t) devices.torch_gc() except Exception: errors.report('Failed face-restoration inference', exc_info=True) - restored_face = tensor2img(cropped_face_t, rgb2bgr=True, min_max=(-1, 1)) - restored_face = restored_face.astype('uint8') + restored_face = rgb_tensor_to_bgr_image(cropped_face_t, min_max=(-1, 1)) + restored_face = (restored_face * 255.0).astype('uint8') face_helper.add_restored_face(restored_face) logger.debug("Merging restored faces into image") diff --git a/requirements.txt b/requirements.txt index b1329c9e3..731a1be7d 100644 --- a/requirements.txt +++ b/requirements.txt @@ -2,7 +2,6 @@ GitPython Pillow accelerate -basicsr blendmodes clean-fid einops diff --git a/requirements_versions.txt b/requirements_versions.txt index edbb6db9e..1e0ccafa7 100644 --- a/requirements_versions.txt +++ b/requirements_versions.txt @@ -1,7 +1,6 @@ GitPython==3.1.32 Pillow==9.5.0 accelerate==0.21.0 -basicsr==1.4.2 blendmodes==2022 clean-fid==0.1.35 einops==0.4.1 From b58ed1b2432c3c7643b39e53f7bb567ea8655aae Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Sat, 30 Dec 2023 18:02:01 +0200 Subject: [PATCH 314/342] Bump numpy to 1.26.2 This avoids it being downgraded during `launch.py` --- requirements_versions.txt | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/requirements_versions.txt b/requirements_versions.txt index edbb6db9e..7ec7abe2e 100644 --- a/requirements_versions.txt +++ b/requirements_versions.txt @@ -13,7 +13,7 @@ inflection==0.5.1 jsonmerge==1.8.0 kornia==0.6.7 lark==1.1.2 -numpy==1.23.5 +numpy==1.26.2 omegaconf==2.2.3 open-clip-torch==2.20.0 piexif==1.1.3 From f651405427dfc6d4ef96ecba7f9c2ceb580263fd Mon Sep 17 00:00:00 2001 From: lanyeeee <1210347077@qq.com> Date: Sun, 31 Dec 2023 01:09:13 +0800 Subject: [PATCH 315/342] remove locks, move init code to __init__ --- modules/api/api.py | 37 +++++++++++++++++++------------------ 1 file changed, 19 insertions(+), 18 deletions(-) diff --git a/modules/api/api.py b/modules/api/api.py index d202cb8d2..fc3921c23 100644 --- a/modules/api/api.py +++ b/modules/api/api.py @@ -251,8 +251,21 @@ class Api: self.default_script_arg_txt2img = [] self.default_script_arg_img2img = [] - self.txt2img_script_arg_init_lock = Lock() - self.img2img_script_arg_init_lock = Lock() + txt2img_script_runner = scripts.scripts_txt2img + img2img_script_runner = scripts.scripts_img2img + + if not txt2img_script_runner.scripts or not img2img_script_runner.scripts: + ui.create_ui() + + if not txt2img_script_runner.scripts: + txt2img_script_runner.initialize_scripts(False) + if not self.default_script_arg_txt2img: + self.default_script_arg_txt2img = self.init_default_script_args(txt2img_script_runner) + + if not img2img_script_runner.scripts: + img2img_script_runner.initialize_scripts(True) + if not self.default_script_arg_img2img: + self.default_script_arg_img2img = self.init_default_script_args(img2img_script_runner) @@ -418,16 +431,10 @@ class Api: task_id = txt2imgreq.force_task_id or create_task_id("txt2img") script_runner = scripts.scripts_txt2img - with self.txt2img_script_arg_init_lock: - if not script_runner.scripts: - script_runner.initialize_scripts(False) - ui.create_ui() - infotext_script_args = {} - self.apply_infotext(txt2imgreq, "txt2img", script_runner=script_runner, mentioned_script_args=infotext_script_args) + infotext_script_args = {} + self.apply_infotext(txt2imgreq, "txt2img", script_runner=script_runner, mentioned_script_args=infotext_script_args) - if not self.default_script_arg_txt2img: - self.default_script_arg_txt2img = self.init_default_script_args(script_runner) selectable_scripts, selectable_script_idx = self.get_selectable_script(txt2imgreq.script_name, script_runner) populate = txt2imgreq.copy(update={ # Override __init__ params @@ -488,16 +495,10 @@ class Api: mask = decode_base64_to_image(mask) script_runner = scripts.scripts_img2img - with self.img2img_script_arg_init_lock: - if not script_runner.scripts: - script_runner.initialize_scripts(True) - ui.create_ui() - infotext_script_args = {} - self.apply_infotext(img2imgreq, "img2img", script_runner=script_runner, mentioned_script_args=infotext_script_args) + infotext_script_args = {} + self.apply_infotext(img2imgreq, "img2img", script_runner=script_runner, mentioned_script_args=infotext_script_args) - if not self.default_script_arg_img2img: - self.default_script_arg_img2img = self.init_default_script_args(script_runner) selectable_scripts, selectable_script_idx = self.get_selectable_script(img2imgreq.script_name, script_runner) populate = img2imgreq.copy(update={ # Override __init__ params From 1465dab71564bb30091479ceabae6c69e3426bc6 Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Sat, 30 Dec 2023 19:44:05 +0200 Subject: [PATCH 316/342] Make Tensorboard a late import (it was implicitly installed by basicsr) --- modules/textual_inversion/textual_inversion.py | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) diff --git a/modules/textual_inversion/textual_inversion.py b/modules/textual_inversion/textual_inversion.py index 04dda585c..c6bcab153 100644 --- a/modules/textual_inversion/textual_inversion.py +++ b/modules/textual_inversion/textual_inversion.py @@ -11,7 +11,6 @@ import safetensors.torch import numpy as np from PIL import Image, PngImagePlugin -from torch.utils.tensorboard import SummaryWriter from modules import shared, devices, sd_hijack, sd_models, images, sd_samplers, sd_hijack_checkpoint, errors, hashes import modules.textual_inversion.dataset @@ -344,6 +343,7 @@ def write_loss(log_directory, filename, step, epoch_len, values): }) def tensorboard_setup(log_directory): + from torch.utils.tensorboard import SummaryWriter os.makedirs(os.path.join(log_directory, "tensorboard"), exist_ok=True) return SummaryWriter( log_dir=os.path.join(log_directory, "tensorboard"), @@ -448,8 +448,12 @@ def train_embedding(id_task, embedding_name, learn_rate, batch_size, gradient_st shared.state.textinfo = f"Preparing dataset from {html.escape(data_root)}..." old_parallel_processing_allowed = shared.parallel_processing_allowed + tensorboard_writer = None if shared.opts.training_enable_tensorboard: - tensorboard_writer = tensorboard_setup(log_directory) + try: + tensorboard_writer = tensorboard_setup(log_directory) + except ImportError: + errors.report("Error initializing tensorboard", exc_info=True) pin_memory = shared.opts.pin_memory @@ -622,7 +626,7 @@ def train_embedding(id_task, embedding_name, learn_rate, batch_size, gradient_st last_saved_image, last_text_info = images.save_image(image, images_dir, "", p.seed, p.prompt, shared.opts.samples_format, processed.infotexts[0], p=p, forced_filename=forced_filename, save_to_dirs=False) last_saved_image += f", prompt: {preview_text}" - if shared.opts.training_enable_tensorboard and shared.opts.training_tensorboard_save_images: + if tensorboard_writer and shared.opts.training_tensorboard_save_images: tensorboard_add_image(tensorboard_writer, f"Validation at epoch {epoch_num}", image, embedding.step) if save_image_with_stored_embedding and os.path.exists(last_saved_file) and embedding_yet_to_be_embedded: From 48a2a1a437a48cc232725cc813242f98483b7697 Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Sat, 30 Dec 2023 19:44:38 +0200 Subject: [PATCH 317/342] Don't wait for 10 minutes for test server to come up --- .github/workflows/run_tests.yaml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/.github/workflows/run_tests.yaml b/.github/workflows/run_tests.yaml index cd5c3f868..f42e4758e 100644 --- a/.github/workflows/run_tests.yaml +++ b/.github/workflows/run_tests.yaml @@ -57,7 +57,7 @@ jobs: 2>&1 | tee output.txt & - name: Run tests run: | - wait-for-it --service 127.0.0.1:7860 -t 600 + wait-for-it --service 127.0.0.1:7860 -t 20 python -m pytest -vv --junitxml=test/results.xml --cov . --cov-report=xml --verify-base-url test - name: Kill test server if: always() From 5fbb13e0da8eb2e26bd2c45ec8ffbb2de669ef47 Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Sat, 30 Dec 2023 20:46:44 +0200 Subject: [PATCH 318/342] Remove `cleanup_models` code --- modules/initialize.py | 3 --- modules/modelloader.py | 50 ------------------------------------------ 2 files changed, 53 deletions(-) diff --git a/modules/initialize.py b/modules/initialize.py index ac95fc6f0..4a3cd98cf 100644 --- a/modules/initialize.py +++ b/modules/initialize.py @@ -54,9 +54,6 @@ def initialize(): initialize_util.configure_sigint_handler() initialize_util.configure_opts_onchange() - from modules import modelloader - modelloader.cleanup_models() - from modules import sd_models sd_models.setup_model() startup_timer.record("setup SD model") diff --git a/modules/modelloader.py b/modules/modelloader.py index f4182559e..5f7aec3e4 100644 --- a/modules/modelloader.py +++ b/modules/modelloader.py @@ -2,7 +2,6 @@ from __future__ import annotations import logging import os -import shutil import importlib from urllib.parse import urlparse @@ -10,7 +9,6 @@ import torch from modules import shared from modules.upscaler import Upscaler, UpscalerLanczos, UpscalerNearest, UpscalerNone -from modules.paths import script_path, models_path logger = logging.getLogger(__name__) @@ -96,54 +94,6 @@ def friendly_name(file: str): return model_name -def cleanup_models(): - # This code could probably be more efficient if we used a tuple list or something to store the src/destinations - # and then enumerate that, but this works for now. In the future, it'd be nice to just have every "model" scaler - # somehow auto-register and just do these things... - root_path = script_path - src_path = models_path - dest_path = os.path.join(models_path, "Stable-diffusion") - move_files(src_path, dest_path, ".ckpt") - move_files(src_path, dest_path, ".safetensors") - src_path = os.path.join(root_path, "ESRGAN") - dest_path = os.path.join(models_path, "ESRGAN") - move_files(src_path, dest_path) - src_path = os.path.join(models_path, "BSRGAN") - dest_path = os.path.join(models_path, "ESRGAN") - move_files(src_path, dest_path, ".pth") - src_path = os.path.join(root_path, "gfpgan") - dest_path = os.path.join(models_path, "GFPGAN") - move_files(src_path, dest_path) - src_path = os.path.join(root_path, "SwinIR") - dest_path = os.path.join(models_path, "SwinIR") - move_files(src_path, dest_path) - src_path = os.path.join(root_path, "repositories/latent-diffusion/experiments/pretrained_models/") - dest_path = os.path.join(models_path, "LDSR") - move_files(src_path, dest_path) - - -def move_files(src_path: str, dest_path: str, ext_filter: str = None): - try: - os.makedirs(dest_path, exist_ok=True) - if os.path.exists(src_path): - for file in os.listdir(src_path): - fullpath = os.path.join(src_path, file) - if os.path.isfile(fullpath): - if ext_filter is not None: - if ext_filter not in file: - continue - print(f"Moving {file} from {src_path} to {dest_path}.") - try: - shutil.move(fullpath, dest_path) - except Exception: - pass - if len(os.listdir(src_path)) == 0: - print(f"Removing empty folder: {src_path}") - shutil.rmtree(src_path, True) - except Exception: - pass - - def load_upscalers(): # We can only do this 'magic' method to dynamically load upscalers if they are referenced, # so we'll try to import any _model.py files before looking in __subclasses__ From af050dcaa75ef40b6b1c3da3361f32fe52786aeb Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Sat, 30 Dec 2023 21:05:59 +0200 Subject: [PATCH 319/342] Soften Spandrel model-architecture check to just a warning --- modules/modelloader.py | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/modules/modelloader.py b/modules/modelloader.py index f4182559e..6b7d697f1 100644 --- a/modules/modelloader.py +++ b/modules/modelloader.py @@ -196,7 +196,9 @@ def load_spandrel_model( import spandrel model = spandrel.ModelLoader(device=device).load_from_file(path) if expected_architecture and model.architecture != expected_architecture: - raise TypeError(f"Model {path} is not a {expected_architecture} model") + logger.warning( + f"Model {path!r} is not a {expected_architecture!r} model (got {model.architecture!r})", + ) if half: model = model.model.half() if dtype: From 393a5b82ba6df06d85f2bf7bbbe0456d3d06115f Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Sat, 30 Dec 2023 21:12:32 +0200 Subject: [PATCH 320/342] Correct RealESRGAN expected architecture type to ESRGAN --- modules/realesrgan_model.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/realesrgan_model.py b/modules/realesrgan_model.py index 2a2be5ad7..65f2e8806 100644 --- a/modules/realesrgan_model.py +++ b/modules/realesrgan_model.py @@ -40,7 +40,7 @@ class UpscalerRealESRGAN(Upscaler): info.local_data_path, device=self.device, half=(not cmd_opts.no_half and not cmd_opts.upcast_sampling), - expected_architecture="RealESRGAN", + expected_architecture="ESRGAN", # "RealESRGAN" isn't a specific thing for Spandrel ) return upscale_with_model( mod, From 8100e901ab0c5b04d289eebb722c8a653b8beef1 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sat, 30 Dec 2023 22:41:53 +0300 Subject: [PATCH 321/342] fix error with RealESRGAN model failing to upscale fp32 image --- modules/upscaler_utils.py | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/modules/upscaler_utils.py b/modules/upscaler_utils.py index 8bdda51c4..39f78a0b7 100644 --- a/modules/upscaler_utils.py +++ b/modules/upscaler_utils.py @@ -16,9 +16,13 @@ def upscale_without_tiling(model, img: Image.Image): img = img[:, :, ::-1] img = np.ascontiguousarray(np.transpose(img, (2, 0, 1))) / 255 img = torch.from_numpy(img).float() - img = img.unsqueeze(0).to(devices.device_esrgan) + + model_weight = next(iter(model.parameters())) + img = img.unsqueeze(0).to(device=model_weight.device, dtype=model_weight.dtype) + with torch.no_grad(): output = model(img) + output = output.squeeze().float().cpu().clamp_(0, 1).numpy() output = 255. * np.moveaxis(output, 0, 2) output = output.astype(np.uint8) From bc5ae74c7d8949bab37e260b16e76889b9968099 Mon Sep 17 00:00:00 2001 From: Learwin <6223515+Learwin@users.noreply.github.com> Date: Sat, 30 Dec 2023 21:52:27 +0100 Subject: [PATCH 322/342] Added negative prompts to extra networks lora --- .../Lora/ui_edit_user_metadata.py | 14 +++++++-- .../Lora/ui_extra_networks_lora.py | 9 ++++++ javascript/extraNetworks.js | 31 +++++++++++++------ modules/ui_extra_networks.py | 5 ++- 4 files changed, 46 insertions(+), 13 deletions(-) diff --git a/extensions-builtin/Lora/ui_edit_user_metadata.py b/extensions-builtin/Lora/ui_edit_user_metadata.py index c70119090..f7859b21f 100644 --- a/extensions-builtin/Lora/ui_edit_user_metadata.py +++ b/extensions-builtin/Lora/ui_edit_user_metadata.py @@ -54,12 +54,14 @@ class LoraUserMetadataEditor(ui_extra_networks_user_metadata.UserMetadataEditor) self.slider_preferred_weight = None self.edit_notes = None - def save_lora_user_metadata(self, name, desc, sd_version, activation_text, preferred_weight, notes): + def save_lora_user_metadata(self, name, desc, sd_version, activation_text, preferred_weight, negative_text, negative_weight, notes): user_metadata = self.get_user_metadata(name) user_metadata["description"] = desc user_metadata["sd version"] = sd_version user_metadata["activation text"] = activation_text user_metadata["preferred weight"] = preferred_weight + user_metadata["negative text"] = negative_text + user_metadata["negative weight"] = negative_weight user_metadata["notes"] = notes self.write_user_metadata(name, user_metadata) @@ -127,6 +129,8 @@ class LoraUserMetadataEditor(ui_extra_networks_user_metadata.UserMetadataEditor) gr.HighlightedText.update(value=gradio_tags, visible=True if tags else False), user_metadata.get('activation text', ''), float(user_metadata.get('preferred weight', 0.0)), + user_metadata.get('negative text', ''), + float(user_metadata.get('negative weight', 0.0)), gr.update(visible=True if tags else False), gr.update(value=self.generate_random_prompt_from_tags(tags), visible=True if tags else False), ] @@ -162,7 +166,8 @@ class LoraUserMetadataEditor(ui_extra_networks_user_metadata.UserMetadataEditor) self.taginfo = gr.HighlightedText(label="Training dataset tags") self.edit_activation_text = gr.Text(label='Activation text', info="Will be added to prompt along with Lora") self.slider_preferred_weight = gr.Slider(label='Preferred weight', info="Set to 0 to disable", minimum=0.0, maximum=2.0, step=0.01) - + self.edit_negative_text = gr.Text(label='Negative prompt', info="Will be added to negative prompts") + self.slider_negative_weight = gr.Slider(label='Preferred negative weight', info="Set to 0 to disable", minimum=0.0, maximum=2.0, step=0.01) with gr.Row() as row_random_prompt: with gr.Column(scale=8): random_prompt = gr.Textbox(label='Random prompt', lines=4, max_lines=4, interactive=False) @@ -198,6 +203,8 @@ class LoraUserMetadataEditor(ui_extra_networks_user_metadata.UserMetadataEditor) self.taginfo, self.edit_activation_text, self.slider_preferred_weight, + self.edit_negative_text, + self.slider_negative_weight, row_random_prompt, random_prompt, ] @@ -211,7 +218,10 @@ class LoraUserMetadataEditor(ui_extra_networks_user_metadata.UserMetadataEditor) self.select_sd_version, self.edit_activation_text, self.slider_preferred_weight, + self.edit_negative_text, + self.slider_negative_weight, self.edit_notes, ] + self.setup_save_handler(self.button_save, self.save_lora_user_metadata, edited_components) diff --git a/extensions-builtin/Lora/ui_extra_networks_lora.py b/extensions-builtin/Lora/ui_extra_networks_lora.py index df02c663b..09ce2a057 100644 --- a/extensions-builtin/Lora/ui_extra_networks_lora.py +++ b/extensions-builtin/Lora/ui_extra_networks_lora.py @@ -45,6 +45,15 @@ class ExtraNetworksPageLora(ui_extra_networks.ExtraNetworksPage): if activation_text: item["prompt"] += " + " + quote_js(" " + activation_text) + negative_prompt = item["user_metadata"].get("negative text") + preferred_negative_weight = item["user_metadata"].get("negative weight") + item["negative_prompt"] = quote_js("") + if negative_prompt: + neg_prompt = negative_prompt + if (preferred_negative_weight > 0): + neg_prompt = '(' + negative_prompt + ':' + str(preferred_negative_weight) + ')' + item["negative_prompt"] = quote_js(neg_prompt) + sd_version = item["user_metadata"].get("sd version") if sd_version in network.SdVersion.__members__: item["sd_version"] = sd_version diff --git a/javascript/extraNetworks.js b/javascript/extraNetworks.js index 98a7abb74..2bb9795dc 100644 --- a/javascript/extraNetworks.js +++ b/javascript/extraNetworks.js @@ -185,8 +185,10 @@ onUiLoaded(setupExtraNetworks); var re_extranet = /<([^:^>]+:[^:]+):[\d.]+>(.*)/; var re_extranet_g = /<([^:^>]+:[^:]+):[\d.]+>/g; -function tryToRemoveExtraNetworkFromPrompt(textarea, text) { - var m = text.match(re_extranet); +var re_extranet_neg = /\(([^:^>]+:[\d.]+)\)/; +var re_extranet_g_neg = /\(([^:^>]+:[\d.]+)\)/g; +function tryToRemoveExtraNetworkFromPrompt(textarea, text, isNeg) { + var m = text.match(isNeg ? re_extranet_neg : re_extranet); var replaced = false; var newTextareaText; if (m) { @@ -194,8 +196,8 @@ function tryToRemoveExtraNetworkFromPrompt(textarea, text) { var extraTextAfterNet = m[2]; var partToSearch = m[1]; var foundAtPosition = -1; - newTextareaText = textarea.value.replaceAll(re_extranet_g, function(found, net, pos) { - m = found.match(re_extranet); + newTextareaText = textarea.value.replaceAll(isNeg ? re_extranet_g_neg : re_extranet_g, function(found, net, pos) { + m = found.match(isNeg ? re_extranet_neg : re_extranet); if (m[1] == partToSearch) { replaced = true; foundAtPosition = pos; @@ -205,7 +207,7 @@ function tryToRemoveExtraNetworkFromPrompt(textarea, text) { }); if (foundAtPosition >= 0) { - if (newTextareaText.substr(foundAtPosition, extraTextAfterNet.length) == extraTextAfterNet) { + if (extraTextAfterNet && newTextareaText.substr(foundAtPosition, extraTextAfterNet.length) == extraTextAfterNet) { newTextareaText = newTextareaText.substr(0, foundAtPosition) + newTextareaText.substr(foundAtPosition + extraTextAfterNet.length); } if (newTextareaText.substr(foundAtPosition - extraTextBeforeNet.length, extraTextBeforeNet.length) == extraTextBeforeNet) { @@ -230,14 +232,23 @@ function tryToRemoveExtraNetworkFromPrompt(textarea, text) { return false; } -function cardClicked(tabname, textToAdd, allowNegativePrompt) { - var textarea = allowNegativePrompt ? activePromptTextarea[tabname] : gradioApp().querySelector("#" + tabname + "_prompt > label > textarea"); +function updatePromptArea(text, textArea, isNeg) { - if (!tryToRemoveExtraNetworkFromPrompt(textarea, textToAdd)) { - textarea.value = textarea.value + opts.extra_networks_add_text_separator + textToAdd; + if (!tryToRemoveExtraNetworkFromPrompt(textArea, text, isNeg)) { + textArea.value = textArea.value + opts.extra_networks_add_text_separator + text; } - updateInput(textarea); + updateInput(textArea); +} + +function cardClicked(tabname, textToAdd, textToAddNegative, allowNegativePrompt) { + if (textToAddNegative.length > 0) { + updatePromptArea(textToAdd, gradioApp().querySelector("#" + tabname + "_prompt > label > textarea")) + updatePromptArea(textToAddNegative, gradioApp().querySelector("#" + tabname + "_neg_prompt > label > textarea"), true) + } else { + var textarea = allowNegativePrompt ? activePromptTextarea[tabname] : gradioApp().querySelector("#" + tabname + "_prompt > label > textarea"); + updatePromptArea(textToAdd, textarea) + } } function saveCardPreview(event, tabname, filename) { diff --git a/modules/ui_extra_networks.py b/modules/ui_extra_networks.py index fe5d3ba33..b8c022413 100644 --- a/modules/ui_extra_networks.py +++ b/modules/ui_extra_networks.py @@ -223,7 +223,10 @@ class ExtraNetworksPage: onclick = item.get("onclick", None) if onclick is None: - onclick = '"' + html.escape(f"""return cardClicked({quote_js(tabname)}, {item["prompt"]}, {"true" if self.allow_negative_prompt else "false"})""") + '"' + if "negative_prompt" in item: + onclick = '"' + html.escape(f"""return cardClicked({quote_js(tabname)}, {item["prompt"]}, {item["negative_prompt"]}, {"true" if self.allow_negative_prompt else "false"})""") + '"' + else: + onclick = '"' + html.escape(f"""return cardClicked({quote_js(tabname)}, {item["prompt"]}, {'""'}, {"true" if self.allow_negative_prompt else "false"})""") + '"' height = f"height: {shared.opts.extra_networks_card_height}px;" if shared.opts.extra_networks_card_height else '' width = f"width: {shared.opts.extra_networks_card_width}px;" if shared.opts.extra_networks_card_width else '' From a2f23f9d22dde87bf2529dcb2854a6a5d3d44278 Mon Sep 17 00:00:00 2001 From: Learwin <6223515+Learwin@users.noreply.github.com> Date: Sat, 30 Dec 2023 22:16:51 +0100 Subject: [PATCH 323/342] Code Style fixes --- extensions-builtin/Lora/ui_extra_networks_lora.py | 4 ++-- javascript/extraNetworks.js | 6 +++--- modules/upscaler_utils.py | 2 +- 3 files changed, 6 insertions(+), 6 deletions(-) diff --git a/extensions-builtin/Lora/ui_extra_networks_lora.py b/extensions-builtin/Lora/ui_extra_networks_lora.py index 09ce2a057..9a6624e3f 100644 --- a/extensions-builtin/Lora/ui_extra_networks_lora.py +++ b/extensions-builtin/Lora/ui_extra_networks_lora.py @@ -52,8 +52,8 @@ class ExtraNetworksPageLora(ui_extra_networks.ExtraNetworksPage): neg_prompt = negative_prompt if (preferred_negative_weight > 0): neg_prompt = '(' + negative_prompt + ':' + str(preferred_negative_weight) + ')' - item["negative_prompt"] = quote_js(neg_prompt) - + item["negative_prompt"] = quote_js(neg_prompt) + sd_version = item["user_metadata"].get("sd version") if sd_version in network.SdVersion.__members__: item["sd_version"] = sd_version diff --git a/javascript/extraNetworks.js b/javascript/extraNetworks.js index 2bb9795dc..f1ad19a66 100644 --- a/javascript/extraNetworks.js +++ b/javascript/extraNetworks.js @@ -243,11 +243,11 @@ function updatePromptArea(text, textArea, isNeg) { function cardClicked(tabname, textToAdd, textToAddNegative, allowNegativePrompt) { if (textToAddNegative.length > 0) { - updatePromptArea(textToAdd, gradioApp().querySelector("#" + tabname + "_prompt > label > textarea")) - updatePromptArea(textToAddNegative, gradioApp().querySelector("#" + tabname + "_neg_prompt > label > textarea"), true) + updatePromptArea(textToAdd, gradioApp().querySelector("#" + tabname + "_prompt > label > textarea")); + updatePromptArea(textToAddNegative, gradioApp().querySelector("#" + tabname + "_neg_prompt > label > textarea"), true); } else { var textarea = allowNegativePrompt ? activePromptTextarea[tabname] : gradioApp().querySelector("#" + tabname + "_prompt > label > textarea"); - updatePromptArea(textToAdd, textarea) + updatePromptArea(textToAdd, textarea); } } diff --git a/modules/upscaler_utils.py b/modules/upscaler_utils.py index 39f78a0b7..1d610dbff 100644 --- a/modules/upscaler_utils.py +++ b/modules/upscaler_utils.py @@ -6,7 +6,7 @@ import torch import tqdm from PIL import Image -from modules import devices, images +from modules import images logger = logging.getLogger(__name__) From 3be90740316f8fbb950b31d440458a5e8ed4beb3 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sun, 31 Dec 2023 00:43:41 +0300 Subject: [PATCH 324/342] fix for the previous fix. --- modules/upscaler_utils.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/upscaler_utils.py b/modules/upscaler_utils.py index 39f78a0b7..dde5d7ad4 100644 --- a/modules/upscaler_utils.py +++ b/modules/upscaler_utils.py @@ -17,7 +17,7 @@ def upscale_without_tiling(model, img: Image.Image): img = np.ascontiguousarray(np.transpose(img, (2, 0, 1))) / 255 img = torch.from_numpy(img).float() - model_weight = next(iter(model.parameters())) + model_weight = next(iter(model.model.parameters())) img = img.unsqueeze(0).to(device=model_weight.device, dtype=model_weight.dtype) with torch.no_grad(): From c0ca6348e8489651df861a101142805c213c66a0 Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Sun, 31 Dec 2023 00:04:47 +0200 Subject: [PATCH 325/342] load_spandrel_model: always return a model descriptor --- modules/modelloader.py | 23 +++++++++++++---------- 1 file changed, 13 insertions(+), 10 deletions(-) diff --git a/modules/modelloader.py b/modules/modelloader.py index 0b89d682c..8bcee08c1 100644 --- a/modules/modelloader.py +++ b/modules/modelloader.py @@ -1,8 +1,9 @@ from __future__ import annotations +import importlib import logging import os -import importlib +from typing import TYPE_CHECKING from urllib.parse import urlparse import torch @@ -10,6 +11,8 @@ import torch from modules import shared from modules.upscaler import Upscaler, UpscalerLanczos, UpscalerNearest, UpscalerNone +if TYPE_CHECKING: + import spandrel logger = logging.getLogger(__name__) @@ -142,17 +145,17 @@ def load_spandrel_model( half: bool = False, dtype: str | None = None, expected_architecture: str | None = None, -): +) -> spandrel.ModelDescriptor: import spandrel - model = spandrel.ModelLoader(device=device).load_from_file(path) - if expected_architecture and model.architecture != expected_architecture: + model_descriptor = spandrel.ModelLoader(device=device).load_from_file(path) + if expected_architecture and model_descriptor.architecture != expected_architecture: logger.warning( - f"Model {path!r} is not a {expected_architecture!r} model (got {model.architecture!r})", + f"Model {path!r} is not a {expected_architecture!r} model (got {model_descriptor.architecture!r})", ) if half: - model = model.model.half() + model_descriptor.model.half() if dtype: - model = model.model.to(dtype=dtype) - model.eval() - logger.debug("Loaded %s from %s (device=%s, half=%s, dtype=%s)", model, path, device, half, dtype) - return model + model_descriptor.model.to(dtype=dtype) + model_descriptor.model.eval() + logger.debug("Loaded %s from %s (device=%s, half=%s, dtype=%s)", model_descriptor, path, device, half, dtype) + return model_descriptor From 777af661a21821994993df3ef566b01df2bb61a0 Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Sun, 31 Dec 2023 00:09:51 +0200 Subject: [PATCH 326/342] Be more clear about Spandrel model nomenclature --- extensions-builtin/SwinIR/scripts/swinir_model.py | 6 +++--- modules/gfpgan_model.py | 10 ++++++---- modules/modelloader.py | 2 +- modules/realesrgan_model.py | 4 ++-- modules/upscaler_utils.py | 2 +- 5 files changed, 13 insertions(+), 11 deletions(-) diff --git a/extensions-builtin/SwinIR/scripts/swinir_model.py b/extensions-builtin/SwinIR/scripts/swinir_model.py index aae159af5..95c7ec648 100644 --- a/extensions-builtin/SwinIR/scripts/swinir_model.py +++ b/extensions-builtin/SwinIR/scripts/swinir_model.py @@ -71,7 +71,7 @@ class UpscalerSwinIR(Upscaler): else: filename = path - model = modelloader.load_spandrel_model( + model_descriptor = modelloader.load_spandrel_model( filename, device=self._get_device(), dtype=devices.dtype, @@ -79,10 +79,10 @@ class UpscalerSwinIR(Upscaler): ) if getattr(opts, 'SWIN_torch_compile', False): try: - model = torch.compile(model) + model_descriptor.model.compile() except Exception: logger.warning("Failed to compile SwinIR model, fallback to JIT", exc_info=True) - return model + return model_descriptor def _get_device(self): return devices.get_device_for('swinir') diff --git a/modules/gfpgan_model.py b/modules/gfpgan_model.py index 48f8ad5e2..445b04092 100644 --- a/modules/gfpgan_model.py +++ b/modules/gfpgan_model.py @@ -3,6 +3,8 @@ from __future__ import annotations import logging import os +import torch + from modules import ( devices, errors, @@ -25,7 +27,7 @@ class FaceRestorerGFPGAN(face_restoration_utils.CommonFaceRestoration): def get_device(self): return devices.device_gfpgan - def load_net(self) -> None: + def load_net(self) -> torch.Module: for model_path in modelloader.load_models( model_path=self.model_path, model_url=model_url, @@ -34,13 +36,13 @@ class FaceRestorerGFPGAN(face_restoration_utils.CommonFaceRestoration): ext_filter=['.pth'], ): if 'GFPGAN' in os.path.basename(model_path): - net = modelloader.load_spandrel_model( + model = modelloader.load_spandrel_model( model_path, device=self.get_device(), expected_architecture='GFPGAN', ).model - net.different_w = True # see https://github.com/chaiNNer-org/spandrel/pull/81 - return net + model.different_w = True # see https://github.com/chaiNNer-org/spandrel/pull/81 + return model raise ValueError("No GFPGAN model found") def restore(self, np_image): diff --git a/modules/modelloader.py b/modules/modelloader.py index 8bcee08c1..a71941375 100644 --- a/modules/modelloader.py +++ b/modules/modelloader.py @@ -143,7 +143,7 @@ def load_spandrel_model( *, device: str | torch.device | None, half: bool = False, - dtype: str | None = None, + dtype: str | torch.dtype | None = None, expected_architecture: str | None = None, ) -> spandrel.ModelDescriptor: import spandrel diff --git a/modules/realesrgan_model.py b/modules/realesrgan_model.py index 65f2e8806..4d35b695c 100644 --- a/modules/realesrgan_model.py +++ b/modules/realesrgan_model.py @@ -36,14 +36,14 @@ class UpscalerRealESRGAN(Upscaler): errors.report(f"Unable to load RealESRGAN model {path}", exc_info=True) return img - mod = modelloader.load_spandrel_model( + model_descriptor = modelloader.load_spandrel_model( info.local_data_path, device=self.device, half=(not cmd_opts.no_half and not cmd_opts.upcast_sampling), expected_architecture="ESRGAN", # "RealESRGAN" isn't a specific thing for Spandrel ) return upscale_with_model( - mod, + model_descriptor, img, tile_size=opts.ESRGAN_tile, tile_overlap=opts.ESRGAN_tile_overlap, diff --git a/modules/upscaler_utils.py b/modules/upscaler_utils.py index dde5d7ad4..174c9bc37 100644 --- a/modules/upscaler_utils.py +++ b/modules/upscaler_utils.py @@ -6,7 +6,7 @@ import torch import tqdm from PIL import Image -from modules import devices, images +from modules import images logger = logging.getLogger(__name__) From 6f86b62a1be7993073ba3a789d522e0b8870605a Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Sat, 30 Dec 2023 22:53:49 +0200 Subject: [PATCH 327/342] Deduplicate tiled inference code from SwinIR/ScuNET --- .../ScuNET/scripts/scunet_model.py | 55 +++----------- .../SwinIR/scripts/swinir_model.py | 57 ++------------- modules/upscaler_utils.py | 72 ++++++++++++++++++- 3 files changed, 87 insertions(+), 97 deletions(-) diff --git a/extensions-builtin/ScuNET/scripts/scunet_model.py b/extensions-builtin/ScuNET/scripts/scunet_model.py index 5f3dd08b3..f799cb76d 100644 --- a/extensions-builtin/ScuNET/scripts/scunet_model.py +++ b/extensions-builtin/ScuNET/scripts/scunet_model.py @@ -3,12 +3,11 @@ import sys import PIL.Image import numpy as np import torch -from tqdm import tqdm import modules.upscaler from modules import devices, modelloader, script_callbacks, errors - from modules.shared import opts +from modules.upscaler_utils import tiled_upscale_2 class UpscalerScuNET(modules.upscaler.Upscaler): @@ -40,47 +39,6 @@ class UpscalerScuNET(modules.upscaler.Upscaler): scalers.append(scaler_data2) self.scalers = scalers - @staticmethod - @torch.no_grad() - def tiled_inference(img, model): - # test the image tile by tile - h, w = img.shape[2:] - tile = opts.SCUNET_tile - tile_overlap = opts.SCUNET_tile_overlap - if tile == 0: - return model(img) - - device = devices.get_device_for('scunet') - assert tile % 8 == 0, "tile size should be a multiple of window_size" - sf = 1 - - stride = tile - tile_overlap - h_idx_list = list(range(0, h - tile, stride)) + [h - tile] - w_idx_list = list(range(0, w - tile, stride)) + [w - tile] - E = torch.zeros(1, 3, h * sf, w * sf, dtype=img.dtype, device=device) - W = torch.zeros_like(E, dtype=devices.dtype, device=device) - - with tqdm(total=len(h_idx_list) * len(w_idx_list), desc="ScuNET tiles") as pbar: - for h_idx in h_idx_list: - - for w_idx in w_idx_list: - - in_patch = img[..., h_idx: h_idx + tile, w_idx: w_idx + tile] - - out_patch = model(in_patch) - out_patch_mask = torch.ones_like(out_patch) - - E[ - ..., h_idx * sf: (h_idx + tile) * sf, w_idx * sf: (w_idx + tile) * sf - ].add_(out_patch) - W[ - ..., h_idx * sf: (h_idx + tile) * sf, w_idx * sf: (w_idx + tile) * sf - ].add_(out_patch_mask) - pbar.update(1) - output = E.div_(W) - - return output - def do_upscale(self, img: PIL.Image.Image, selected_file): devices.torch_gc() @@ -104,7 +62,16 @@ class UpscalerScuNET(modules.upscaler.Upscaler): _img[:, :, :h, :w] = torch_img # pad image torch_img = _img - torch_output = self.tiled_inference(torch_img, model).squeeze(0) + with torch.no_grad(): + torch_output = tiled_upscale_2( + torch_img, + model, + tile_size=opts.SCUNET_tile, + tile_overlap=opts.SCUNET_tile_overlap, + scale=1, + device=devices.get_device_for('scunet'), + desc="ScuNET tiles", + ).squeeze(0) torch_output = torch_output[:, :h * 1, :w * 1] # remove padding, if any np_output: np.ndarray = torch_output.float().cpu().clamp_(0, 1).numpy() del torch_img, torch_output diff --git a/extensions-builtin/SwinIR/scripts/swinir_model.py b/extensions-builtin/SwinIR/scripts/swinir_model.py index 95c7ec648..8a555c794 100644 --- a/extensions-builtin/SwinIR/scripts/swinir_model.py +++ b/extensions-builtin/SwinIR/scripts/swinir_model.py @@ -4,11 +4,11 @@ import sys import numpy as np import torch from PIL import Image -from tqdm import tqdm from modules import modelloader, devices, script_callbacks, shared -from modules.shared import opts, state +from modules.shared import opts from modules.upscaler import Upscaler, UpscalerData +from modules.upscaler_utils import tiled_upscale_2 SWINIR_MODEL_URL = "https://github.com/JingyunLiang/SwinIR/releases/download/v0.0/003_realSR_BSRGAN_DFOWMFC_s64w8_SwinIR-L_x4_GAN.pth" @@ -110,14 +110,14 @@ def upscale( w_pad = (w_old // window_size + 1) * window_size - w_old img = torch.cat([img, torch.flip(img, [2])], 2)[:, :, : h_old + h_pad, :] img = torch.cat([img, torch.flip(img, [3])], 3)[:, :, :, : w_old + w_pad] - output = inference( + output = tiled_upscale_2( img, model, - tile=tile, + tile_size=tile, tile_overlap=tile_overlap, - window_size=window_size, scale=scale, device=device, + desc="SwinIR tiles", ) output = output[..., : h_old * scale, : w_old * scale] output = output.data.squeeze().float().cpu().clamp_(0, 1).numpy() @@ -129,53 +129,6 @@ def upscale( return Image.fromarray(output, "RGB") -def inference( - img, - model, - *, - tile: int, - tile_overlap: int, - window_size: int, - scale: int, - device, -): - # test the image tile by tile - b, c, h, w = img.size() - tile = min(tile, h, w) - assert tile % window_size == 0, "tile size should be a multiple of window_size" - sf = scale - - stride = tile - tile_overlap - h_idx_list = list(range(0, h - tile, stride)) + [h - tile] - w_idx_list = list(range(0, w - tile, stride)) + [w - tile] - E = torch.zeros(b, c, h * sf, w * sf, dtype=devices.dtype, device=device).type_as(img) - W = torch.zeros_like(E, dtype=devices.dtype, device=device) - - with tqdm(total=len(h_idx_list) * len(w_idx_list), desc="SwinIR tiles") as pbar: - for h_idx in h_idx_list: - if state.interrupted or state.skipped: - break - - for w_idx in w_idx_list: - if state.interrupted or state.skipped: - break - - in_patch = img[..., h_idx: h_idx + tile, w_idx: w_idx + tile] - out_patch = model(in_patch) - out_patch_mask = torch.ones_like(out_patch) - - E[ - ..., h_idx * sf: (h_idx + tile) * sf, w_idx * sf: (w_idx + tile) * sf - ].add_(out_patch) - W[ - ..., h_idx * sf: (h_idx + tile) * sf, w_idx * sf: (w_idx + tile) * sf - ].add_(out_patch_mask) - pbar.update(1) - output = E.div_(W) - - return output - - def on_ui_settings(): import gradio as gr diff --git a/modules/upscaler_utils.py b/modules/upscaler_utils.py index 174c9bc37..8e4138543 100644 --- a/modules/upscaler_utils.py +++ b/modules/upscaler_utils.py @@ -6,7 +6,7 @@ import torch import tqdm from PIL import Image -from modules import images +from modules import images, shared logger = logging.getLogger(__name__) @@ -68,3 +68,73 @@ def upscale_with_model( overlap=grid.overlap * scale_factor, ) return images.combine_grid(newgrid) + + +def tiled_upscale_2( + img, + model, + *, + tile_size: int, + tile_overlap: int, + scale: int, + device, + desc="Tiled upscale", +): + # Alternative implementation of `upscale_with_model` originally used by + # SwinIR and ScuNET. It differs from `upscale_with_model` in that tiling and + # weighting is done in PyTorch space, as opposed to `images.Grid` doing it in + # Pillow space without weighting. + b, c, h, w = img.size() + tile_size = min(tile_size, h, w) + + if tile_size <= 0: + logger.debug("Upscaling %s without tiling", img.shape) + return model(img) + + stride = tile_size - tile_overlap + h_idx_list = list(range(0, h - tile_size, stride)) + [h - tile_size] + w_idx_list = list(range(0, w - tile_size, stride)) + [w - tile_size] + result = torch.zeros( + b, + c, + h * scale, + w * scale, + device=device, + ).type_as(img) + weights = torch.zeros_like(result) + logger.debug("Upscaling %s to %s with tiles", img.shape, result.shape) + with tqdm.tqdm(total=len(h_idx_list) * len(w_idx_list), desc=desc) as pbar: + for h_idx in h_idx_list: + if shared.state.interrupted or shared.state.skipped: + break + + for w_idx in w_idx_list: + if shared.state.interrupted or shared.state.skipped: + break + + in_patch = img[ + ..., + h_idx : h_idx + tile_size, + w_idx : w_idx + tile_size, + ] + out_patch = model(in_patch) + + result[ + ..., + h_idx * scale : (h_idx + tile_size) * scale, + w_idx * scale : (w_idx + tile_size) * scale, + ].add_(out_patch) + + out_patch_mask = torch.ones_like(out_patch) + + weights[ + ..., + h_idx * scale : (h_idx + tile_size) * scale, + w_idx * scale : (w_idx + tile_size) * scale, + ].add_(out_patch_mask) + + pbar.update(1) + + output = result.div_(weights) + + return output From 5768afc776a66bb94e77a9c1daebeea58fa731d5 Mon Sep 17 00:00:00 2001 From: Aarni Koskela Date: Sun, 31 Dec 2023 00:20:30 +0200 Subject: [PATCH 328/342] Add utility to inspect a model's parameters (to get dtype/device) --- modules/devices.py | 3 ++- modules/interrogate.py | 3 ++- modules/sd_models_xl.py | 3 ++- modules/torch_utils.py | 17 +++++++++++++++++ modules/upscaler_utils.py | 5 +++-- modules/xlmr.py | 5 ++++- modules/xlmr_m18.py | 5 ++++- test/test_torch_utils.py | 19 +++++++++++++++++++ 8 files changed, 53 insertions(+), 7 deletions(-) create mode 100644 modules/torch_utils.py create mode 100644 test/test_torch_utils.py diff --git a/modules/devices.py b/modules/devices.py index c956207f3..bd6bd579b 100644 --- a/modules/devices.py +++ b/modules/devices.py @@ -4,6 +4,7 @@ from functools import lru_cache import torch from modules import errors, shared +from modules.torch_utils import get_param if sys.platform == "darwin": from modules import mac_specific @@ -131,7 +132,7 @@ patch_module_list = [ def manual_cast_forward(self, *args, **kwargs): - org_dtype = next(self.parameters()).dtype + org_dtype = get_param(self).dtype self.to(dtype) args = [arg.to(dtype) if isinstance(arg, torch.Tensor) else arg for arg in args] kwargs = {k: v.to(dtype) if isinstance(v, torch.Tensor) else v for k, v in kwargs.items()} diff --git a/modules/interrogate.py b/modules/interrogate.py index 3045560d0..5be5a10f3 100644 --- a/modules/interrogate.py +++ b/modules/interrogate.py @@ -11,6 +11,7 @@ from torchvision import transforms from torchvision.transforms.functional import InterpolationMode from modules import devices, paths, shared, lowvram, modelloader, errors +from modules.torch_utils import get_param blip_image_eval_size = 384 clip_model_name = 'ViT-L/14' @@ -131,7 +132,7 @@ class InterrogateModels: self.clip_model = self.clip_model.to(devices.device_interrogate) - self.dtype = next(self.clip_model.parameters()).dtype + self.dtype = get_param(self.clip_model).dtype def send_clip_to_ram(self): if not shared.opts.interrogate_keep_models_in_memory: diff --git a/modules/sd_models_xl.py b/modules/sd_models_xl.py index 1de31b0da..c3602a7e1 100644 --- a/modules/sd_models_xl.py +++ b/modules/sd_models_xl.py @@ -6,6 +6,7 @@ import sgm.models.diffusion import sgm.modules.diffusionmodules.denoiser_scaling import sgm.modules.diffusionmodules.discretizer from modules import devices, shared, prompt_parser +from modules.torch_utils import get_param def get_learned_conditioning(self: sgm.models.diffusion.DiffusionEngine, batch: prompt_parser.SdConditioning | list[str]): @@ -90,7 +91,7 @@ sgm.modules.GeneralConditioner.get_target_prompt_token_count = get_target_prompt def extend_sdxl(model): """this adds a bunch of parameters to make SDXL model look a bit more like SD1.5 to the rest of the codebase.""" - dtype = next(model.model.diffusion_model.parameters()).dtype + dtype = get_param(model.model.diffusion_model).dtype model.model.diffusion_model.dtype = dtype model.model.conditioning_key = 'crossattn' model.cond_stage_key = 'txt' diff --git a/modules/torch_utils.py b/modules/torch_utils.py new file mode 100644 index 000000000..e5b52393e --- /dev/null +++ b/modules/torch_utils.py @@ -0,0 +1,17 @@ +from __future__ import annotations + +import torch.nn + + +def get_param(model) -> torch.nn.Parameter: + """ + Find the first parameter in a model or module. + """ + if hasattr(model, "model") and hasattr(model.model, "parameters"): + # Unpeel a model descriptor to get at the actual Torch module. + model = model.model + + for param in model.parameters(): + return param + + raise ValueError(f"No parameters found in model {model!r}") diff --git a/modules/upscaler_utils.py b/modules/upscaler_utils.py index 8e4138543..c60e3beb8 100644 --- a/modules/upscaler_utils.py +++ b/modules/upscaler_utils.py @@ -7,6 +7,7 @@ import tqdm from PIL import Image from modules import images, shared +from modules.torch_utils import get_param logger = logging.getLogger(__name__) @@ -17,8 +18,8 @@ def upscale_without_tiling(model, img: Image.Image): img = np.ascontiguousarray(np.transpose(img, (2, 0, 1))) / 255 img = torch.from_numpy(img).float() - model_weight = next(iter(model.model.parameters())) - img = img.unsqueeze(0).to(device=model_weight.device, dtype=model_weight.dtype) + param = get_param(model) + img = img.unsqueeze(0).to(device=param.device, dtype=param.dtype) with torch.no_grad(): output = model(img) diff --git a/modules/xlmr.py b/modules/xlmr.py index a407a3cad..6e000a56e 100644 --- a/modules/xlmr.py +++ b/modules/xlmr.py @@ -5,6 +5,9 @@ from transformers.models.xlm_roberta.configuration_xlm_roberta import XLMRoberta from transformers import XLMRobertaModel,XLMRobertaTokenizer from typing import Optional +from modules.torch_utils import get_param + + class BertSeriesConfig(BertConfig): def __init__(self, vocab_size=30522, hidden_size=768, num_hidden_layers=12, num_attention_heads=12, intermediate_size=3072, hidden_act="gelu", hidden_dropout_prob=0.1, attention_probs_dropout_prob=0.1, max_position_embeddings=512, type_vocab_size=2, initializer_range=0.02, layer_norm_eps=1e-12, pad_token_id=0, position_embedding_type="absolute", use_cache=True, classifier_dropout=None,project_dim=512, pooler_fn="average",learn_encoder=False,model_type='bert',**kwargs): @@ -62,7 +65,7 @@ class BertSeriesModelWithTransformation(BertPreTrainedModel): self.post_init() def encode(self,c): - device = next(self.parameters()).device + device = get_param(self).device text = self.tokenizer(c, truncation=True, max_length=77, diff --git a/modules/xlmr_m18.py b/modules/xlmr_m18.py index a727e8655..e3e819610 100644 --- a/modules/xlmr_m18.py +++ b/modules/xlmr_m18.py @@ -5,6 +5,9 @@ from transformers.models.xlm_roberta.configuration_xlm_roberta import XLMRoberta from transformers import XLMRobertaModel,XLMRobertaTokenizer from typing import Optional +from modules.torch_utils import get_param + + class BertSeriesConfig(BertConfig): def __init__(self, vocab_size=30522, hidden_size=768, num_hidden_layers=12, num_attention_heads=12, intermediate_size=3072, hidden_act="gelu", hidden_dropout_prob=0.1, attention_probs_dropout_prob=0.1, max_position_embeddings=512, type_vocab_size=2, initializer_range=0.02, layer_norm_eps=1e-12, pad_token_id=0, position_embedding_type="absolute", use_cache=True, classifier_dropout=None,project_dim=512, pooler_fn="average",learn_encoder=False,model_type='bert',**kwargs): @@ -68,7 +71,7 @@ class BertSeriesModelWithTransformation(BertPreTrainedModel): self.post_init() def encode(self,c): - device = next(self.parameters()).device + device = get_param(self).device text = self.tokenizer(c, truncation=True, max_length=77, diff --git a/test/test_torch_utils.py b/test/test_torch_utils.py new file mode 100644 index 000000000..f1aec8329 --- /dev/null +++ b/test/test_torch_utils.py @@ -0,0 +1,19 @@ +import types + +import pytest +import torch + +from modules.torch_utils import get_param + + +@pytest.mark.parametrize("wrapped", [True, False]) +def test_get_param(wrapped): + mod = torch.nn.Linear(1, 1) + cpu = torch.device("cpu") + mod.to(dtype=torch.float16, device=cpu) + if wrapped: + # more or less how spandrel wraps a thing + mod = types.SimpleNamespace(model=mod) + p = get_param(mod) + assert p.dtype == torch.float16 + assert p.device == cpu From d4945f4422e5a0bf31a6dbe4c1aeedd78c09eacb Mon Sep 17 00:00:00 2001 From: Learwin <6223515+Learwin@users.noreply.github.com> Date: Sun, 31 Dec 2023 13:22:30 +0100 Subject: [PATCH 329/342] Removed weight slider for negative prompts --- extensions-builtin/Lora/ui_edit_user_metadata.py | 7 +------ extensions-builtin/Lora/ui_extra_networks_lora.py | 6 +----- 2 files changed, 2 insertions(+), 11 deletions(-) diff --git a/extensions-builtin/Lora/ui_edit_user_metadata.py b/extensions-builtin/Lora/ui_edit_user_metadata.py index f7859b21f..3160aecfa 100644 --- a/extensions-builtin/Lora/ui_edit_user_metadata.py +++ b/extensions-builtin/Lora/ui_edit_user_metadata.py @@ -54,14 +54,13 @@ class LoraUserMetadataEditor(ui_extra_networks_user_metadata.UserMetadataEditor) self.slider_preferred_weight = None self.edit_notes = None - def save_lora_user_metadata(self, name, desc, sd_version, activation_text, preferred_weight, negative_text, negative_weight, notes): + def save_lora_user_metadata(self, name, desc, sd_version, activation_text, preferred_weight, negative_text, notes): user_metadata = self.get_user_metadata(name) user_metadata["description"] = desc user_metadata["sd version"] = sd_version user_metadata["activation text"] = activation_text user_metadata["preferred weight"] = preferred_weight user_metadata["negative text"] = negative_text - user_metadata["negative weight"] = negative_weight user_metadata["notes"] = notes self.write_user_metadata(name, user_metadata) @@ -130,7 +129,6 @@ class LoraUserMetadataEditor(ui_extra_networks_user_metadata.UserMetadataEditor) user_metadata.get('activation text', ''), float(user_metadata.get('preferred weight', 0.0)), user_metadata.get('negative text', ''), - float(user_metadata.get('negative weight', 0.0)), gr.update(visible=True if tags else False), gr.update(value=self.generate_random_prompt_from_tags(tags), visible=True if tags else False), ] @@ -167,7 +165,6 @@ class LoraUserMetadataEditor(ui_extra_networks_user_metadata.UserMetadataEditor) self.edit_activation_text = gr.Text(label='Activation text', info="Will be added to prompt along with Lora") self.slider_preferred_weight = gr.Slider(label='Preferred weight', info="Set to 0 to disable", minimum=0.0, maximum=2.0, step=0.01) self.edit_negative_text = gr.Text(label='Negative prompt', info="Will be added to negative prompts") - self.slider_negative_weight = gr.Slider(label='Preferred negative weight', info="Set to 0 to disable", minimum=0.0, maximum=2.0, step=0.01) with gr.Row() as row_random_prompt: with gr.Column(scale=8): random_prompt = gr.Textbox(label='Random prompt', lines=4, max_lines=4, interactive=False) @@ -204,7 +201,6 @@ class LoraUserMetadataEditor(ui_extra_networks_user_metadata.UserMetadataEditor) self.edit_activation_text, self.slider_preferred_weight, self.edit_negative_text, - self.slider_negative_weight, row_random_prompt, random_prompt, ] @@ -219,7 +215,6 @@ class LoraUserMetadataEditor(ui_extra_networks_user_metadata.UserMetadataEditor) self.edit_activation_text, self.slider_preferred_weight, self.edit_negative_text, - self.slider_negative_weight, self.edit_notes, ] diff --git a/extensions-builtin/Lora/ui_extra_networks_lora.py b/extensions-builtin/Lora/ui_extra_networks_lora.py index 9a6624e3f..e714fac46 100644 --- a/extensions-builtin/Lora/ui_extra_networks_lora.py +++ b/extensions-builtin/Lora/ui_extra_networks_lora.py @@ -46,13 +46,9 @@ class ExtraNetworksPageLora(ui_extra_networks.ExtraNetworksPage): item["prompt"] += " + " + quote_js(" " + activation_text) negative_prompt = item["user_metadata"].get("negative text") - preferred_negative_weight = item["user_metadata"].get("negative weight") item["negative_prompt"] = quote_js("") if negative_prompt: - neg_prompt = negative_prompt - if (preferred_negative_weight > 0): - neg_prompt = '(' + negative_prompt + ':' + str(preferred_negative_weight) + ')' - item["negative_prompt"] = quote_js(neg_prompt) + item["negative_prompt"] = quote_js('(' + negative_prompt + ':1)') sd_version = item["user_metadata"].get("sd version") if sd_version in network.SdVersion.__members__: From b6f74e936e4de3b8d190bffaf3bed67d6d4bd211 Mon Sep 17 00:00:00 2001 From: Learwin <6223515+Learwin@users.noreply.github.com> Date: Sun, 31 Dec 2023 13:36:36 +0100 Subject: [PATCH 330/342] Revert change from linting for unrelated file --- modules/upscaler_utils.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/upscaler_utils.py b/modules/upscaler_utils.py index 1d610dbff..39f78a0b7 100644 --- a/modules/upscaler_utils.py +++ b/modules/upscaler_utils.py @@ -6,7 +6,7 @@ import torch import tqdm from PIL import Image -from modules import images +from modules import devices, images logger = logging.getLogger(__name__) From a70dfb64a86b9b6d869deffdb0ffebe980365473 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Sun, 31 Dec 2023 22:38:30 +0300 Subject: [PATCH 331/342] change import statements for #14478 --- modules/devices.py | 4 ++-- modules/interrogate.py | 5 ++--- modules/sd_models_xl.py | 4 ++-- modules/upscaler_utils.py | 5 ++--- modules/xlmr.py | 4 ++-- modules/xlmr_m18.py | 5 ++--- test/test_torch_utils.py | 4 ++-- 7 files changed, 14 insertions(+), 17 deletions(-) diff --git a/modules/devices.py b/modules/devices.py index bd6bd579b..ff279ac50 100644 --- a/modules/devices.py +++ b/modules/devices.py @@ -4,7 +4,7 @@ from functools import lru_cache import torch from modules import errors, shared -from modules.torch_utils import get_param +from modules import torch_utils if sys.platform == "darwin": from modules import mac_specific @@ -132,7 +132,7 @@ patch_module_list = [ def manual_cast_forward(self, *args, **kwargs): - org_dtype = get_param(self).dtype + org_dtype = torch_utils.get_param(self).dtype self.to(dtype) args = [arg.to(dtype) if isinstance(arg, torch.Tensor) else arg for arg in args] kwargs = {k: v.to(dtype) if isinstance(v, torch.Tensor) else v for k, v in kwargs.items()} diff --git a/modules/interrogate.py b/modules/interrogate.py index 5be5a10f3..35a627ca1 100644 --- a/modules/interrogate.py +++ b/modules/interrogate.py @@ -10,8 +10,7 @@ import torch.hub from torchvision import transforms from torchvision.transforms.functional import InterpolationMode -from modules import devices, paths, shared, lowvram, modelloader, errors -from modules.torch_utils import get_param +from modules import devices, paths, shared, lowvram, modelloader, errors, torch_utils blip_image_eval_size = 384 clip_model_name = 'ViT-L/14' @@ -132,7 +131,7 @@ class InterrogateModels: self.clip_model = self.clip_model.to(devices.device_interrogate) - self.dtype = get_param(self.clip_model).dtype + self.dtype = torch_utils.get_param(self.clip_model).dtype def send_clip_to_ram(self): if not shared.opts.interrogate_keep_models_in_memory: diff --git a/modules/sd_models_xl.py b/modules/sd_models_xl.py index c3602a7e1..0de17af3d 100644 --- a/modules/sd_models_xl.py +++ b/modules/sd_models_xl.py @@ -6,7 +6,7 @@ import sgm.models.diffusion import sgm.modules.diffusionmodules.denoiser_scaling import sgm.modules.diffusionmodules.discretizer from modules import devices, shared, prompt_parser -from modules.torch_utils import get_param +from modules import torch_utils def get_learned_conditioning(self: sgm.models.diffusion.DiffusionEngine, batch: prompt_parser.SdConditioning | list[str]): @@ -91,7 +91,7 @@ sgm.modules.GeneralConditioner.get_target_prompt_token_count = get_target_prompt def extend_sdxl(model): """this adds a bunch of parameters to make SDXL model look a bit more like SD1.5 to the rest of the codebase.""" - dtype = get_param(model.model.diffusion_model).dtype + dtype = torch_utils.get_param(model.model.diffusion_model).dtype model.model.diffusion_model.dtype = dtype model.model.conditioning_key = 'crossattn' model.cond_stage_key = 'txt' diff --git a/modules/upscaler_utils.py b/modules/upscaler_utils.py index c60e3beb8..f5cb92d51 100644 --- a/modules/upscaler_utils.py +++ b/modules/upscaler_utils.py @@ -6,8 +6,7 @@ import torch import tqdm from PIL import Image -from modules import images, shared -from modules.torch_utils import get_param +from modules import images, shared, torch_utils logger = logging.getLogger(__name__) @@ -18,7 +17,7 @@ def upscale_without_tiling(model, img: Image.Image): img = np.ascontiguousarray(np.transpose(img, (2, 0, 1))) / 255 img = torch.from_numpy(img).float() - param = get_param(model) + param = torch_utils.get_param(model) img = img.unsqueeze(0).to(device=param.device, dtype=param.dtype) with torch.no_grad(): diff --git a/modules/xlmr.py b/modules/xlmr.py index 6e000a56e..319771b7b 100644 --- a/modules/xlmr.py +++ b/modules/xlmr.py @@ -5,7 +5,7 @@ from transformers.models.xlm_roberta.configuration_xlm_roberta import XLMRoberta from transformers import XLMRobertaModel,XLMRobertaTokenizer from typing import Optional -from modules.torch_utils import get_param +from modules import torch_utils class BertSeriesConfig(BertConfig): @@ -65,7 +65,7 @@ class BertSeriesModelWithTransformation(BertPreTrainedModel): self.post_init() def encode(self,c): - device = get_param(self).device + device = torch_utils.get_param(self).device text = self.tokenizer(c, truncation=True, max_length=77, diff --git a/modules/xlmr_m18.py b/modules/xlmr_m18.py index e3e819610..f60555049 100644 --- a/modules/xlmr_m18.py +++ b/modules/xlmr_m18.py @@ -4,8 +4,7 @@ import torch from transformers.models.xlm_roberta.configuration_xlm_roberta import XLMRobertaConfig from transformers import XLMRobertaModel,XLMRobertaTokenizer from typing import Optional - -from modules.torch_utils import get_param +from modules import torch_utils class BertSeriesConfig(BertConfig): @@ -71,7 +70,7 @@ class BertSeriesModelWithTransformation(BertPreTrainedModel): self.post_init() def encode(self,c): - device = get_param(self).device + device = torch_utils.get_param(self).device text = self.tokenizer(c, truncation=True, max_length=77, diff --git a/test/test_torch_utils.py b/test/test_torch_utils.py index f1aec8329..23ccb93a4 100644 --- a/test/test_torch_utils.py +++ b/test/test_torch_utils.py @@ -3,7 +3,7 @@ import types import pytest import torch -from modules.torch_utils import get_param +from modules import torch_utils @pytest.mark.parametrize("wrapped", [True, False]) @@ -14,6 +14,6 @@ def test_get_param(wrapped): if wrapped: # more or less how spandrel wraps a thing mod = types.SimpleNamespace(model=mod) - p = get_param(mod) + p = torch_utils.get_param(mod) assert p.dtype == torch.float16 assert p.device == cpu From 00901bfbe0095303554f4440b4c12fac262e2e89 Mon Sep 17 00:00:00 2001 From: w-e-w <40751091+w-e-w@users.noreply.github.com> Date: Mon, 1 Jan 2024 15:47:57 +0900 Subject: [PATCH 332/342] handle selectable script_index is None --- modules/scripts.py | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/modules/scripts.py b/modules/scripts.py index 3a7669118..017aed5a0 100644 --- a/modules/scripts.py +++ b/modules/scripts.py @@ -696,6 +696,8 @@ class ScriptRunner: self.setup_ui_for_section(None, self.selectable_scripts) def select_script(script_index): + if script_index is None: + script_index = 0 selected_script = self.selectable_scripts[script_index - 1] if script_index>0 else None return [gr.update(visible=selected_script == s) for s in self.selectable_scripts] @@ -739,7 +741,7 @@ class ScriptRunner: def run(self, p, *args): script_index = args[0] - if script_index == 0: + if script_index == 0 or script_index is None: return None script = self.selectable_scripts[script_index-1] From 5692bf1517c3409ad46262c56e65f256389825b1 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Mon, 1 Jan 2024 11:11:14 +0300 Subject: [PATCH 333/342] add missing field for DDIM sampler that was breaking img2img --- modules/sd_samplers_timesteps.py | 1 + 1 file changed, 1 insertion(+) diff --git a/modules/sd_samplers_timesteps.py b/modules/sd_samplers_timesteps.py index b17a8f93c..f8afa8bd7 100644 --- a/modules/sd_samplers_timesteps.py +++ b/modules/sd_samplers_timesteps.py @@ -80,6 +80,7 @@ class CompVisSampler(sd_samplers_common.Sampler): self.eta_default = 0.0 self.model_wrap_cfg = CFGDenoiserTimesteps(self) + self.model_wrap = self.model_wrap_cfg.inner_model def get_timesteps(self, p, steps): discard_next_to_last_sigma = self.config is not None and self.config.options.get('discard_next_to_last_sigma', False) From 003b91f08361c99ecdd97257624d81a2046d3823 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Mon, 1 Jan 2024 13:45:01 +0300 Subject: [PATCH 334/342] rename generation_parameters_copypaste module to infotext --- modules/{generation_parameters_copypaste.py => infotext.py} | 0 1 file changed, 0 insertions(+), 0 deletions(-) rename modules/{generation_parameters_copypaste.py => infotext.py} (100%) diff --git a/modules/generation_parameters_copypaste.py b/modules/infotext.py similarity index 100% rename from modules/generation_parameters_copypaste.py rename to modules/infotext.py From c5496c76461c90bd186ae8804aa65a33cd136d48 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Mon, 1 Jan 2024 13:52:37 +0300 Subject: [PATCH 335/342] infotext.py: add support for old modules.generation_parameters_copypaste name --- modules/infotext.py | 3 +++ 1 file changed, 3 insertions(+) diff --git a/modules/infotext.py b/modules/infotext.py index 86a36c327..bcbeb0fdc 100644 --- a/modules/infotext.py +++ b/modules/infotext.py @@ -4,12 +4,15 @@ import io import json import os import re +import sys import gradio as gr from modules.paths import data_path from modules import shared, ui_tempdir, script_callbacks, processing from PIL import Image +sys.modules['modules.generation_parameters_copypaste'] = sys.modules[__name__] # alias for old name + re_param_code = r'\s*(\w[\w \-/]+):\s*("(?:\\.|[^\\"])+"|[^,]*)(?:,|$)' re_param = re.compile(re_param_code) re_imagesize = re.compile(r"^(\d+)x(\d+)$") From d859cec696a953dbfd6f69f7735e68661748d579 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Mon, 1 Jan 2024 13:53:12 +0300 Subject: [PATCH 336/342] infotext.py: rename usages in the codebase --- .../scripts/extra_options_section.py | 4 ++-- modules/api/api.py | 10 +++++----- modules/img2img.py | 2 +- modules/postprocessing.py | 4 ++-- modules/processing.py | 4 ++-- modules/processing_scripts/refiner.py | 2 +- modules/processing_scripts/seed.py | 2 +- modules/shared_items.py | 4 ++-- modules/txt2img.py | 2 +- modules/ui.py | 4 ++-- modules/ui_common.py | 4 ++-- modules/ui_extra_networks.py | 2 +- modules/ui_extra_networks_user_metadata.py | 4 ++-- modules/ui_postprocessing.py | 2 +- 14 files changed, 25 insertions(+), 25 deletions(-) diff --git a/extensions-builtin/extra-options-section/scripts/extra_options_section.py b/extensions-builtin/extra-options-section/scripts/extra_options_section.py index ac2c3de46..8aa901fd4 100644 --- a/extensions-builtin/extra-options-section/scripts/extra_options_section.py +++ b/extensions-builtin/extra-options-section/scripts/extra_options_section.py @@ -1,7 +1,7 @@ import math import gradio as gr -from modules import scripts, shared, ui_components, ui_settings, generation_parameters_copypaste +from modules import scripts, shared, ui_components, ui_settings, infotext from modules.ui_components import FormColumn @@ -25,7 +25,7 @@ class ExtraOptionsSection(scripts.Script): extra_options = shared.opts.extra_options_img2img if is_img2img else shared.opts.extra_options_txt2img elem_id_tabname = "extra_options_" + ("img2img" if is_img2img else "txt2img") - mapping = {k: v for v, k in generation_parameters_copypaste.infotext_to_setting_name_mapping} + mapping = {k: v for v, k in infotext.infotext_to_setting_name_mapping} with gr.Blocks() as interface: with gr.Accordion("Options", open=False, elem_id=elem_id_tabname) if shared.opts.extra_options_accordion and extra_options else gr.Group(elem_id=elem_id_tabname): diff --git a/modules/api/api.py b/modules/api/api.py index 843c59b04..0e2807de2 100644 --- a/modules/api/api.py +++ b/modules/api/api.py @@ -17,7 +17,7 @@ from fastapi.encoders import jsonable_encoder from secrets import compare_digest import modules.shared as shared -from modules import sd_samplers, deepbooru, sd_hijack, images, scripts, ui, postprocessing, errors, restart, shared_items, script_callbacks, generation_parameters_copypaste, sd_models +from modules import sd_samplers, deepbooru, sd_hijack, images, scripts, ui, postprocessing, errors, restart, shared_items, script_callbacks, infotext, sd_models from modules.api import models from modules.shared import opts from modules.processing import StableDiffusionProcessingTxt2Img, StableDiffusionProcessingImg2Img, process_images @@ -369,9 +369,9 @@ class Api: if not request.infotext: return {} - possible_fields = generation_parameters_copypaste.paste_fields[tabname]["fields"] + possible_fields = infotext.paste_fields[tabname]["fields"] set_fields = request.model_dump(exclude_unset=True) if hasattr(request, "request") else request.dict(exclude_unset=True) # pydantic v1/v2 have differenrt names for this - params = generation_parameters_copypaste.parse_generation_parameters(request.infotext) + params = infotext.parse_generation_parameters(request.infotext) def get_field_value(field, params): value = field.function(params) if field.function else params.get(field.label) @@ -408,7 +408,7 @@ class Api: if request.override_settings is None: request.override_settings = {} - overriden_settings = generation_parameters_copypaste.get_override_settings(params) + overriden_settings = infotext.get_override_settings(params) for _, setting_name, value in overriden_settings: if setting_name not in request.override_settings: request.override_settings[setting_name] = value @@ -584,7 +584,7 @@ class Api: if geninfo is None: geninfo = "" - params = generation_parameters_copypaste.parse_generation_parameters(geninfo) + params = infotext.parse_generation_parameters(geninfo) script_callbacks.infotext_pasted_callback(geninfo, params) return models.PNGInfoResponse(info=geninfo, items=items, parameters=params) diff --git a/modules/img2img.py b/modules/img2img.py index c583290a0..75b3d346f 100644 --- a/modules/img2img.py +++ b/modules/img2img.py @@ -7,7 +7,7 @@ from PIL import Image, ImageOps, ImageFilter, ImageEnhance, UnidentifiedImageErr import gradio as gr from modules import images as imgutil -from modules.generation_parameters_copypaste import create_override_settings_dict, parse_generation_parameters +from modules.infotext import create_override_settings_dict, parse_generation_parameters from modules.processing import Processed, StableDiffusionProcessingImg2Img, process_images from modules.shared import opts, state from modules.sd_models import get_closet_checkpoint_match diff --git a/modules/postprocessing.py b/modules/postprocessing.py index 0c59fad48..f776f7b69 100644 --- a/modules/postprocessing.py +++ b/modules/postprocessing.py @@ -2,7 +2,7 @@ import os from PIL import Image -from modules import shared, images, devices, scripts, scripts_postprocessing, ui_common, generation_parameters_copypaste +from modules import shared, images, devices, scripts, scripts_postprocessing, ui_common, infotext from modules.shared import opts @@ -86,7 +86,7 @@ def run_postprocessing(extras_mode, image, image_folder, input_dir, output_dir, basename = '' forced_filename = None - infotext = ", ".join([k if k == v else f'{k}: {generation_parameters_copypaste.quote(v)}' for k, v in pp.info.items() if v is not None]) + infotext = ", ".join([k if k == v else f'{k}: {infotext.quote(v)}' for k, v in pp.info.items() if v is not None]) if opts.enable_pnginfo: pp.image.info = existing_pnginfo diff --git a/modules/processing.py b/modules/processing.py index 7789f9a42..b30df60db 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -16,7 +16,7 @@ from skimage import exposure from typing import Any import modules.sd_hijack -from modules import devices, prompt_parser, masking, sd_samplers, lowvram, generation_parameters_copypaste, extra_networks, sd_vae_approx, scripts, sd_samplers_common, sd_unet, errors, rng +from modules import devices, prompt_parser, masking, sd_samplers, lowvram, infotext, extra_networks, sd_vae_approx, scripts, sd_samplers_common, sd_unet, errors, rng from modules.rng import slerp # noqa: F401 from modules.sd_hijack import model_hijack from modules.sd_samplers_common import images_tensor_to_samples, decode_first_stage, approximation_indexes @@ -733,7 +733,7 @@ def create_infotext(p, all_prompts, all_seeds, all_subseeds, comments=None, iter "User": p.user if opts.add_user_name_to_info else None, } - generation_params_text = ", ".join([k if k == v else f'{k}: {generation_parameters_copypaste.quote(v)}' for k, v in generation_params.items() if v is not None]) + generation_params_text = ", ".join([k if k == v else f'{k}: {infotext.quote(v)}' for k, v in generation_params.items() if v is not None]) prompt_text = p.main_prompt if use_main_prompt else all_prompts[index] negative_prompt_text = f"\nNegative prompt: {p.main_negative_prompt if use_main_prompt else all_negative_prompts[index]}" if all_negative_prompts[index] else "" diff --git a/modules/processing_scripts/refiner.py b/modules/processing_scripts/refiner.py index cefad32b7..e9941413f 100644 --- a/modules/processing_scripts/refiner.py +++ b/modules/processing_scripts/refiner.py @@ -1,7 +1,7 @@ import gradio as gr from modules import scripts, sd_models -from modules.generation_parameters_copypaste import PasteField +from modules.infotext import PasteField from modules.ui_common import create_refresh_button from modules.ui_components import InputAccordion diff --git a/modules/processing_scripts/seed.py b/modules/processing_scripts/seed.py index a3e16a12e..602932785 100644 --- a/modules/processing_scripts/seed.py +++ b/modules/processing_scripts/seed.py @@ -3,7 +3,7 @@ import json import gradio as gr from modules import scripts, ui, errors -from modules.generation_parameters_copypaste import PasteField +from modules.infotext import PasteField from modules.shared import cmd_opts from modules.ui_components import ToolButton diff --git a/modules/shared_items.py b/modules/shared_items.py index 991971ad0..e13924720 100644 --- a/modules/shared_items.py +++ b/modules/shared_items.py @@ -67,14 +67,14 @@ def reload_hypernetworks(): def get_infotext_names(): - from modules import generation_parameters_copypaste, shared + from modules import infotext, shared res = {} for info in shared.opts.data_labels.values(): if info.infotext: res[info.infotext] = 1 - for tab_data in generation_parameters_copypaste.paste_fields.values(): + for tab_data in infotext.paste_fields.values(): for _, name in tab_data.get("fields") or []: if isinstance(name, str): res[name] = 1 diff --git a/modules/txt2img.py b/modules/txt2img.py index e4e18ceb6..3a481915f 100644 --- a/modules/txt2img.py +++ b/modules/txt2img.py @@ -2,7 +2,7 @@ from contextlib import closing import modules.scripts from modules import processing -from modules.generation_parameters_copypaste import create_override_settings_dict +from modules.infotext import create_override_settings_dict from modules.shared import opts import modules.shared as shared from modules.ui import plaintext_to_html diff --git a/modules/ui.py b/modules/ui.py index 9db2407ed..6451e14c1 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -21,14 +21,14 @@ from modules.ui_gradio_extensions import reload_javascript from modules.shared import opts, cmd_opts -import modules.generation_parameters_copypaste as parameters_copypaste +import modules.infotext as parameters_copypaste import modules.hypernetworks.ui as hypernetworks_ui import modules.textual_inversion.ui as textual_inversion_ui import modules.textual_inversion.textual_inversion as textual_inversion import modules.shared as shared from modules import prompt_parser from modules.sd_hijack import model_hijack -from modules.generation_parameters_copypaste import image_from_url_text, PasteField +from modules.infotext import image_from_url_text, PasteField create_setting_component = ui_settings.create_setting_component diff --git a/modules/ui_common.py b/modules/ui_common.py index 032ec4af7..fd32676f9 100644 --- a/modules/ui_common.py +++ b/modules/ui_common.py @@ -8,10 +8,10 @@ import gradio as gr import subprocess as sp from modules import call_queue, shared -from modules.generation_parameters_copypaste import image_from_url_text +from modules.infotext import image_from_url_text import modules.images from modules.ui_components import ToolButton -import modules.generation_parameters_copypaste as parameters_copypaste +import modules.infotext as parameters_copypaste folder_symbol = '\U0001f4c2' # 📂 refresh_symbol = '\U0001f504' # 🔄 diff --git a/modules/ui_extra_networks.py b/modules/ui_extra_networks.py index b8c022413..790af1356 100644 --- a/modules/ui_extra_networks.py +++ b/modules/ui_extra_networks.py @@ -10,7 +10,7 @@ import json import html from fastapi.exceptions import HTTPException -from modules.generation_parameters_copypaste import image_from_url_text +from modules.infotext import image_from_url_text from modules.ui_components import ToolButton extra_pages = [] diff --git a/modules/ui_extra_networks_user_metadata.py b/modules/ui_extra_networks_user_metadata.py index 36a807fcd..87aeb6f33 100644 --- a/modules/ui_extra_networks_user_metadata.py +++ b/modules/ui_extra_networks_user_metadata.py @@ -5,7 +5,7 @@ import os.path import gradio as gr -from modules import generation_parameters_copypaste, images, sysinfo, errors, ui_extra_networks +from modules import infotext, images, sysinfo, errors, ui_extra_networks class UserMetadataEditor: @@ -181,7 +181,7 @@ class UserMetadataEditor: index = len(gallery) - 1 if index >= len(gallery) else index img_info = gallery[index if index >= 0 else 0] - image = generation_parameters_copypaste.image_from_url_text(img_info) + image = infotext.image_from_url_text(img_info) geninfo, items = images.read_info_from_image(image) images.save_image_with_geninfo(image, geninfo, item["local_preview"]) diff --git a/modules/ui_postprocessing.py b/modules/ui_postprocessing.py index 13d888e48..b74a15323 100644 --- a/modules/ui_postprocessing.py +++ b/modules/ui_postprocessing.py @@ -1,6 +1,6 @@ import gradio as gr from modules import scripts, shared, ui_common, postprocessing, call_queue, ui_toprow -import modules.generation_parameters_copypaste as parameters_copypaste +import modules.infotext as parameters_copypaste def create_ui(): From d613cd17c72c753bd1e314dff74dc22d9a949374 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Mon, 1 Jan 2024 14:38:29 +0300 Subject: [PATCH 337/342] add automatic backwards version compatibility --- modules/infotext.py | 4 +++- modules/infotext_versions.py | 35 +++++++++++++++++++++++++++++++++++ modules/shared_options.py | 1 + 3 files changed, 39 insertions(+), 1 deletion(-) create mode 100644 modules/infotext_versions.py diff --git a/modules/infotext.py b/modules/infotext.py index bcbeb0fdc..7f30446b9 100644 --- a/modules/infotext.py +++ b/modules/infotext.py @@ -8,7 +8,7 @@ import sys import gradio as gr from modules.paths import data_path -from modules import shared, ui_tempdir, script_callbacks, processing +from modules import shared, ui_tempdir, script_callbacks, processing, infotext_versions, errors from PIL import Image sys.modules['modules.generation_parameters_copypaste'] = sys.modules[__name__] # alias for old name @@ -342,6 +342,8 @@ Steps: 20, Sampler: Euler a, CFG scale: 7, Seed: 965400086, Size: 512x512, Model if "Cache FP16 weight for LoRA" not in res and res["FP8 weight"] != "Disable": res["Cache FP16 weight for LoRA"] = False + infotext_versions.backcompat(res) + skip = set(shared.opts.infotext_skip_pasting) res = {k: v for k, v in res.items() if k not in skip} diff --git a/modules/infotext_versions.py b/modules/infotext_versions.py new file mode 100644 index 000000000..01e885a26 --- /dev/null +++ b/modules/infotext_versions.py @@ -0,0 +1,35 @@ +from modules import shared +from packaging import version +import re + + +v160 = version.parse("1.6.0") + + +def parse_version(text): + if text is None: + return None + + m = re.match(r'([^-]+-[^-]+)-.*', text) + if m: + text = m.group(1) + + try: + return version.parse(text) + except Exception as e: + return None + + +def backcompat(d): + """Checks infotext Version field, and enables backwards compatibility options according to it.""" + + if not shared.opts.auto_backcompat: + return + + ver = parse_version(d.get("Version")) + if ver is None: + return + + if ver < v160: + d["Old prompt editing timelines"] = True + diff --git a/modules/shared_options.py b/modules/shared_options.py index 752a4f125..281591da8 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -212,6 +212,7 @@ options_templates.update(options_section(('optimizations', "Optimizations", "sd" })) options_templates.update(options_section(('compatibility', "Compatibility", "sd"), { + "auto_backcompat": OptionInfo(True, "Automatic backward compatibility").info("automatically enable options for backwards compatibility when importing generation parameters from infotext that has program version."), "use_old_emphasis_implementation": OptionInfo(False, "Use old emphasis implementation. Can be useful to reproduce old seeds."), "use_old_karras_scheduler_sigmas": OptionInfo(False, "Use old karras scheduler sigmas (0.1 to 10)."), "no_dpmpp_sde_batch_determinism": OptionInfo(False, "Do not make DPM++ SDE deterministic across different batch sizes."), From 45b7bba3d06f2d4bc2fffc210cbfcb357b86add6 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Mon, 1 Jan 2024 14:51:56 +0300 Subject: [PATCH 338/342] add automatic version support for zero terminal SNR noise schedule option from #14145 --- modules/infotext_versions.py | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/modules/infotext_versions.py b/modules/infotext_versions.py index 01e885a26..9a204d842 100644 --- a/modules/infotext_versions.py +++ b/modules/infotext_versions.py @@ -4,6 +4,7 @@ import re v160 = version.parse("1.6.0") +v170_tsnr = version.parse("v1.7.0-225") def parse_version(text): @@ -33,3 +34,6 @@ def backcompat(d): if ver < v160: d["Old prompt editing timelines"] = True + if ver < v170_tsnr: + d["Downcast alphas_cumprod"] = True + From d8126be578c7d4579c0f2ee4adbe35500bc71ce6 Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Mon, 1 Jan 2024 15:00:39 +0300 Subject: [PATCH 339/342] linter --- modules/infotext.py | 2 +- modules/infotext_versions.py | 2 +- modules/postprocessing.py | 2 +- 3 files changed, 3 insertions(+), 3 deletions(-) diff --git a/modules/infotext.py b/modules/infotext.py index 7f30446b9..26e9b9493 100644 --- a/modules/infotext.py +++ b/modules/infotext.py @@ -8,7 +8,7 @@ import sys import gradio as gr from modules.paths import data_path -from modules import shared, ui_tempdir, script_callbacks, processing, infotext_versions, errors +from modules import shared, ui_tempdir, script_callbacks, processing, infotext_versions from PIL import Image sys.modules['modules.generation_parameters_copypaste'] = sys.modules[__name__] # alias for old name diff --git a/modules/infotext_versions.py b/modules/infotext_versions.py index 9a204d842..a5afeebf1 100644 --- a/modules/infotext_versions.py +++ b/modules/infotext_versions.py @@ -17,7 +17,7 @@ def parse_version(text): try: return version.parse(text) - except Exception as e: + except Exception: return None diff --git a/modules/postprocessing.py b/modules/postprocessing.py index f776f7b69..facea899f 100644 --- a/modules/postprocessing.py +++ b/modules/postprocessing.py @@ -2,7 +2,7 @@ import os from PIL import Image -from modules import shared, images, devices, scripts, scripts_postprocessing, ui_common, infotext +from modules import shared, images, devices, scripts, scripts_postprocessing, ui_common from modules.shared import opts From 0743ee9b3eda8dd4ceea625d710031577201f4ad Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Mon, 1 Jan 2024 15:50:47 +0300 Subject: [PATCH 340/342] re-layout checkboxes for XYZ grid a bit --- scripts/xyz_grid.py | 13 ++++++------- 1 file changed, 6 insertions(+), 7 deletions(-) diff --git a/scripts/xyz_grid.py b/scripts/xyz_grid.py index 34267c2c3..2d5509947 100644 --- a/scripts/xyz_grid.py +++ b/scripts/xyz_grid.py @@ -438,17 +438,16 @@ class Script(scripts.Script): with gr.Column(): draw_legend = gr.Checkbox(label='Draw legend', value=True, elem_id=self.elem_id("draw_legend")) no_fixed_seeds = gr.Checkbox(label='Keep -1 for seeds', value=False, elem_id=self.elem_id("no_fixed_seeds")) + with gr.Row(): + vary_seeds_x = gr.Checkbox(label='Vary seeds for X', value=False, min_width=80, elem_id=self.elem_id("vary_seeds_x"), tooltip="Use different seeds for images along X axis.") + vary_seeds_y = gr.Checkbox(label='Vary seeds for Y', value=False, min_width=80, elem_id=self.elem_id("vary_seeds_y"), tooltip="Use different seeds for images along Y axis.") + vary_seeds_z = gr.Checkbox(label='Vary seeds for Z', value=False, min_width=80, elem_id=self.elem_id("vary_seeds_z"), tooltip="Use different seeds for images along Z axis.") with gr.Column(): include_lone_images = gr.Checkbox(label='Include Sub Images', value=False, elem_id=self.elem_id("include_lone_images")) include_sub_grids = gr.Checkbox(label='Include Sub Grids', value=False, elem_id=self.elem_id("include_sub_grids")) - with gr.Column(): - vary_seeds_x = gr.Checkbox(label='Vary seed on X axis', value=False, elem_id=self.elem_id("vary_seeds_x")) - vary_seeds_y = gr.Checkbox(label='Vary seed on Y axis', value=False, elem_id=self.elem_id("vary_seeds_y")) - vary_seeds_z = gr.Checkbox(label='Vary seed on Z axis', value=False, elem_id=self.elem_id("vary_seeds_z")) + csv_mode = gr.Checkbox(label='Use text inputs instead of dropdowns', value=False, elem_id=self.elem_id("csv_mode")) with gr.Column(): margin_size = gr.Slider(label="Grid margins (px)", minimum=0, maximum=500, value=0, step=2, elem_id=self.elem_id("margin_size")) - with gr.Column(): - csv_mode = gr.Checkbox(label='Use text inputs instead of dropdowns', value=False, elem_id=self.elem_id("csv_mode")) with gr.Row(variant="compact", elem_id="swap_axes"): swap_xy_axes_button = gr.Button(value="Swap X/Y axes", elem_id="xy_grid_swap_axes_button") @@ -531,7 +530,7 @@ class Script(scripts.Script): return [x_type, x_values, x_values_dropdown, y_type, y_values, y_values_dropdown, z_type, z_values, z_values_dropdown, draw_legend, include_lone_images, include_sub_grids, no_fixed_seeds, vary_seeds_x, vary_seeds_y, vary_seeds_z, margin_size, csv_mode] - def run(self, p, x_type, x_values, x_values_dropdown, y_type, y_values, y_values_dropdown, z_type, z_values, z_values_dropdown, draw_legend, include_lone_images, include_sub_grids, no_fixed_seeds, margin_size, csv_mode): + def run(self, p, x_type, x_values, x_values_dropdown, y_type, y_values, y_values_dropdown, z_type, z_values, z_values_dropdown, draw_legend, include_lone_images, include_sub_grids, no_fixed_seeds, vary_seeds_x, vary_seeds_y, vary_seeds_z, margin_size, csv_mode): x_type, y_type, z_type = x_type or 0, y_type or 0, z_type or 0 # if axle type is None set to 0 if not no_fixed_seeds: From ac0ecf3b4b9d147743c04f0ff4ddc4cf4595e11d Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Mon, 1 Jan 2024 16:28:58 +0300 Subject: [PATCH 341/342] option to convert VAE to bfloat16 (implementation of #9295) --- modules/processing.py | 23 ++++++++++++++++++----- modules/shared_options.py | 1 + 2 files changed, 19 insertions(+), 5 deletions(-) diff --git a/modules/processing.py b/modules/processing.py index 846e4796a..f06568821 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -628,20 +628,33 @@ def decode_latent_batch(model, batch, target_device=None, check_for_nans=False): sample = decode_first_stage(model, batch[i:i + 1])[0] if check_for_nans: + try: devices.test_for_nans(sample, "vae") except devices.NansException as e: - if devices.dtype_vae == torch.float32 or not shared.opts.auto_vae_precision: + if shared.opts.auto_vae_precision_bfloat16: + autofix_dtype = torch.bfloat16 + autofix_dtype_text = "bfloat16" + autofix_dtype_setting = "Automatically convert VAE to bfloat16" + autofix_dtype_comment = "" + elif shared.opts.auto_vae_precision: + autofix_dtype = torch.float32 + autofix_dtype_text = "32-bit float" + autofix_dtype_setting = "Automatically revert VAE to 32-bit floats" + autofix_dtype_comment = "\nTo always start with 32-bit VAE, use --no-half-vae commandline flag." + else: + raise e + + if devices.dtype_vae == autofix_dtype: raise e errors.print_error_explanation( "A tensor with all NaNs was produced in VAE.\n" - "Web UI will now convert VAE into 32-bit float and retry.\n" - "To disable this behavior, disable the 'Automatically revert VAE to 32-bit floats' setting.\n" - "To always start with 32-bit VAE, use --no-half-vae commandline flag." + f"Web UI will now convert VAE into {autofix_dtype_text} and retry.\n" + f"To disable this behavior, disable the '{autofix_dtype_setting}' setting.{autofix_dtype_comment}" ) - devices.dtype_vae = torch.float32 + devices.dtype_vae = autofix_dtype model.first_stage_model.to(devices.dtype_vae) batch = batch.to(devices.dtype_vae) diff --git a/modules/shared_options.py b/modules/shared_options.py index ce06f022e..e813546f6 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -177,6 +177,7 @@ For img2img, VAE is used to process user's input image before the sampling, and "sd_vae_checkpoint_cache": OptionInfo(0, "VAE Checkpoints to cache in RAM", gr.Slider, {"minimum": 0, "maximum": 10, "step": 1}), "sd_vae": OptionInfo("Automatic", "SD VAE", gr.Dropdown, lambda: {"choices": shared_items.sd_vae_items()}, refresh=shared_items.refresh_vae_list, infotext='VAE').info("choose VAE model: Automatic = use one with same filename as checkpoint; None = use VAE from checkpoint"), "sd_vae_overrides_per_model_preferences": OptionInfo(True, "Selected VAE overrides per-model preferences").info("you can set per-model VAE either by editing user metadata for checkpoints, or by making the VAE have same name as checkpoint"), + "auto_vae_precision_bfloat16": OptionInfo(False, "Automatically convert VAE to bfloat16").info("triggers when a tensor with NaNs is produced in VAE; disabling the option in this case will result in a black square image; if enabled, overrides the option below"), "auto_vae_precision": OptionInfo(True, "Automatically revert VAE to 32-bit floats").info("triggers when a tensor with NaNs is produced in VAE; disabling the option in this case will result in a black square image"), "sd_vae_encode_method": OptionInfo("Full", "VAE type for encode", gr.Radio, {"choices": ["Full", "TAESD"]}, infotext='VAE Encoder').info("method to encode image to latent (use in img2img, hires-fix or inpaint mask)"), "sd_vae_decode_method": OptionInfo("Full", "VAE type for decode", gr.Radio, {"choices": ["Full", "TAESD"]}, infotext='VAE Decoder').info("method to decode latent to image"), From 0aa7c53c0b9469849377aff83f43c9f75c19b3fa Mon Sep 17 00:00:00 2001 From: AUTOMATIC1111 <16777216c@gmail.com> Date: Mon, 1 Jan 2024 16:50:59 +0300 Subject: [PATCH 342/342] fix borked merge, rename fields to better match what they do, change setting default to true for #13653 --- modules/call_queue.py | 2 +- modules/img2img.py | 2 +- modules/processing.py | 2 +- modules/shared_options.py | 2 +- modules/shared_state.py | 12 ++++++------ modules/ui_toprow.py | 8 +++++++- scripts/loopback.py | 4 ++-- scripts/xyz_grid.py | 2 +- 8 files changed, 20 insertions(+), 14 deletions(-) diff --git a/modules/call_queue.py b/modules/call_queue.py index 01c6d17f6..bcd7c5462 100644 --- a/modules/call_queue.py +++ b/modules/call_queue.py @@ -78,7 +78,7 @@ def wrap_gradio_call(func, extra_outputs=None, add_stats=False): shared.state.skipped = False shared.state.interrupted = False - shared.state.interrupted_next = False + shared.state.stopping_generation = False shared.state.job_count = 0 if not add_stats: diff --git a/modules/img2img.py b/modules/img2img.py index 829faa818..e7e8e2510 100644 --- a/modules/img2img.py +++ b/modules/img2img.py @@ -51,7 +51,7 @@ def process_batch(p, input_dir, output_dir, inpaint_mask_dir, args, to_scale=Fal if state.skipped: state.skipped = False - if state.interrupted or state.interrupted_next: + if state.interrupted or state.stopping_generation: break try: diff --git a/modules/processing.py b/modules/processing.py index 00de2ed2e..f55b85ed3 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -865,7 +865,7 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed: if state.skipped: state.skipped = False - if state.interrupted or state.interrupted_next: + if state.interrupted or state.stopping_generation: break sd_models.reload_model_weights() # model can be changed for example by refiner diff --git a/modules/shared_options.py b/modules/shared_options.py index 7852e0ea3..7581e276e 100644 --- a/modules/shared_options.py +++ b/modules/shared_options.py @@ -120,7 +120,6 @@ options_templates.update(options_section(('system', "System", "system"), { "disable_mmap_load_safetensors": OptionInfo(False, "Disable memmapping for loading .safetensors files.").info("fixes very slow loading speed in some cases"), "hide_ldm_prints": OptionInfo(True, "Prevent Stability-AI's ldm/sgm modules from printing noise to console."), "dump_stacks_on_signal": OptionInfo(False, "Print stack traces before exiting the program with ctrl+c."), - "interrupt_after_current": OptionInfo(False, "Interrupt generation after current image is finished on batch processing"), })) options_templates.update(options_section(('API', "API", "system"), { @@ -286,6 +285,7 @@ options_templates.update(options_section(('ui_alternatives', "UI alternatives", "hires_fix_show_prompts": OptionInfo(False, "Hires fix: show hires prompt and negative prompt").needs_reload_ui(), "txt2img_settings_accordion": OptionInfo(False, "Settings in txt2img hidden under Accordion").needs_reload_ui(), "img2img_settings_accordion": OptionInfo(False, "Settings in img2img hidden under Accordion").needs_reload_ui(), + "interrupt_after_current": OptionInfo(True, "Don't Interrupt in the middle").info("when using Interrupt button, if generating more than one image, stop after the generation of an image has finished, instead of immediately"), })) options_templates.update(options_section(('ui', "User interface", "ui"), { diff --git a/modules/shared_state.py b/modules/shared_state.py index 532fdcd8d..33996691c 100644 --- a/modules/shared_state.py +++ b/modules/shared_state.py @@ -12,7 +12,7 @@ log = logging.getLogger(__name__) class State: skipped = False interrupted = False - interrupted_next = False + stopping_generation = False job = "" job_no = 0 job_count = 0 @@ -80,9 +80,9 @@ class State: self.interrupted = True log.info("Received interrupt request") - def interrupt_next(self): - self.interrupted_next = True - log.info("Received interrupt request, interrupt after current job") + def stop_generating(self): + self.stopping_generation = True + log.info("Received stop generating request") def nextjob(self): if shared.opts.live_previews_enable and shared.opts.show_progress_every_n_steps == -1: @@ -96,7 +96,7 @@ class State: obj = { "skipped": self.skipped, "interrupted": self.interrupted, - "interrupted_next": self.interrupted_next, + "stopping_generation": self.stopping_generation, "job": self.job, "job_count": self.job_count, "job_timestamp": self.job_timestamp, @@ -120,7 +120,7 @@ class State: self.id_live_preview = 0 self.skipped = False self.interrupted = False - self.interrupted_next = False + self.stopping_generation = False self.textinfo = None self.job = job devices.torch_gc() diff --git a/modules/ui_toprow.py b/modules/ui_toprow.py index 9caf8faa2..1abc9117b 100644 --- a/modules/ui_toprow.py +++ b/modules/ui_toprow.py @@ -106,8 +106,14 @@ class Toprow: outputs=[], ) + def interrupt_function(): + if shared.state.job_count > 1 and shared.opts.interrupt_after_current: + shared.state.stop_generating() + else: + shared.state.interrupt() + self.interrupt.click( - fn=lambda: shared.state.interrupt(), + fn=interrupt_function, inputs=[], outputs=[], ) diff --git a/scripts/loopback.py b/scripts/loopback.py index ad921269a..800ee882a 100644 --- a/scripts/loopback.py +++ b/scripts/loopback.py @@ -95,7 +95,7 @@ class Script(scripts.Script): processed = processing.process_images(p) # Generation cancelled. - if state.interrupted or state.interrupted_next: + if state.interrupted or state.stopping_generation: break if initial_seed is None: @@ -122,7 +122,7 @@ class Script(scripts.Script): p.inpainting_fill = original_inpainting_fill - if state.interrupted or state.interrupted_next: + if state.interrupted or state.stopping_generation: break if len(history) > 1: diff --git a/scripts/xyz_grid.py b/scripts/xyz_grid.py index 2deff365f..2f385ebf2 100644 --- a/scripts/xyz_grid.py +++ b/scripts/xyz_grid.py @@ -696,7 +696,7 @@ class Script(scripts.Script): grid_infotext = [None] * (1 + len(zs)) def cell(x, y, z, ix, iy, iz): - if shared.state.interrupted or state.interrupted_next: + if shared.state.interrupted or state.stopping_generation: return Processed(p, [], p.seed, "") pc = copy(p)