Commit Graph

3456 Commits

Author SHA1 Message Date
Max Audron 6b3981c068 clean up unused script_path imports 2023-01-27 14:44:34 +01:00
Max Audron 14c0884fd0 use python importlib to load and execute extension modules
previously module attributes like __file__ where not set correctly,
leading to scripts getting the directory of the stable-diffusion repo
location instead of their own script.

This causes problem when loading user data from an external location
using the --data-dir flag, as extensions would look for their own code
in the stable-diffusion repo location instead of the data dir location.

Using pythons importlib functions sets the modules specs correctly and
executes them. But this will break extensions if they build paths based
on the previously incorrect __file__ attribute.
2023-01-27 14:44:34 +01:00
Max Audron 5eee2ac398 add data-dir flag and set all user data directories based on it 2023-01-27 14:44:30 +01:00
AUTOMATIC 9beb794e0b clarify the option to disable NaN check. 2023-01-27 13:08:00 +03:00
AUTOMATIC 6f31d2210c support detecting midas model
fix broken api for checkpoint list
2023-01-27 11:54:19 +03:00
AUTOMATIC d2ac95fa7b remove the need to place configs near models 2023-01-27 11:28:12 +03:00
AUTOMATIC 7a14c8ab45 add an option to enable sections from extras tab in txt2img/img2img
fix some style inconsistenices
2023-01-26 23:31:32 +03:00
AUTOMATIC1111 645f4e7ef8
Merge pull request #7234 from brkirch/fix-full-previews
Fix full previews and--no-half-vae to work correctly with --upcast-sampling
2023-01-26 14:48:43 +03:00
brkirch 10421f93c3 Fix full previews, --no-half-vae 2023-01-26 01:43:35 -05:00
AUTOMATIC 6cff440182 fix prompt editing break after first batch in img2img 2023-01-25 23:25:40 +03:00
AUTOMATIC d1d6ce2983 add edit_image_conditioning from my earlier edits in case there's an attempt to inegrate pix2pix properly
this allows to use pix2pix model in img2img though it won't work well this way
2023-01-25 23:25:25 +03:00
AUTOMATIC1111 3cead6983e
Merge pull request #7197 from mcmonkey4eva/fix-ti-symlinks
allow symlinks in the textual inversion embeddings folder
2023-01-25 22:59:12 +03:00
AUTOMATIC1111 a85e22a127
Merge pull request #7201 from brkirch/update-macos-defaults
Update default Mac command line arguments to use --upcast-sampling instead of --no-half
2023-01-25 22:57:17 +03:00
brkirch e0df864b8c Update arguments to use --upcast-sampling 2023-01-25 13:19:06 -05:00
AUTOMATIC 15e89ef0f6 fix for unet hijack breaking the train tab 2023-01-25 20:11:01 +03:00
AUTOMATIC 789d47f832 make clicking extra networks button one more time close the extra networks UI 2023-01-25 19:55:31 +03:00
Alex "mcmonkey" Goodwin e179b6098a allow symlinks in the textual inversion embeddings folder 2023-01-25 08:48:40 -08:00
AUTOMATIC 635499e832 add pix2pix credits 2023-01-25 19:42:26 +03:00
AUTOMATIC1111 1574e96729
Merge pull request #6510 from brkirch/unet16-upcast-precision
Add upcast options, full precision sampling from float16 UNet and upcasting attention for inference using SD 2.1 models without --no-half
2023-01-25 19:12:29 +03:00
AUTOMATIC1111 1982ef6890
Merge pull request #7138 from mykeehu/patch-4
Fix extra network thumbs label color
2023-01-25 18:59:11 +03:00
AUTOMATIC 57c1baa774 change to code for live preview fix on OSX to be bit more obvious 2023-01-25 18:56:23 +03:00
AUTOMATIC1111 23dafe6d86
Merge pull request #7151 from brkirch/fix-approx-nn
Fix Approx NN previews changing first generation result
2023-01-25 18:48:25 +03:00
AUTOMATIC1111 11485659dc
Merge pull request #7195 from Klace/instruct-pix2pix_model_load
Add instruct-pix2pix hijack
2023-01-25 18:33:15 +03:00
Kyle bd9b55ee90 Update requirements transformers==4.25.1
Update requirement for transformers to version 4.25.1 to allow instruct-pix2pix demo code to work
2023-01-25 09:41:41 -05:00
Kyle ee0a0da324 Add instruct-pix2pix hijack
Allows loading instruct-pix2pix models via same method as inpainting models in sd_models.py and sd_hijack_ip2p.py

Adds ddpm_edit.py necessary for instruct-pix2pix
2023-01-25 08:53:23 -05:00
AUTOMATIC1111 d5ce044bcd
Merge pull request #7146 from EllangoK/master
Adds X/Y/Z Grid Script
2023-01-25 11:56:26 +03:00
AUTOMATIC 1bfec873fa add an experimental option to apply loras to outputs rather than inputs 2023-01-25 11:29:46 +03:00
brkirch e3b53fd295 Add UI setting for upcasting attention to float32
Adds "Upcast cross attention layer to float32" option in Stable Diffusion settings. This allows for generating images using SD 2.1 models without --no-half or xFormers.

In order to make upcasting cross attention layer optimizations possible it is necessary to indent several sections of code in sd_hijack_optimizations.py so that a context manager can be used to disable autocast. Also, even though Stable Diffusion (and Diffusers) only upcast q and k, unfortunately my findings were that most of the cross attention layer optimizations could not function unless v is upcast also.
2023-01-25 01:13:04 -05:00
brkirch 84d9ce30cb Add option for float32 sampling with float16 UNet
This also handles type casting so that ROCm and MPS torch devices work correctly without --no-half. One cast is required for deepbooru in deepbooru_model.py, some explicit casting is required for img2img and inpainting. depth_model can't be converted to float16 or it won't work correctly on some systems (it's known to have issues on MPS) so in sd_models.py model.depth_model is removed for model.half().
2023-01-25 01:13:02 -05:00
AUTOMATIC 48a15821de remove the pip install stuff because it does not work as i hoped it would 2023-01-25 00:49:16 +03:00
AUTOMATIC bef1931895 add fastapi to requirements 2023-01-24 23:50:04 +03:00
AUTOMATIC 93fad28a97 print progress when installing torch
add PIP_INSTALLER_LOCATION env var to install pip if it's not installed
remove accidental call to accelerate when venv is disabled
add another env var to skip venv - SKIP_VENV
2023-01-24 21:13:05 +03:00
AUTOMATIC 5228ec8bda remove fairscale requirement, add fake fairscale to make BLIP not complain about it mk2 2023-01-24 20:30:43 +03:00
AUTOMATIC 28189985e6 remove fairscale requirement, add fake fairscale to make BLIP not complain about it 2023-01-24 20:24:27 +03:00
AUTOMATIC dac45299dd make git commands not fail for extensions when you have spaces in webui directory 2023-01-24 20:22:19 +03:00
EllangoK ec8774729e swaps xyz axes internally if one costs more 2023-01-24 02:53:35 -05:00
EllangoK e46bfa5a9e handling sub grids and merging into one 2023-01-24 02:24:32 -05:00
EllangoK 9fc354e130 implements most of xyz grid script 2023-01-24 02:22:40 -05:00
EllangoK d30ac02f28 renamed xy to xyz grid
this is mostly just so git can detect it properly
2023-01-24 02:21:32 -05:00
AUTOMATIC 602a1864b0 also return the removed field to sdapi/v1/upscalers because someone might have relied on it existing 2023-01-24 10:09:30 +03:00
AUTOMATIC 42a70d7477 repair sdapi/v1/upscalers returning bogus results 2023-01-24 10:05:45 +03:00
AUTOMATIC1111 8b903322e6
Merge pull request #7140 from vladmandic/api-decode-image
Add exception handling to API image decode
2023-01-24 09:54:20 +03:00
AUTOMATIC1111 848ef919b3
Merge pull request #7144 from catboxanon/swinir-interrupt
Make SwinIR upscaler interruptible and skippable
2023-01-24 09:51:53 +03:00
AUTOMATIC1111 393e09c1c3
Merge pull request #7148 from acncagua/improvement_launch.py
Set Linux xformers 0.0.16RC425
2023-01-24 09:38:10 +03:00
brkirch f64af77adc Fix different first gen with Approx NN previews
The loading of the model for approx nn live previews can change the internal state of PyTorch, resulting in a different image. This can be avoided by preloading the approx nn model in advance.
2023-01-23 22:49:20 -05:00
acncagua 078e16e4d3 Set Linux xformers 0.0.16RC425 2023-01-24 12:21:07 +09:00
catboxanon 3c47b05036
Also make SwinIR skippable 2023-01-23 22:00:27 -05:00
catboxanon f993525820
Make SwinIR interruptible 2023-01-23 21:50:59 -05:00
Vladimir Mandic 45e270dfc8
add image decod exception handling 2023-01-23 17:11:22 -05:00
Mykeehu 82a28bfe35
Fix extra network thumbs label color
Added white color for labels.
2023-01-23 22:36:27 +01:00