stable-diffusion-webui/modules/sd_samplers_timesteps.py

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

167 lines
6.3 KiB
Python
Raw Normal View History

import torch
import inspect
2023-08-08 13:27:32 -06:00
import sys
from modules import devices, sd_samplers_common, sd_samplers_timesteps_impl
from modules.sd_samplers_cfg_denoiser import CFGDenoiser
2023-08-16 19:45:19 -06:00
from modules.script_callbacks import ExtraNoiseParams, extra_noise_callback
from modules.shared import opts
import modules.shared as shared
samplers_timesteps = [
2023-08-08 12:07:18 -06:00
('DDIM', sd_samplers_timesteps_impl.ddim, ['ddim'], {}),
('PLMS', sd_samplers_timesteps_impl.plms, ['plms'], {}),
('UniPC', sd_samplers_timesteps_impl.unipc, ['unipc'], {}),
]
samplers_data_timesteps = [
sd_samplers_common.SamplerData(label, lambda model, funcname=funcname: CompVisSampler(funcname, model), aliases, options)
for label, funcname, aliases, options in samplers_timesteps
]
class CompVisTimestepsDenoiser(torch.nn.Module):
def __init__(self, model, *args, **kwargs):
super().__init__(*args, **kwargs)
self.inner_model = model
def forward(self, input, timesteps, **kwargs):
return self.inner_model.apply_model(input, timesteps, **kwargs)
class CompVisTimestepsVDenoiser(torch.nn.Module):
def __init__(self, model, *args, **kwargs):
super().__init__(*args, **kwargs)
self.inner_model = model
def predict_eps_from_z_and_v(self, x_t, t, v):
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)
e_t = self.predict_eps_from_z_and_v(input, timesteps, model_output)
return e_t
class CFGDenoiserTimesteps(CFGDenoiser):
2023-08-08 13:09:40 -06:00
def __init__(self, sampler):
super().__init__(sampler)
2023-08-08 13:09:40 -06:00
self.alphas = shared.sd_model.alphas_cumprod
self.mask_before_denoising = True
def get_pred_x0(self, x_in, x_out, sigma):
2023-08-13 23:49:02 -06:00
ts = sigma.to(dtype=int)
2023-08-13 23:49:02 -06:00
a_t = self.alphas[ts][:, None, None, None]
sqrt_one_minus_at = (1 - a_t).sqrt()
pred_x0 = (x_in - sqrt_one_minus_at * x_out) / a_t.sqrt()
return pred_x0
2023-08-08 13:09:40 -06:00
@property
def inner_model(self):
if self.model_wrap is None:
denoiser = CompVisTimestepsVDenoiser if shared.sd_model.parameterization == "v" else CompVisTimestepsDenoiser
self.model_wrap = denoiser(shared.sd_model)
return self.model_wrap
class CompVisSampler(sd_samplers_common.Sampler):
def __init__(self, funcname, sd_model):
super().__init__(funcname)
self.eta_option_field = 'eta_ddim'
self.eta_infotext_field = 'Eta DDIM'
self.eta_default = 0.0
2023-08-08 13:09:40 -06:00
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)
if opts.always_discard_next_to_last_sigma and not discard_next_to_last_sigma:
discard_next_to_last_sigma = True
p.extra_generation_params["Discard penultimate sigma"] = True
steps += 1 if discard_next_to_last_sigma else 0
timesteps = torch.clip(torch.asarray(list(range(0, 1000, 1000 // steps)), device=devices.device) + 1, 0, 999)
return timesteps
def sample_img2img(self, p, x, noise, conditioning, unconditional_conditioning, steps=None, image_conditioning=None):
steps, t_enc = sd_samplers_common.setup_img2img_steps(p, steps)
timesteps = self.get_timesteps(p, steps)
timesteps_sched = timesteps[:t_enc]
alphas_cumprod = shared.sd_model.alphas_cumprod
sqrt_alpha_cumprod = torch.sqrt(alphas_cumprod[timesteps[t_enc]])
sqrt_one_minus_alpha_cumprod = torch.sqrt(1 - alphas_cumprod[timesteps[t_enc]])
xi = x * sqrt_alpha_cumprod + noise * sqrt_one_minus_alpha_cumprod
2023-08-15 00:19:19 -06:00
if opts.img2img_extra_noise > 0:
p.extra_generation_params["Extra noise"] = opts.img2img_extra_noise
extra_noise_params = ExtraNoiseParams(noise, x, xi)
2023-08-16 19:45:19 -06:00
extra_noise_callback(extra_noise_params)
noise = extra_noise_params.noise
2023-08-15 00:19:19 -06:00
xi += noise * opts.img2img_extra_noise * sqrt_alpha_cumprod
extra_params_kwargs = self.initialize(p)
parameters = inspect.signature(self.func).parameters
if 'timesteps' in parameters:
extra_params_kwargs['timesteps'] = timesteps_sched
if 'is_img2img' in parameters:
extra_params_kwargs['is_img2img'] = True
self.model_wrap_cfg.init_latent = x
self.last_latent = x
2023-08-08 13:09:40 -06:00
self.sampler_extra_args = {
'cond': conditioning,
'image_cond': image_conditioning,
'uncond': unconditional_conditioning,
'cond_scale': p.cfg_scale,
's_min_uncond': self.s_min_uncond
}
2023-08-08 13:09:40 -06:00
samples = self.launch_sampling(t_enc + 1, lambda: self.func(self.model_wrap_cfg, xi, extra_args=self.sampler_extra_args, disable=False, callback=self.callback_state, **extra_params_kwargs))
2024-01-27 12:30:12 -07:00
self.add_infotext(p)
return samples
def sample(self, p, x, conditioning, unconditional_conditioning, steps=None, image_conditioning=None):
steps = steps or p.steps
timesteps = self.get_timesteps(p, steps)
extra_params_kwargs = self.initialize(p)
parameters = inspect.signature(self.func).parameters
if 'timesteps' in parameters:
extra_params_kwargs['timesteps'] = timesteps
self.last_latent = x
2023-08-08 13:09:40 -06:00
self.sampler_extra_args = {
'cond': conditioning,
'image_cond': image_conditioning,
'uncond': unconditional_conditioning,
'cond_scale': p.cfg_scale,
's_min_uncond': self.s_min_uncond
2023-08-08 13:09:40 -06:00
}
samples = self.launch_sampling(steps, lambda: self.func(self.model_wrap_cfg, x, extra_args=self.sampler_extra_args, disable=False, callback=self.callback_state, **extra_params_kwargs))
2024-01-27 12:30:12 -07:00
self.add_infotext(p)
return samples
2023-08-08 13:27:32 -06:00
sys.modules['modules.sd_samplers_compvis'] = sys.modules[__name__]
VanillaStableDiffusionSampler = CompVisSampler # temp. compatibility with older extensions