stable-diffusion-webui/modules/sd_samplers_cfg_denoiser.py

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

204 lines
9.5 KiB
Python
Raw Normal View History

import torch
2023-08-08 23:35:31 +08:00
from modules import prompt_parser, devices, sd_samplers_common
from modules.shared import opts, state
import modules.shared as shared
from modules.script_callbacks import CFGDenoiserParams, cfg_denoiser_callback
2023-02-11 10:18:38 +08:00
from modules.script_callbacks import CFGDenoisedParams, cfg_denoised_callback
from modules.script_callbacks import AfterCFGCallbackParams, cfg_after_cfg_callback
2023-07-12 02:16:43 +08:00
def catenate_conds(conds):
if not isinstance(conds[0], dict):
return torch.cat(conds)
return {key: torch.cat([x[key] for x in conds]) for key in conds[0].keys()}
def subscript_cond(cond, a, b):
if not isinstance(cond, dict):
return cond[a:b]
return {key: vec[a:b] for key, vec in cond.items()}
def pad_cond(tensor, repeats, empty):
if not isinstance(tensor, dict):
return torch.cat([tensor, empty.repeat((tensor.shape[0], repeats, 1))], axis=1)
tensor['crossattn'] = pad_cond(tensor['crossattn'], repeats, empty)
return tensor
class CFGDenoiser(torch.nn.Module):
"""
Classifier free guidance denoiser. A wrapper for stable diffusion model (specifically for unet)
that can take a noisy picture and produce a noise-free picture using two guidances (prompts)
instead of one. Originally, the second prompt is just an empty string, but we use non-empty
negative prompt.
"""
def __init__(self, model, sampler):
super().__init__()
self.inner_model = model
self.mask = None
self.nmask = None
self.init_latent = None
2022-09-15 18:10:16 +08:00
self.step = 0
self.image_cfg_scale = None
self.padded_cond_uncond = False
self.sampler = sampler
def combine_denoised(self, x_out, conds_list, uncond, cond_scale):
denoised_uncond = x_out[-uncond.shape[0]:]
denoised = torch.clone(denoised_uncond)
for i, conds in enumerate(conds_list):
for cond_index, weight in conds:
denoised[i] += (x_out[cond_index] - denoised_uncond[i]) * (weight * cond_scale)
return denoised
def combine_denoised_for_edit_model(self, x_out, cond_scale):
out_cond, out_img_cond, out_uncond = x_out.chunk(3)
denoised = out_uncond + cond_scale * (out_cond - out_img_cond) + self.image_cfg_scale * (out_img_cond - out_uncond)
return denoised
def get_pred_x0(self, x_in, x_out, sigma):
return x_out
2023-03-29 06:18:28 +08:00
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
# at self.image_cfg_scale == 1.0 produced results for edit model are the same as with normal sampling,
# so is_edit_model is set to False to support AND composition.
is_edit_model = shared.sd_model.cond_stage_key == "edit" and self.image_cfg_scale is not None and self.image_cfg_scale != 1.0
conds_list, tensor = prompt_parser.reconstruct_multicond_batch(cond, self.step)
2022-09-15 18:10:16 +08:00
uncond = prompt_parser.reconstruct_cond_batch(uncond, self.step)
2023-05-10 16:05:02 +08:00
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 self.mask is not None:
x = self.init_latent * self.mask + self.nmask * x
batch_size = len(conds_list)
repeats = [len(conds_list[i]) for i in range(batch_size)]
if shared.sd_model.model.conditioning_key == "crossattn-adm":
image_uncond = torch.zeros_like(image_cond)
2023-07-12 02:16:43 +08:00
make_condition_dict = lambda c_crossattn, c_adm: {"c_crossattn": [c_crossattn], "c_adm": c_adm}
else:
image_uncond = image_cond
2023-07-12 02:16:43 +08:00
if isinstance(uncond, dict):
make_condition_dict = lambda c_crossattn, c_concat: {**c_crossattn, "c_concat": [c_concat]}
else:
make_condition_dict = lambda c_crossattn, c_concat: {"c_crossattn": [c_crossattn], "c_concat": [c_concat]}
if not is_edit_model:
x_in = torch.cat([torch.stack([x[i] for _ in range(n)]) for i, n in enumerate(repeats)] + [x])
sigma_in = torch.cat([torch.stack([sigma[i] for _ in range(n)]) for i, n in enumerate(repeats)] + [sigma])
image_cond_in = torch.cat([torch.stack([image_cond[i] for _ in range(n)]) for i, n in enumerate(repeats)] + [image_uncond])
else:
x_in = torch.cat([torch.stack([x[i] for _ in range(n)]) for i, n in enumerate(repeats)] + [x] + [x])
sigma_in = torch.cat([torch.stack([sigma[i] for _ in range(n)]) for i, n in enumerate(repeats)] + [sigma] + [sigma])
image_cond_in = torch.cat([torch.stack([image_cond[i] for _ in range(n)]) for i, n in enumerate(repeats)] + [image_uncond] + [torch.zeros_like(self.init_latent)])
2023-02-24 13:04:23 +08:00
denoiser_params = CFGDenoiserParams(x_in, image_cond_in, sigma_in, state.sampling_step, state.sampling_steps, tensor, uncond)
cfg_denoiser_callback(denoiser_params)
x_in = denoiser_params.x
image_cond_in = denoiser_params.image_cond
sigma_in = denoiser_params.sigma
tensor = denoiser_params.text_cond
uncond = denoiser_params.text_uncond
skip_uncond = False
2022-10-31 07:48:33 +08:00
# alternating uncond allows for higher thresholds without the quality loss normally expected from raising it
if self.step % 2 and s_min_uncond > 0 and sigma[0] < s_min_uncond and not is_edit_model:
skip_uncond = True
x_in = x_in[:-batch_size]
sigma_in = sigma_in[:-batch_size]
2023-03-29 06:18:28 +08:00
self.padded_cond_uncond = False
if shared.opts.pad_cond_uncond and tensor.shape[1] != uncond.shape[1]:
empty = shared.sd_model.cond_stage_model_empty_prompt
num_repeats = (tensor.shape[1] - uncond.shape[1]) // empty.shape[1]
if num_repeats < 0:
2023-07-12 02:16:43 +08:00
tensor = pad_cond(tensor, -num_repeats, empty)
self.padded_cond_uncond = True
elif num_repeats > 0:
2023-07-12 02:16:43 +08:00
uncond = pad_cond(uncond, num_repeats, empty)
self.padded_cond_uncond = True
if tensor.shape[1] == uncond.shape[1] or skip_uncond:
if is_edit_model:
2023-07-12 02:16:43 +08:00
cond_in = catenate_conds([tensor, uncond, uncond])
elif skip_uncond:
cond_in = tensor
else:
2023-07-12 02:16:43 +08:00
cond_in = catenate_conds([tensor, uncond])
if shared.batch_cond_uncond:
2023-07-12 02:16:43 +08:00
x_out = self.inner_model(x_in, sigma_in, cond=make_condition_dict(cond_in, image_cond_in))
else:
x_out = torch.zeros_like(x_in)
for batch_offset in range(0, x_out.shape[0], batch_size):
a = batch_offset
b = a + batch_size
2023-07-13 04:52:43 +08:00
x_out[a:b] = self.inner_model(x_in[a:b], sigma_in[a:b], cond=make_condition_dict(subscript_cond(cond_in, a, b), image_cond_in[a:b]))
else:
x_out = torch.zeros_like(x_in)
batch_size = batch_size*2 if shared.batch_cond_uncond else batch_size
for batch_offset in range(0, tensor.shape[0], batch_size):
a = batch_offset
b = min(a + batch_size, tensor.shape[0])
if not is_edit_model:
2023-07-12 02:16:43 +08:00
c_crossattn = subscript_cond(tensor, a, b)
else:
c_crossattn = torch.cat([tensor[a:b]], uncond)
x_out[a:b] = self.inner_model(x_in[a:b], sigma_in[a:b], cond=make_condition_dict(c_crossattn, image_cond_in[a:b]))
if not skip_uncond:
2023-07-12 02:16:43 +08:00
x_out[-uncond.shape[0]:] = self.inner_model(x_in[-uncond.shape[0]:], sigma_in[-uncond.shape[0]:], cond=make_condition_dict(uncond, image_cond_in[-uncond.shape[0]:]))
denoised_image_indexes = [x[0][0] for x in conds_list]
if skip_uncond:
fake_uncond = torch.cat([x_out[i:i+1] for i in denoised_image_indexes])
x_out = torch.cat([x_out, fake_uncond]) # we skipped uncond denoising, so we put cond-denoised image to where the uncond-denoised image should be
denoised_params = CFGDenoisedParams(x_out, state.sampling_step, state.sampling_steps, self.inner_model)
2023-02-11 10:18:38 +08:00
cfg_denoised_callback(denoised_params)
devices.test_for_nans(x_out, "unet")
if is_edit_model:
denoised = self.combine_denoised_for_edit_model(x_out, cond_scale)
elif skip_uncond:
denoised = self.combine_denoised(x_out, conds_list, uncond, 1.0)
else:
denoised = self.combine_denoised(x_out, conds_list, uncond, cond_scale)
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)
if opts.live_preview_content == "Prompt":
preview = self.sampler.last_latent
elif opts.live_preview_content == "Negative prompt":
preview = self.get_pred_x0(x_in[-uncond.shape[0]:], x_out[-uncond.shape[0]:], sigma)
else:
preview = self.get_pred_x0(torch.cat([x_in[i:i+1] for i in denoised_image_indexes]), torch.cat([denoised[i:i+1] for i in denoised_image_indexes]), sigma)
sd_samplers_common.store_latent(preview)
after_cfg_callback_params = AfterCFGCallbackParams(denoised, state.sampling_step, state.sampling_steps)
cfg_after_cfg_callback(after_cfg_callback_params)
denoised = after_cfg_callback_params.x
2022-09-15 18:10:16 +08:00
self.step += 1
return denoised