stable-diffusion-webui/scripts/img2imgalt.py

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

219 lines
9.0 KiB
Python
Raw Normal View History

2022-09-15 00:41:55 +08:00
from collections import namedtuple
2022-09-12 06:55:34 +08:00
import numpy as np
from tqdm import trange
import modules.scripts as scripts
import gradio as gr
2023-03-21 06:43:01 +08:00
from modules import processing, shared, sd_samplers, sd_samplers_common
2022-09-12 06:55:34 +08:00
import torch
import k_diffusion as K
def find_noise_for_image(p, cond, uncond, cfg_scale, steps):
x = p.init_latent
s_in = x.new_ones([x.shape[0]])
2023-03-21 06:42:36 +08:00
if shared.sd_model.parameterization == "v":
dnw = K.external.CompVisVDenoiser(shared.sd_model)
skip = 1
else:
dnw = K.external.CompVisDenoiser(shared.sd_model)
skip = 0
2022-09-12 06:55:34 +08:00
sigmas = dnw.get_sigmas(steps).flip(0)
shared.state.sampling_steps = steps
for i in trange(1, len(sigmas)):
shared.state.sampling_step += 1
x_in = torch.cat([x] * 2)
sigma_in = torch.cat([sigmas[i] * s_in] * 2)
cond_in = torch.cat([uncond, cond])
image_conditioning = torch.cat([p.image_conditioning] * 2)
cond_in = {"c_concat": [image_conditioning], "c_crossattn": [cond_in]}
2023-03-21 06:42:36 +08:00
c_out, c_in = [K.utils.append_dims(k, x_in.ndim) for k in dnw.get_scalings(sigma_in)[skip:]]
2022-09-12 06:55:34 +08:00
t = dnw.sigma_to_t(sigma_in)
eps = shared.sd_model.apply_model(x_in * c_in, t, cond=cond_in)
denoised_uncond, denoised_cond = (x_in + eps * c_out).chunk(2)
denoised = denoised_uncond + (denoised_cond - denoised_uncond) * cfg_scale
d = (x - denoised) / sigmas[i]
dt = sigmas[i] - sigmas[i - 1]
x = x + d * dt
sd_samplers_common.store_latent(x)
2022-09-12 06:55:34 +08:00
# This shouldn't be necessary, but solved some VRAM issues
del x_in, sigma_in, cond_in, c_out, c_in, t,
del eps, denoised_uncond, denoised_cond, denoised, d, dt
shared.state.nextjob()
return x / x.std()
2022-09-15 00:41:55 +08:00
Cached = namedtuple("Cached", ["noise", "cfg_scale", "steps", "latent", "original_prompt", "original_negative_prompt", "sigma_adjustment"])
# Based on changes suggested by briansemrau in https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/736
def find_noise_for_image_sigma_adjustment(p, cond, uncond, cfg_scale, steps):
x = p.init_latent
s_in = x.new_ones([x.shape[0]])
2023-03-21 06:42:36 +08:00
if shared.sd_model.parameterization == "v":
dnw = K.external.CompVisVDenoiser(shared.sd_model)
skip = 1
else:
dnw = K.external.CompVisDenoiser(shared.sd_model)
skip = 0
sigmas = dnw.get_sigmas(steps).flip(0)
shared.state.sampling_steps = steps
for i in trange(1, len(sigmas)):
shared.state.sampling_step += 1
x_in = torch.cat([x] * 2)
sigma_in = torch.cat([sigmas[i - 1] * s_in] * 2)
cond_in = torch.cat([uncond, cond])
image_conditioning = torch.cat([p.image_conditioning] * 2)
cond_in = {"c_concat": [image_conditioning], "c_crossattn": [cond_in]}
2023-03-21 06:42:36 +08:00
c_out, c_in = [K.utils.append_dims(k, x_in.ndim) for k in dnw.get_scalings(sigma_in)[skip:]]
if i == 1:
t = dnw.sigma_to_t(torch.cat([sigmas[i] * s_in] * 2))
else:
t = dnw.sigma_to_t(sigma_in)
eps = shared.sd_model.apply_model(x_in * c_in, t, cond=cond_in)
denoised_uncond, denoised_cond = (x_in + eps * c_out).chunk(2)
denoised = denoised_uncond + (denoised_cond - denoised_uncond) * cfg_scale
if i == 1:
d = (x - denoised) / (2 * sigmas[i])
else:
d = (x - denoised) / sigmas[i - 1]
dt = sigmas[i] - sigmas[i - 1]
x = x + d * dt
sd_samplers_common.store_latent(x)
# This shouldn't be necessary, but solved some VRAM issues
del x_in, sigma_in, cond_in, c_out, c_in, t,
del eps, denoised_uncond, denoised_cond, denoised, d, dt
shared.state.nextjob()
return x / sigmas[-1]
2022-09-15 00:41:55 +08:00
2022-09-12 06:55:34 +08:00
class Script(scripts.Script):
2022-09-15 00:41:55 +08:00
def __init__(self):
self.cache = None
2022-09-12 06:55:34 +08:00
def title(self):
return "img2img alternative test"
def show(self, is_img2img):
return is_img2img
2023-03-21 06:43:01 +08:00
def ui(self, is_img2img):
info = gr.Markdown('''
* `CFG Scale` should be 2 or lower.
''')
override_sampler = gr.Checkbox(label="Override `Sampling method` to Euler?(this method is built for it)", value=True, elem_id=self.elem_id("override_sampler"))
override_prompt = gr.Checkbox(label="Override `prompt` to the same value as `original prompt`?(and `negative prompt`)", value=True, elem_id=self.elem_id("override_prompt"))
original_prompt = gr.Textbox(label="Original prompt", lines=1, elem_id=self.elem_id("original_prompt"))
original_negative_prompt = gr.Textbox(label="Original negative prompt", lines=1, elem_id=self.elem_id("original_negative_prompt"))
override_steps = gr.Checkbox(label="Override `Sampling Steps` to the same value as `Decode steps`?", value=True, elem_id=self.elem_id("override_steps"))
st = gr.Slider(label="Decode steps", minimum=1, maximum=150, step=1, value=50, elem_id=self.elem_id("st"))
override_strength = gr.Checkbox(label="Override `Denoising strength` to 1?", value=True, elem_id=self.elem_id("override_strength"))
cfg = gr.Slider(label="Decode CFG scale", minimum=0.0, maximum=15.0, step=0.1, value=1.0, elem_id=self.elem_id("cfg"))
randomness = gr.Slider(label="Randomness", minimum=0.0, maximum=1.0, step=0.01, value=0.0, elem_id=self.elem_id("randomness"))
sigma_adjustment = gr.Checkbox(label="Sigma adjustment for finding noise for image", value=False, elem_id=self.elem_id("sigma_adjustment"))
2022-09-12 06:55:34 +08:00
return [
info,
override_sampler,
override_prompt, original_prompt, original_negative_prompt,
override_steps, st,
override_strength,
cfg, randomness, sigma_adjustment,
]
def run(self, p, _, override_sampler, override_prompt, original_prompt, original_negative_prompt, override_steps, st, override_strength, cfg, randomness, sigma_adjustment):
# Override
if override_sampler:
p.sampler_name = "Euler"
if override_prompt:
p.prompt = original_prompt
p.negative_prompt = original_negative_prompt
if override_steps:
p.steps = st
if override_strength:
p.denoising_strength = 1.0
2022-09-12 06:55:34 +08:00
def sample_extra(conditioning, unconditional_conditioning, seeds, subseeds, subseed_strength, prompts):
2022-09-15 00:41:55 +08:00
lat = (p.init_latent.cpu().numpy() * 10).astype(int)
same_params = self.cache is not None and self.cache.cfg_scale == cfg and self.cache.steps == st \
and self.cache.original_prompt == original_prompt \
and self.cache.original_negative_prompt == original_negative_prompt \
and self.cache.sigma_adjustment == sigma_adjustment
2022-09-15 00:41:55 +08:00
same_everything = same_params and self.cache.latent.shape == lat.shape and np.abs(self.cache.latent-lat).sum() < 100
2022-09-12 06:55:34 +08:00
2022-09-15 00:41:55 +08:00
if same_everything:
rec_noise = self.cache.noise
2022-09-12 06:55:34 +08:00
else:
shared.state.job_count += 1
cond = p.sd_model.get_learned_conditioning(p.batch_size * [original_prompt])
uncond = p.sd_model.get_learned_conditioning(p.batch_size * [original_negative_prompt])
if sigma_adjustment:
rec_noise = find_noise_for_image_sigma_adjustment(p, cond, uncond, cfg, st)
else:
rec_noise = find_noise_for_image(p, cond, uncond, cfg, st)
self.cache = Cached(rec_noise, cfg, st, lat, original_prompt, original_negative_prompt, sigma_adjustment)
2022-09-12 06:55:34 +08:00
2022-10-12 13:02:28 +08:00
rand_noise = processing.create_random_tensors(p.init_latent.shape[1:], seeds=seeds, subseeds=subseeds, subseed_strength=p.subseed_strength, seed_resize_from_h=p.seed_resize_from_h, seed_resize_from_w=p.seed_resize_from_w, p=p)
combined_noise = ((1 - randomness) * rec_noise + randomness * rand_noise) / ((randomness**2 + (1-randomness)**2) ** 0.5)
sampler = sd_samplers.create_sampler(p.sampler_name, p.sd_model)
2022-09-12 06:55:34 +08:00
sigmas = sampler.model_wrap.get_sigmas(p.steps)
noise_dt = combined_noise - (p.init_latent / sigmas[0])
p.seed = p.seed + 1
return sampler.sample_img2img(p, p.init_latent, noise_dt, conditioning, unconditional_conditioning, image_conditioning=p.image_conditioning)
2022-09-16 14:40:43 +08:00
2022-09-12 06:55:34 +08:00
p.sample = sample_extra
2022-09-21 00:07:09 +08:00
p.extra_generation_params["Decode prompt"] = original_prompt
p.extra_generation_params["Decode negative prompt"] = original_negative_prompt
p.extra_generation_params["Decode CFG scale"] = cfg
p.extra_generation_params["Decode steps"] = st
p.extra_generation_params["Randomness"] = randomness
p.extra_generation_params["Sigma Adjustment"] = sigma_adjustment
2022-09-12 06:55:34 +08:00
processed = processing.process_images(p)
return processed