Merge remote-tracking branch 'origin/master'

This commit is contained in:
AUTOMATIC 2022-10-02 15:49:59 +03:00
commit 5f561ee95d
3 changed files with 24 additions and 15 deletions

View File

@ -79,7 +79,7 @@ class StableDiffusionProcessing:
self.paste_to = None
self.color_corrections = None
self.denoising_strength: float = 0
self.sampler_noise_scheduler_override = None
self.ddim_discretize = opts.ddim_discretize
self.s_churn = opts.s_churn
self.s_tmin = opts.s_tmin
@ -130,7 +130,7 @@ class Processed:
self.s_tmin = p.s_tmin
self.s_tmax = p.s_tmax
self.s_noise = p.s_noise
self.sampler_noise_scheduler_override = p.sampler_noise_scheduler_override
self.prompt = self.prompt if type(self.prompt) != list else self.prompt[0]
self.negative_prompt = self.negative_prompt if type(self.negative_prompt) != list else self.negative_prompt[0]
self.seed = int(self.seed if type(self.seed) != list else self.seed[0])

View File

@ -290,7 +290,10 @@ class KDiffusionSampler:
def sample_img2img(self, p, x, noise, conditioning, unconditional_conditioning, steps=None):
steps, t_enc = setup_img2img_steps(p, steps)
sigmas = self.model_wrap.get_sigmas(steps)
if p.sampler_noise_scheduler_override:
sigmas = p.sampler_noise_scheduler_override(steps)
else:
sigmas = self.model_wrap.get_sigmas(steps)
noise = noise * sigmas[steps - t_enc - 1]
xi = x + noise
@ -306,7 +309,10 @@ class KDiffusionSampler:
def sample(self, p, x, conditioning, unconditional_conditioning, steps=None):
steps = steps or p.steps
sigmas = self.model_wrap.get_sigmas(steps)
if p.sampler_noise_scheduler_override:
sigmas = p.sampler_noise_scheduler_override(steps)
else:
sigmas = self.model_wrap.get_sigmas(steps)
x = x * sigmas[0]
extra_params_kwargs = self.initialize(p)

View File

@ -5,6 +5,7 @@ import numpy as np
import torch
from PIL import Image
from basicsr.utils.download_util import load_file_from_url
from tqdm import tqdm
from modules import modelloader
from modules.paths import models_path
@ -122,18 +123,20 @@ def inference(img, model, tile, tile_overlap, window_size, scale):
E = torch.zeros(b, c, h * sf, w * sf, dtype=torch.half, device=device).type_as(img)
W = torch.zeros_like(E, dtype=torch.half, device=device)
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)
with tqdm(total=len(h_idx_list) * len(w_idx_list), desc="SwinIR 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)
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