Commit d2c7ad2f authored by JashoBell's avatar JashoBell

Merge branch 'master' of https://github.com/AUTOMATIC1111/stable-diffusion-webui into Base

parents 5a797a56 23a0ec04
...@@ -16,3 +16,4 @@ __pycache__ ...@@ -16,3 +16,4 @@ __pycache__
/webui-user.bat /webui-user.bat
/webui-user.sh /webui-user.sh
/interrogate /interrogate
/user.css
...@@ -51,7 +51,7 @@ Alternatively, use [Google Colab](https://colab.research.google.com/drive/1Iy-xW ...@@ -51,7 +51,7 @@ Alternatively, use [Google Colab](https://colab.research.google.com/drive/1Iy-xW
1. Install [Python 3.10.6](https://www.python.org/downloads/windows/), checking "Add Python to PATH" 1. Install [Python 3.10.6](https://www.python.org/downloads/windows/), checking "Add Python to PATH"
2. Install [git](https://git-scm.com/download/win). 2. Install [git](https://git-scm.com/download/win).
3. Download the stable-diffusion-webui repository, for example by running `git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git`. 3. Download the stable-diffusion-webui repository, for example by running `git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git`.
4. Place `model.ckpt` in the base directory, alongside `webui.py`. 4. Place `model.ckpt` in the `models` directory.
5. _*(Optional)*_ Place `GFPGANv1.3.pth` in the base directory, alongside `webui.py`. 5. _*(Optional)*_ Place `GFPGANv1.3.pth` in the base directory, alongside `webui.py`.
6. Run `webui-user.bat` from Windows Explorer as normal, non-administrate, user. 6. Run `webui-user.bat` from Windows Explorer as normal, non-administrate, user.
...@@ -81,6 +81,7 @@ The documentation was moved from this README over to the project's [wiki](https: ...@@ -81,6 +81,7 @@ The documentation was moved from this README over to the project's [wiki](https:
- Ideas for optimizations - https://github.com/basujindal/stable-diffusion - Ideas for optimizations - https://github.com/basujindal/stable-diffusion
- Doggettx - Cross Attention layer optimization - https://github.com/Doggettx/stable-diffusion, original idea for prompt editing. - Doggettx - Cross Attention layer optimization - https://github.com/Doggettx/stable-diffusion, original idea for prompt editing.
- Idea for SD upscale - https://github.com/jquesnelle/txt2imghd - Idea for SD upscale - https://github.com/jquesnelle/txt2imghd
- Noise generation for outpainting mk2 - https://github.com/parlance-zz/g-diffuser-bot
- CLIP interrogator idea and borrowing some code - https://github.com/pharmapsychotic/clip-interrogator - CLIP interrogator idea and borrowing some code - https://github.com/pharmapsychotic/clip-interrogator
- Initial Gradio script - posted on 4chan by an Anonymous user. Thank you Anonymous user. - Initial Gradio script - posted on 4chan by an Anonymous user. Thank you Anonymous user.
- (You) - (You)
...@@ -48,3 +48,13 @@ def randn(seed, shape): ...@@ -48,3 +48,13 @@ def randn(seed, shape):
torch.manual_seed(seed) torch.manual_seed(seed)
return torch.randn(shape, device=device) return torch.randn(shape, device=device)
def randn_without_seed(shape):
# Pytorch currently doesn't handle setting randomness correctly when the metal backend is used.
if device.type == 'mps':
generator = torch.Generator(device=cpu)
noise = torch.randn(shape, generator=generator, device=cpu).to(device)
return noise
return torch.randn(shape, device=device)
...@@ -36,6 +36,7 @@ def run_extras(image, image_folder, gfpgan_visibility, codeformer_visibility, co ...@@ -36,6 +36,7 @@ def run_extras(image, image_folder, gfpgan_visibility, codeformer_visibility, co
outpath = opts.outdir_samples or opts.outdir_extras_samples outpath = opts.outdir_samples or opts.outdir_extras_samples
outputs = []
for image in imageArr: for image in imageArr:
existing_pnginfo = image.info or {} existing_pnginfo = image.info or {}
...@@ -91,7 +92,9 @@ def run_extras(image, image_folder, gfpgan_visibility, codeformer_visibility, co ...@@ -91,7 +92,9 @@ def run_extras(image, image_folder, gfpgan_visibility, codeformer_visibility, co
images.save_image(image, path=outpath, basename="", seed=None, prompt=None, extension=opts.samples_format, info=info, short_filename=True, no_prompt=True, grid=False, pnginfo_section_name="extras", existing_info=existing_pnginfo) images.save_image(image, path=outpath, basename="", seed=None, prompt=None, extension=opts.samples_format, info=info, short_filename=True, no_prompt=True, grid=False, pnginfo_section_name="extras", existing_info=existing_pnginfo)
return imageArr, plaintext_to_html(info), '' outputs.append(image)
return outputs, plaintext_to_html(info), ''
def run_pnginfo(image): def run_pnginfo(image):
...@@ -108,8 +111,9 @@ def run_pnginfo(image): ...@@ -108,8 +111,9 @@ def run_pnginfo(image):
items['exif comment'] = exif_comment items['exif comment'] = exif_comment
for field in ['jfif', 'jfif_version', 'jfif_unit', 'jfif_density', 'dpi', 'exif']: for field in ['jfif', 'jfif_version', 'jfif_unit', 'jfif_density', 'dpi', 'exif',
del items[field] 'loop', 'background', 'timestamp', 'duration']:
items.pop(field, None)
info = '' info = ''
......
...@@ -274,7 +274,7 @@ def apply_filename_pattern(x, p, seed, prompt): ...@@ -274,7 +274,7 @@ def apply_filename_pattern(x, p, seed, prompt):
x = x.replace("[height]", str(p.height)) x = x.replace("[height]", str(p.height))
x = x.replace("[sampler]", sd_samplers.samplers[p.sampler_index].name) x = x.replace("[sampler]", sd_samplers.samplers[p.sampler_index].name)
x = x.replace("[model_hash]", shared.sd_model_hash) x = x.replace("[model_hash]", shared.sd_model.sd_model_hash)
x = x.replace("[date]", datetime.date.today().isoformat()) x = x.replace("[date]", datetime.date.today().isoformat())
if cmd_opts.hide_ui_dir_config: if cmd_opts.hide_ui_dir_config:
...@@ -353,13 +353,12 @@ def save_image(image, path, basename, seed=None, prompt=None, extension='png', i ...@@ -353,13 +353,12 @@ def save_image(image, path, basename, seed=None, prompt=None, extension='png', i
}) })
if extension.lower() in ("jpg", "jpeg", "webp"): if extension.lower() in ("jpg", "jpeg", "webp"):
image.save(fullfn, quality=opts.jpeg_quality, exif_bytes=exif_bytes()) image.save(fullfn, quality=opts.jpeg_quality)
if opts.enable_pnginfo and info is not None:
piexif.insert(exif_bytes(), fullfn)
else: else:
image.save(fullfn, quality=opts.jpeg_quality, pnginfo=pnginfo) image.save(fullfn, quality=opts.jpeg_quality, pnginfo=pnginfo)
if extension.lower() == "webp":
piexif.insert(exif_bytes, fullfn)
target_side_length = 4000 target_side_length = 4000
oversize = image.width > target_side_length or image.height > target_side_length oversize = image.width > target_side_length or image.height > target_side_length
if opts.export_for_4chan and (oversize or os.stat(fullfn).st_size > 4 * 1024 * 1024): if opts.export_for_4chan and (oversize or os.stat(fullfn).st_size > 4 * 1024 * 1024):
...@@ -370,7 +369,9 @@ def save_image(image, path, basename, seed=None, prompt=None, extension='png', i ...@@ -370,7 +369,9 @@ def save_image(image, path, basename, seed=None, prompt=None, extension='png', i
elif oversize: elif oversize:
image = image.resize((image.width * target_side_length // image.height, target_side_length), LANCZOS) image = image.resize((image.width * target_side_length // image.height, target_side_length), LANCZOS)
image.save(fullfn_without_extension + ".jpg", quality=opts.jpeg_quality, exif_bytes=exif_bytes()) image.save(fullfn_without_extension + ".jpg", quality=opts.jpeg_quality)
if opts.enable_pnginfo and info is not None:
piexif.insert(exif_bytes(), fullfn_without_extension + ".jpg")
if opts.save_txt and info is not None: if opts.save_txt and info is not None:
with open(f"{fullfn_without_extension}.txt", "w", encoding="utf8") as file: with open(f"{fullfn_without_extension}.txt", "w", encoding="utf8") as file:
......
import threading
import time
from collections import defaultdict
import torch
class MemUsageMonitor(threading.Thread):
run_flag = None
device = None
disabled = False
opts = None
data = None
def __init__(self, name, device, opts):
threading.Thread.__init__(self)
self.name = name
self.device = device
self.opts = opts
self.daemon = True
self.run_flag = threading.Event()
self.data = defaultdict(int)
def run(self):
if self.disabled:
return
while True:
self.run_flag.wait()
torch.cuda.reset_peak_memory_stats()
self.data.clear()
if self.opts.memmon_poll_rate <= 0:
self.run_flag.clear()
continue
self.data["min_free"] = torch.cuda.mem_get_info()[0]
while self.run_flag.is_set():
free, total = torch.cuda.mem_get_info() # calling with self.device errors, torch bug?
self.data["min_free"] = min(self.data["min_free"], free)
time.sleep(1 / self.opts.memmon_poll_rate)
def dump_debug(self):
print(self, 'recorded data:')
for k, v in self.read().items():
print(k, -(v // -(1024 ** 2)))
print(self, 'raw torch memory stats:')
tm = torch.cuda.memory_stats(self.device)
for k, v in tm.items():
if 'bytes' not in k:
continue
print('\t' if 'peak' in k else '', k, -(v // -(1024 ** 2)))
print(torch.cuda.memory_summary())
def monitor(self):
self.run_flag.set()
def read(self):
free, total = torch.cuda.mem_get_info()
self.data["total"] = total
torch_stats = torch.cuda.memory_stats(self.device)
self.data["active_peak"] = torch_stats["active_bytes.all.peak"]
self.data["reserved_peak"] = torch_stats["reserved_bytes.all.peak"]
self.data["system_peak"] = total - self.data["min_free"]
return self.data
def stop(self):
self.run_flag.clear()
return self.read()
...@@ -119,8 +119,18 @@ def slerp(val, low, high): ...@@ -119,8 +119,18 @@ def slerp(val, low, high):
return res return res
def create_random_tensors(shape, seeds, subseeds=None, subseed_strength=0.0, seed_resize_from_h=0, seed_resize_from_w=0): def create_random_tensors(shape, seeds, subseeds=None, subseed_strength=0.0, seed_resize_from_h=0, seed_resize_from_w=0, p=None):
xs = [] xs = []
# if we have multiple seeds, this means we are working with batch size>1; this then
# enables the generation of additional tensors with noise that the sampler will use during its processing.
# Using those pre-genrated tensors instead of siimple torch.randn allows a batch with seeds [100, 101] to
# produce the same images as with two batches [100], [101].
if p is not None and p.sampler is not None and len(seeds) > 1 and opts.enable_batch_seeds:
sampler_noises = [[] for _ in range(p.sampler.number_of_needed_noises(p))]
else:
sampler_noises = None
for i, seed in enumerate(seeds): for i, seed in enumerate(seeds):
noise_shape = shape if seed_resize_from_h <= 0 or seed_resize_from_w <= 0 else (shape[0], seed_resize_from_h//8, seed_resize_from_w//8) noise_shape = shape if seed_resize_from_h <= 0 or seed_resize_from_w <= 0 else (shape[0], seed_resize_from_h//8, seed_resize_from_w//8)
...@@ -155,9 +165,17 @@ def create_random_tensors(shape, seeds, subseeds=None, subseed_strength=0.0, see ...@@ -155,9 +165,17 @@ def create_random_tensors(shape, seeds, subseeds=None, subseed_strength=0.0, see
x[:, ty:ty+h, tx:tx+w] = noise[:, dy:dy+h, dx:dx+w] x[:, ty:ty+h, tx:tx+w] = noise[:, dy:dy+h, dx:dx+w]
noise = x noise = x
if sampler_noises is not None:
cnt = p.sampler.number_of_needed_noises(p)
for j in range(cnt):
sampler_noises[j].append(devices.randn_without_seed(tuple(noise_shape)))
xs.append(noise) xs.append(noise)
if sampler_noises is not None:
p.sampler.sampler_noises = [torch.stack(n).to(shared.device) for n in sampler_noises]
x = torch.stack(xs).to(shared.device) x = torch.stack(xs).to(shared.device)
return x return x
...@@ -170,7 +188,11 @@ def fix_seed(p): ...@@ -170,7 +188,11 @@ def fix_seed(p):
def process_images(p: StableDiffusionProcessing) -> Processed: def process_images(p: StableDiffusionProcessing) -> Processed:
"""this is the main loop that both txt2img and img2img use; it calls func_init once inside all the scopes and func_sample once per batch""" """this is the main loop that both txt2img and img2img use; it calls func_init once inside all the scopes and func_sample once per batch"""
assert p.prompt is not None if type(p.prompt) == list:
assert(len(p.prompt) > 0)
else:
assert p.prompt is not None
devices.torch_gc() devices.torch_gc()
fix_seed(p) fix_seed(p)
...@@ -209,7 +231,7 @@ def process_images(p: StableDiffusionProcessing) -> Processed: ...@@ -209,7 +231,7 @@ def process_images(p: StableDiffusionProcessing) -> Processed:
"Seed": all_seeds[index], "Seed": all_seeds[index],
"Face restoration": (opts.face_restoration_model if p.restore_faces else None), "Face restoration": (opts.face_restoration_model if p.restore_faces else None),
"Size": f"{p.width}x{p.height}", "Size": f"{p.width}x{p.height}",
"Model hash": (None if not opts.add_model_hash_to_info or not shared.sd_model_hash else shared.sd_model_hash), "Model hash": (None if not opts.add_model_hash_to_info or not shared.sd_model.sd_model_hash else shared.sd_model.sd_model_hash),
"Batch size": (None if p.batch_size < 2 else p.batch_size), "Batch size": (None if p.batch_size < 2 else p.batch_size),
"Batch pos": (None if p.batch_size < 2 else position_in_batch), "Batch pos": (None if p.batch_size < 2 else position_in_batch),
"Variation seed": (None if p.subseed_strength == 0 else all_subseeds[index]), "Variation seed": (None if p.subseed_strength == 0 else all_subseeds[index]),
...@@ -247,6 +269,9 @@ def process_images(p: StableDiffusionProcessing) -> Processed: ...@@ -247,6 +269,9 @@ def process_images(p: StableDiffusionProcessing) -> Processed:
seeds = all_seeds[n * p.batch_size:(n + 1) * p.batch_size] seeds = all_seeds[n * p.batch_size:(n + 1) * p.batch_size]
subseeds = all_subseeds[n * p.batch_size:(n + 1) * p.batch_size] subseeds = all_subseeds[n * p.batch_size:(n + 1) * p.batch_size]
if (len(prompts) == 0):
break
#uc = p.sd_model.get_learned_conditioning(len(prompts) * [p.negative_prompt]) #uc = p.sd_model.get_learned_conditioning(len(prompts) * [p.negative_prompt])
#c = p.sd_model.get_learned_conditioning(prompts) #c = p.sd_model.get_learned_conditioning(prompts)
uc = prompt_parser.get_learned_conditioning(len(prompts) * [p.negative_prompt], p.steps) uc = prompt_parser.get_learned_conditioning(len(prompts) * [p.negative_prompt], p.steps)
...@@ -257,7 +282,7 @@ def process_images(p: StableDiffusionProcessing) -> Processed: ...@@ -257,7 +282,7 @@ def process_images(p: StableDiffusionProcessing) -> Processed:
comments[comment] = 1 comments[comment] = 1
# we manually generate all input noises because each one should have a specific seed # we manually generate all input noises because each one should have a specific seed
x = create_random_tensors([opt_C, p.height // opt_f, p.width // opt_f], 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) x = create_random_tensors([opt_C, p.height // opt_f, p.width // opt_f], 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)
if p.n_iter > 1: if p.n_iter > 1:
shared.state.job = f"Batch {n+1} out of {p.n_iter}" shared.state.job = f"Batch {n+1} out of {p.n_iter}"
......
import glob
import os.path
import sys
from collections import namedtuple
import torch
from omegaconf import OmegaConf
from ldm.util import instantiate_from_config
from modules import shared
CheckpointInfo = namedtuple("CheckpointInfo", ['filename', 'title', 'hash'])
checkpoints_list = {}
try:
# this silences the annoying "Some weights of the model checkpoint were not used when initializing..." message at start.
from transformers import logging
logging.set_verbosity_error()
except Exception:
pass
def list_models():
checkpoints_list.clear()
model_dir = os.path.abspath(shared.cmd_opts.ckpt_dir)
def modeltitle(path, h):
abspath = os.path.abspath(path)
if abspath.startswith(model_dir):
name = abspath.replace(model_dir, '')
else:
name = os.path.basename(path)
if name.startswith("\\") or name.startswith("/"):
name = name[1:]
return f'{name} [{h}]'
cmd_ckpt = shared.cmd_opts.ckpt
if os.path.exists(cmd_ckpt):
h = model_hash(cmd_ckpt)
title = modeltitle(cmd_ckpt, h)
checkpoints_list[title] = CheckpointInfo(cmd_ckpt, title, h)
elif cmd_ckpt is not None and cmd_ckpt != shared.default_sd_model_file:
print(f"Checkpoint in --ckpt argument not found: {cmd_ckpt}", file=sys.stderr)
if os.path.exists(model_dir):
for filename in glob.glob(model_dir + '/**/*.ckpt', recursive=True):
h = model_hash(filename)
title = modeltitle(filename, h)
checkpoints_list[title] = CheckpointInfo(filename, title, h)
def model_hash(filename):
try:
with open(filename, "rb") as file:
import hashlib
m = hashlib.sha256()
file.seek(0x100000)
m.update(file.read(0x10000))
return m.hexdigest()[0:8]
except FileNotFoundError:
return 'NOFILE'
def select_checkpoint():
model_checkpoint = shared.opts.sd_model_checkpoint
checkpoint_info = checkpoints_list.get(model_checkpoint, None)
if checkpoint_info is not None:
return checkpoint_info
if len(checkpoints_list) == 0:
print(f"Checkpoint {model_checkpoint} not found and no other checkpoints found", file=sys.stderr)
return None
checkpoint_info = next(iter(checkpoints_list.values()))
if model_checkpoint is not None:
print(f"Checkpoint {model_checkpoint} not found; loading fallback {checkpoint_info.title}", file=sys.stderr)
return checkpoint_info
def load_model_weights(model, checkpoint_file, sd_model_hash):
print(f"Loading weights [{sd_model_hash}] from {checkpoint_file}")
pl_sd = torch.load(checkpoint_file, map_location="cpu")
if "global_step" in pl_sd:
print(f"Global Step: {pl_sd['global_step']}")
sd = pl_sd["state_dict"]
model.load_state_dict(sd, strict=False)
if shared.cmd_opts.opt_channelslast:
model.to(memory_format=torch.channels_last)
if not shared.cmd_opts.no_half:
model.half()
model.sd_model_hash = sd_model_hash
model.sd_model_checkpint = checkpoint_file
def load_model():
from modules import lowvram, sd_hijack
checkpoint_info = select_checkpoint()
sd_config = OmegaConf.load(shared.cmd_opts.config)
sd_model = instantiate_from_config(sd_config.model)
load_model_weights(sd_model, checkpoint_info.filename, checkpoint_info.hash)
if shared.cmd_opts.lowvram or shared.cmd_opts.medvram:
lowvram.setup_for_low_vram(sd_model, shared.cmd_opts.medvram)
else:
sd_model.to(shared.device)
sd_hijack.model_hijack.hijack(sd_model)
sd_model.eval()
print(f"Model loaded.")
return sd_model
def reload_model_weights(sd_model, info=None):
from modules import lowvram, devices
checkpoint_info = info or select_checkpoint()
if sd_model.sd_model_checkpint == checkpoint_info.filename:
return
if shared.cmd_opts.lowvram or shared.cmd_opts.medvram:
lowvram.send_everything_to_cpu()
else:
sd_model.to(devices.cpu)
load_model_weights(sd_model, checkpoint_info.filename, checkpoint_info.hash)
if not shared.cmd_opts.lowvram and not shared.cmd_opts.medvram:
sd_model.to(devices.device)
print(f"Weights loaded.")
return sd_model
...@@ -38,6 +38,17 @@ samplers = [ ...@@ -38,6 +38,17 @@ samplers = [
samplers_for_img2img = [x for x in samplers if x.name != 'PLMS'] samplers_for_img2img = [x for x in samplers if x.name != 'PLMS']
def setup_img2img_steps(p):
if opts.img2img_fix_steps:
steps = int(p.steps / min(p.denoising_strength, 0.999))
t_enc = p.steps - 1
else:
steps = p.steps
t_enc = int(min(p.denoising_strength, 0.999) * steps)
return steps, t_enc
def sample_to_image(samples): def sample_to_image(samples):
x_sample = shared.sd_model.decode_first_stage(samples[0:1].type(shared.sd_model.dtype))[0] x_sample = shared.sd_model.decode_first_stage(samples[0:1].type(shared.sd_model.dtype))[0]
x_sample = torch.clamp((x_sample + 1.0) / 2.0, min=0.0, max=1.0) x_sample = torch.clamp((x_sample + 1.0) / 2.0, min=0.0, max=1.0)
...@@ -80,8 +91,12 @@ class VanillaStableDiffusionSampler: ...@@ -80,8 +91,12 @@ class VanillaStableDiffusionSampler:
self.mask = None self.mask = None
self.nmask = None self.nmask = None
self.init_latent = None self.init_latent = None
self.sampler_noises = None
self.step = 0 self.step = 0
def number_of_needed_noises(self, p):
return 0
def p_sample_ddim_hook(self, x_dec, cond, ts, unconditional_conditioning, *args, **kwargs): def p_sample_ddim_hook(self, x_dec, cond, ts, unconditional_conditioning, *args, **kwargs):
cond = prompt_parser.reconstruct_cond_batch(cond, self.step) cond = prompt_parser.reconstruct_cond_batch(cond, self.step)
unconditional_conditioning = prompt_parser.reconstruct_cond_batch(unconditional_conditioning, self.step) unconditional_conditioning = prompt_parser.reconstruct_cond_batch(unconditional_conditioning, self.step)
...@@ -101,13 +116,13 @@ class VanillaStableDiffusionSampler: ...@@ -101,13 +116,13 @@ class VanillaStableDiffusionSampler:
return res return res
def sample_img2img(self, p, x, noise, conditioning, unconditional_conditioning): def sample_img2img(self, p, x, noise, conditioning, unconditional_conditioning):
t_enc = int(min(p.denoising_strength, 0.999) * p.steps) steps, t_enc = setup_img2img_steps(p)
# existing code fails with cetain step counts, like 9 # existing code fails with cetain step counts, like 9
try: try:
self.sampler.make_schedule(ddim_num_steps=p.steps, verbose=False) self.sampler.make_schedule(ddim_num_steps=steps, verbose=False)
except Exception: except Exception:
self.sampler.make_schedule(ddim_num_steps=p.steps+1, verbose=False) self.sampler.make_schedule(ddim_num_steps=steps+1, verbose=False)
x1 = self.sampler.stochastic_encode(x, torch.tensor([t_enc] * int(x.shape[0])).to(shared.device), noise=noise) x1 = self.sampler.stochastic_encode(x, torch.tensor([t_enc] * int(x.shape[0])).to(shared.device), noise=noise)
...@@ -115,6 +130,7 @@ class VanillaStableDiffusionSampler: ...@@ -115,6 +130,7 @@ class VanillaStableDiffusionSampler:
self.mask = p.mask self.mask = p.mask
self.nmask = p.nmask self.nmask = p.nmask
self.init_latent = p.init_latent self.init_latent = p.init_latent
self.step = 0
samples = self.sampler.decode(x1, conditioning, t_enc, unconditional_guidance_scale=p.cfg_scale, unconditional_conditioning=unconditional_conditioning) samples = self.sampler.decode(x1, conditioning, t_enc, unconditional_guidance_scale=p.cfg_scale, unconditional_conditioning=unconditional_conditioning)
...@@ -127,6 +143,7 @@ class VanillaStableDiffusionSampler: ...@@ -127,6 +143,7 @@ class VanillaStableDiffusionSampler:
self.mask = None self.mask = None
self.nmask = None self.nmask = None
self.init_latent = None self.init_latent = None
self.step = 0
# existing code fails with cetin step counts, like 9 # existing code fails with cetin step counts, like 9
try: try:
...@@ -183,42 +200,82 @@ def extended_trange(count, *args, **kwargs): ...@@ -183,42 +200,82 @@ def extended_trange(count, *args, **kwargs):
shared.total_tqdm.update() shared.total_tqdm.update()
class TorchHijack:
def __init__(self, kdiff_sampler):
self.kdiff_sampler = kdiff_sampler
def __getattr__(self, item):
if item == 'randn_like':
return self.kdiff_sampler.randn_like
if hasattr(torch, item):
return getattr(torch, item)
raise AttributeError("'{}' object has no attribute '{}'".format(type(self).__name__, item))
class KDiffusionSampler: class KDiffusionSampler:
def __init__(self, funcname, sd_model): def __init__(self, funcname, sd_model):
self.model_wrap = k_diffusion.external.CompVisDenoiser(sd_model, quantize=shared.opts.enable_quantization) self.model_wrap = k_diffusion.external.CompVisDenoiser(sd_model, quantize=shared.opts.enable_quantization)
self.funcname = funcname self.funcname = funcname
self.func = getattr(k_diffusion.sampling, self.funcname) self.func = getattr(k_diffusion.sampling, self.funcname)
self.model_wrap_cfg = CFGDenoiser(self.model_wrap) self.model_wrap_cfg = CFGDenoiser(self.model_wrap)
self.sampler_noises = None
self.sampler_noise_index = 0
def callback_state(self, d): def callback_state(self, d):
store_latent(d["denoised"]) store_latent(d["denoised"])
def number_of_needed_noises(self, p):
return p.steps
def randn_like(self, x):
noise = self.sampler_noises[self.sampler_noise_index] if self.sampler_noises is not None and self.sampler_noise_index < len(self.sampler_noises) else None
if noise is not None and x.shape == noise.shape:
res = noise
else:
res = torch.randn_like(x)
self.sampler_noise_index += 1
return res
def sample_img2img(self, p, x, noise, conditioning, unconditional_conditioning): def sample_img2img(self, p, x, noise, conditioning, unconditional_conditioning):
t_enc = int(min(p.denoising_strength, 0.999) * p.steps) steps, t_enc = setup_img2img_steps(p)
sigmas = self.model_wrap.get_sigmas(p.steps)
sigmas = self.model_wrap.get_sigmas(steps)
noise = noise * sigmas[p.steps - t_enc - 1] noise = noise * sigmas[steps - t_enc - 1]
xi = x + noise xi = x + noise
sigma_sched = sigmas[p.steps - t_enc - 1:] sigma_sched = sigmas[steps - t_enc - 1:]
self.model_wrap_cfg.mask = p.mask self.model_wrap_cfg.mask = p.mask
self.model_wrap_cfg.nmask = p.nmask self.model_wrap_cfg.nmask = p.nmask
self.model_wrap_cfg.init_latent = p.init_latent self.model_wrap_cfg.init_latent = p.init_latent
self.model_wrap.step = 0
if hasattr(k_diffusion.sampling, 'trange'): if hasattr(k_diffusion.sampling, 'trange'):
k_diffusion.sampling.trange = lambda *args, **kwargs: extended_trange(*args, **kwargs) k_diffusion.sampling.trange = lambda *args, **kwargs: extended_trange(*args, **kwargs)
if self.sampler_noises is not None:
k_diffusion.sampling.torch = TorchHijack(self)
return self.func(self.model_wrap_cfg, xi, sigma_sched, extra_args={'cond': conditioning, 'uncond': unconditional_conditioning, 'cond_scale': p.cfg_scale}, disable=False, callback=self.callback_state) return self.func(self.model_wrap_cfg, xi, sigma_sched, extra_args={'cond': conditioning, 'uncond': unconditional_conditioning, 'cond_scale': p.cfg_scale}, disable=False, callback=self.callback_state)
def sample(self, p, x, conditioning, unconditional_conditioning): def sample(self, p, x, conditioning, unconditional_conditioning):
sigmas = self.model_wrap.get_sigmas(p.steps) sigmas = self.model_wrap.get_sigmas(p.steps)
x = x * sigmas[0] x = x * sigmas[0]
self.model_wrap_cfg.step = 0
if hasattr(k_diffusion.sampling, 'trange'): if hasattr(k_diffusion.sampling, 'trange'):
k_diffusion.sampling.trange = lambda *args, **kwargs: extended_trange(*args, **kwargs) k_diffusion.sampling.trange = lambda *args, **kwargs: extended_trange(*args, **kwargs)
if self.sampler_noises is not None:
k_diffusion.sampling.torch = TorchHijack(self)
samples_ddim = self.func(self.model_wrap_cfg, x, sigmas, extra_args={'cond': conditioning, 'uncond': unconditional_conditioning, 'cond_scale': p.cfg_scale}, disable=False, callback=self.callback_state) samples_ddim = self.func(self.model_wrap_cfg, x, sigmas, extra_args={'cond': conditioning, 'uncond': unconditional_conditioning, 'cond_scale': p.cfg_scale}, disable=False, callback=self.callback_state)
return samples_ddim return samples_ddim
...@@ -12,14 +12,16 @@ from modules.paths import script_path, sd_path ...@@ -12,14 +12,16 @@ from modules.paths import script_path, sd_path
from modules.devices import get_optimal_device from modules.devices import get_optimal_device
import modules.styles import modules.styles
import modules.interrogate import modules.interrogate
import modules.memmon
import modules.sd_models
sd_model_file = os.path.join(script_path, 'model.ckpt') sd_model_file = os.path.join(script_path, 'model.ckpt')
if not os.path.exists(sd_model_file): default_sd_model_file = sd_model_file
sd_model_file = "models/ldm/stable-diffusion-v1/model.ckpt"
parser = argparse.ArgumentParser() parser = argparse.ArgumentParser()
parser.add_argument("--config", type=str, default=os.path.join(sd_path, "configs/stable-diffusion/v1-inference.yaml"), help="path to config which constructs model",) parser.add_argument("--config", type=str, default=os.path.join(sd_path, "configs/stable-diffusion/v1-inference.yaml"), help="path to config which constructs model",)
parser.add_argument("--ckpt", type=str, default=os.path.join(sd_path, sd_model_file), help="path to checkpoint of model",) parser.add_argument("--ckpt", type=str, default=sd_model_file, help="path to checkpoint of stable diffusion model; this checkpoint will be added to the list of checkpoints and loaded by default if you don't have a checkpoint selected in settings",)
parser.add_argument("--ckpt-dir", type=str, default=os.path.join(script_path, 'models'), help="path to directory with stable diffusion checkpoints",)
parser.add_argument("--gfpgan-dir", type=str, help="GFPGAN directory", default=('./src/gfpgan' if os.path.exists('./src/gfpgan') else './GFPGAN')) parser.add_argument("--gfpgan-dir", type=str, help="GFPGAN directory", default=('./src/gfpgan' if os.path.exists('./src/gfpgan') else './GFPGAN'))
parser.add_argument("--gfpgan-model", type=str, help="GFPGAN model file name", default='GFPGANv1.3.pth') parser.add_argument("--gfpgan-model", type=str, help="GFPGAN model file name", default='GFPGANv1.3.pth')
parser.add_argument("--no-half", action='store_true', help="do not switch the model to 16-bit floats") parser.add_argument("--no-half", action='store_true', help="do not switch the model to 16-bit floats")
...@@ -87,13 +89,17 @@ interrogator = modules.interrogate.InterrogateModels("interrogate") ...@@ -87,13 +89,17 @@ interrogator = modules.interrogate.InterrogateModels("interrogate")
face_restorers = [] face_restorers = []
modules.sd_models.list_models()
class Options: class Options:
class OptionInfo: class OptionInfo:
def __init__(self, default=None, label="", component=None, component_args=None): def __init__(self, default=None, label="", component=None, component_args=None, onchange=None):
self.default = default self.default = default
self.label = label self.label = label
self.component = component self.component = component
self.component_args = component_args self.component_args = component_args
self.onchange = onchange
data = None data = None
hide_dirs = {"visible": False} if cmd_opts.hide_ui_dir_config else None hide_dirs = {"visible": False} if cmd_opts.hide_ui_dir_config else None
...@@ -125,9 +131,11 @@ class Options: ...@@ -125,9 +131,11 @@ class Options:
"enable_pnginfo": OptionInfo(True, "Save text information about generation parameters as chunks to png files"), "enable_pnginfo": OptionInfo(True, "Save text information about generation parameters as chunks to png files"),
"add_model_hash_to_info": OptionInfo(False, "Add model hash to generation information"), "add_model_hash_to_info": OptionInfo(False, "Add model hash to generation information"),
"img2img_color_correction": OptionInfo(False, "Apply color correction to img2img results to match original colors."), "img2img_color_correction": OptionInfo(False, "Apply color correction to img2img results to match original colors."),
"img2img_fix_steps": OptionInfo(False, "With img2img, do exactly the amount of steps the slider specifies (normaly you'd do less with less denoising)."),
"enable_quantization": OptionInfo(False, "Enable quantization in K samplers for sharper and cleaner results. This may change existing seeds. Requires restart to apply."), "enable_quantization": OptionInfo(False, "Enable quantization in K samplers for sharper and cleaner results. This may change existing seeds. Requires restart to apply."),
"font": OptionInfo("", "Font for image grids that have text"), "font": OptionInfo("", "Font for image grids that have text"),
"enable_emphasis": OptionInfo(True, "Use (text) to make model pay more attention to text and [text] to make it pay less attention"), "enable_emphasis": OptionInfo(True, "Use (text) to make model pay more attention to text and [text] to make it pay less attention"),
"enable_batch_seeds": OptionInfo(True, "Make K-diffusion samplers produce same images in a batch as when making a single image"),
"save_txt": OptionInfo(False, "Create a text file next to every image with generation parameters."), "save_txt": OptionInfo(False, "Create a text file next to every image with generation parameters."),
"ESRGAN_tile": OptionInfo(192, "Tile size for upscaling. 0 = no tiling.", gr.Slider, {"minimum": 0, "maximum": 512, "step": 16}), "ESRGAN_tile": OptionInfo(192, "Tile size for upscaling. 0 = no tiling.", gr.Slider, {"minimum": 0, "maximum": 512, "step": 16}),
"ESRGAN_tile_overlap": OptionInfo(8, "Tile overlap, in pixels for upscaling. Low values = visible seam.", gr.Slider, {"minimum": 0, "maximum": 48, "step": 1}), "ESRGAN_tile_overlap": OptionInfo(8, "Tile overlap, in pixels for upscaling. Low values = visible seam.", gr.Slider, {"minimum": 0, "maximum": 48, "step": 1}),
...@@ -136,6 +144,7 @@ class Options: ...@@ -136,6 +144,7 @@ class Options:
"show_progressbar": OptionInfo(True, "Show progressbar"), "show_progressbar": OptionInfo(True, "Show progressbar"),
"show_progress_every_n_steps": OptionInfo(0, "Show show image creation progress every N sampling steps. Set 0 to disable.", gr.Slider, {"minimum": 0, "maximum": 32, "step": 1}), "show_progress_every_n_steps": OptionInfo(0, "Show show image creation progress every N sampling steps. Set 0 to disable.", gr.Slider, {"minimum": 0, "maximum": 32, "step": 1}),
"multiple_tqdm": OptionInfo(True, "Add a second progress bar to the console that shows progress for an entire job. Broken in PyCharm console."), "multiple_tqdm": OptionInfo(True, "Add a second progress bar to the console that shows progress for an entire job. Broken in PyCharm console."),
"memmon_poll_rate": OptionInfo(8, "VRAM usage polls per second during generation. Set to 0 to disable.", gr.Slider, {"minimum": 0, "maximum": 40, "step":1}),
"face_restoration_model": OptionInfo(None, "Face restoration model", gr.Radio, lambda: {"choices": [x.name() for x in face_restorers]}), "face_restoration_model": OptionInfo(None, "Face restoration model", gr.Radio, lambda: {"choices": [x.name() for x in face_restorers]}),
"code_former_weight": OptionInfo(0.5, "CodeFormer weight parameter; 0 = maximum effect; 1 = minimum effect", gr.Slider, {"minimum": 0, "maximum": 1, "step": 0.01}), "code_former_weight": OptionInfo(0.5, "CodeFormer weight parameter; 0 = maximum effect; 1 = minimum effect", gr.Slider, {"minimum": 0, "maximum": 1, "step": 0.01}),
"save_images_before_face_restoration": OptionInfo(False, "Save a copy of image before doing face restoration."), "save_images_before_face_restoration": OptionInfo(False, "Save a copy of image before doing face restoration."),
...@@ -146,6 +155,7 @@ class Options: ...@@ -146,6 +155,7 @@ class Options:
"interrogate_clip_min_length": OptionInfo(24, "Interrogate: minimum description length (excluding artists, etc..)", gr.Slider, {"minimum": 1, "maximum": 128, "step": 1}), "interrogate_clip_min_length": OptionInfo(24, "Interrogate: minimum description length (excluding artists, etc..)", gr.Slider, {"minimum": 1, "maximum": 128, "step": 1}),
"interrogate_clip_max_length": OptionInfo(48, "Interrogate: maximum description length", gr.Slider, {"minimum": 1, "maximum": 256, "step": 1}), "interrogate_clip_max_length": OptionInfo(48, "Interrogate: maximum description length", gr.Slider, {"minimum": 1, "maximum": 256, "step": 1}),
"interrogate_clip_dict_limit": OptionInfo(1500, "Interrogate: maximum number of lines in text file (0 = No limit)"), "interrogate_clip_dict_limit": OptionInfo(1500, "Interrogate: maximum number of lines in text file (0 = No limit)"),
"sd_model_checkpoint": OptionInfo(None, "Stable Diffusion checkpoint", gr.Radio, lambda: {"choices": [x.title for x in modules.sd_models.checkpoints_list.values()]}),
} }
def __init__(self): def __init__(self):
...@@ -176,6 +186,10 @@ class Options: ...@@ -176,6 +186,10 @@ class Options:
with open(filename, "r", encoding="utf8") as file: with open(filename, "r", encoding="utf8") as file:
self.data = json.load(file) self.data = json.load(file)
def onchange(self, key, func):
item = self.data_labels.get(key)
item.onchange = func
opts = Options() opts = Options()
if os.path.exists(config_filename): if os.path.exists(config_filename):
...@@ -184,7 +198,6 @@ if os.path.exists(config_filename): ...@@ -184,7 +198,6 @@ if os.path.exists(config_filename):
sd_upscalers = [] sd_upscalers = []
sd_model = None sd_model = None
sd_model_hash = ''
progress_print_out = sys.stdout progress_print_out = sys.stdout
...@@ -215,3 +228,6 @@ class TotalTQDM: ...@@ -215,3 +228,6 @@ class TotalTQDM:
total_tqdm = TotalTQDM() total_tqdm = TotalTQDM()
mem_mon = modules.memmon.MemUsageMonitor("MemMon", device, opts)
mem_mon.start()
...@@ -119,6 +119,7 @@ def save_files(js_data, images, index): ...@@ -119,6 +119,7 @@ def save_files(js_data, images, index):
def wrap_gradio_call(func): def wrap_gradio_call(func):
def f(*args, **kwargs): def f(*args, **kwargs):
shared.mem_mon.monitor()
t = time.perf_counter() t = time.perf_counter()
try: try:
...@@ -135,8 +136,20 @@ def wrap_gradio_call(func): ...@@ -135,8 +136,20 @@ def wrap_gradio_call(func):
elapsed = time.perf_counter() - t elapsed = time.perf_counter() - t
mem_stats = {k: -(v//-(1024*1024)) for k,v in shared.mem_mon.stop().items()}
active_peak = mem_stats['active_peak']
reserved_peak = mem_stats['reserved_peak']
sys_peak = '?' if opts.memmon_poll_rate <= 0 else mem_stats['system_peak']
sys_total = mem_stats['total']
sys_pct = '?' if opts.memmon_poll_rate <= 0 else round(sys_peak/sys_total * 100, 2)
vram_tooltip = "Torch active: Peak amount of VRAM used by Torch during generation, excluding cached data.&#013;" \
"Torch reserved: Peak amount of VRAM allocated by Torch, including all active and cached data.&#013;" \
"Sys VRAM: Peak amount of VRAM allocation across all applications / total GPU VRAM (peak utilization%)."
vram_html = '' if opts.memmon_poll_rate == 0 else f"<p class='vram' title='{vram_tooltip}'>Torch active/reserved: {active_peak}/{reserved_peak} MiB, <wbr>Sys VRAM: {sys_peak}/{sys_total} MiB ({sys_pct}%)</p>"
# last item is always HTML # last item is always HTML
res[-1] = res[-1] + f"<p class='performance'>Time taken: {elapsed:.2f}s</p>" res[-1] += f"<div class='performance'><p class='time'>Time taken: <wbr>{elapsed:.2f}s</p>{vram_html}</div>"
shared.state.interrupted = False shared.state.interrupted = False
...@@ -324,6 +337,8 @@ def create_ui(txt2img, img2img, run_extras, run_pnginfo): ...@@ -324,6 +337,8 @@ def create_ui(txt2img, img2img, run_extras, run_pnginfo):
custom_inputs = modules.scripts.scripts_txt2img.setup_ui(is_img2img=False) custom_inputs = modules.scripts.scripts_txt2img.setup_ui(is_img2img=False)
with gr.Column(variant='panel'): with gr.Column(variant='panel'):
progressbar = gr.HTML(elem_id="progressbar")
with gr.Group(): with gr.Group():
txt2img_preview = gr.Image(elem_id='txt2img_preview', visible=False) txt2img_preview = gr.Image(elem_id='txt2img_preview', visible=False)
txt2img_gallery = gr.Gallery(label='Output', elem_id='txt2img_gallery').style(grid=4) txt2img_gallery = gr.Gallery(label='Output', elem_id='txt2img_gallery').style(grid=4)
...@@ -336,8 +351,6 @@ def create_ui(txt2img, img2img, run_extras, run_pnginfo): ...@@ -336,8 +351,6 @@ def create_ui(txt2img, img2img, run_extras, run_pnginfo):
send_to_extras = gr.Button('Send to extras') send_to_extras = gr.Button('Send to extras')
interrupt = gr.Button('Interrupt') interrupt = gr.Button('Interrupt')
progressbar = gr.HTML(elem_id="progressbar")
with gr.Group(): with gr.Group():
html_info = gr.HTML() html_info = gr.HTML()
generation_info = gr.Textbox(visible=False) generation_info = gr.Textbox(visible=False)
...@@ -461,6 +474,8 @@ def create_ui(txt2img, img2img, run_extras, run_pnginfo): ...@@ -461,6 +474,8 @@ def create_ui(txt2img, img2img, run_extras, run_pnginfo):
custom_inputs = modules.scripts.scripts_img2img.setup_ui(is_img2img=True) custom_inputs = modules.scripts.scripts_img2img.setup_ui(is_img2img=True)
with gr.Column(variant='panel'): with gr.Column(variant='panel'):
progressbar = gr.HTML(elem_id="progressbar")
with gr.Group(): with gr.Group():
img2img_preview = gr.Image(elem_id='img2img_preview', visible=False) img2img_preview = gr.Image(elem_id='img2img_preview', visible=False)
img2img_gallery = gr.Gallery(label='Output', elem_id='img2img_gallery').style(grid=4) img2img_gallery = gr.Gallery(label='Output', elem_id='img2img_gallery').style(grid=4)
...@@ -474,7 +489,6 @@ def create_ui(txt2img, img2img, run_extras, run_pnginfo): ...@@ -474,7 +489,6 @@ def create_ui(txt2img, img2img, run_extras, run_pnginfo):
interrupt = gr.Button('Interrupt') interrupt = gr.Button('Interrupt')
img2img_save_style = gr.Button('Save prompt as style') img2img_save_style = gr.Button('Save prompt as style')
progressbar = gr.HTML(elem_id="progressbar")
with gr.Group(): with gr.Group():
html_info = gr.HTML() html_info = gr.HTML()
...@@ -649,7 +663,7 @@ def create_ui(txt2img, img2img, run_extras, run_pnginfo): ...@@ -649,7 +663,7 @@ def create_ui(txt2img, img2img, run_extras, run_pnginfo):
image = gr.Image(label="Source", source="upload", interactive=True, type="pil") image = gr.Image(label="Source", source="upload", interactive=True, type="pil")
with gr.TabItem('Batch Process'): with gr.TabItem('Batch Process'):
image_batch = gr.File(label="Batch Process", file_count="multiple", source="upload", interactive=True, type="file") image_batch = gr.File(label="Batch Process", file_count="multiple", interactive=True, type="file")
upscaling_resize = gr.Slider(minimum=1.0, maximum=4.0, step=0.05, label="Resize", value=2) upscaling_resize = gr.Slider(minimum=1.0, maximum=4.0, step=0.05, label="Resize", value=2)
...@@ -745,7 +759,12 @@ def create_ui(txt2img, img2img, run_extras, run_pnginfo): ...@@ -745,7 +759,12 @@ def create_ui(txt2img, img2img, run_extras, run_pnginfo):
if comp_args and isinstance(comp_args, dict) and comp_args.get('visible') is False: if comp_args and isinstance(comp_args, dict) and comp_args.get('visible') is False:
continue continue
oldval = opts.data.get(key, None)
opts.data[key] = value opts.data[key] = value
if oldval != value and opts.data_labels[key].onchange is not None:
opts.data_labels[key].onchange()
up.append(comp.update(value=value)) up.append(comp.update(value=value))
opts.save(shared.config_filename) opts.save(shared.config_filename)
...@@ -782,6 +801,11 @@ def create_ui(txt2img, img2img, run_extras, run_pnginfo): ...@@ -782,6 +801,11 @@ def create_ui(txt2img, img2img, run_extras, run_pnginfo):
with open(os.path.join(script_path, "style.css"), "r", encoding="utf8") as file: with open(os.path.join(script_path, "style.css"), "r", encoding="utf8") as file:
css = file.read() css = file.read()
if os.path.exists(os.path.join(script_path, "user.css")):
with open(os.path.join(script_path, "user.css"), "r", encoding="utf8") as file:
usercss = file.read()
css += usercss
if not cmd_opts.no_progressbar_hiding: if not cmd_opts.no_progressbar_hiding:
css += css_hide_progressbar css += css_hide_progressbar
......
...@@ -66,6 +66,8 @@ titles = { ...@@ -66,6 +66,8 @@ titles = {
"Style 2": "Style to apply; styles have components for both positive and negative prompts and apply to both", "Style 2": "Style to apply; styles have components for both positive and negative prompts and apply to both",
"Apply style": "Insert selected styles into prompt fields", "Apply style": "Insert selected styles into prompt fields",
"Create style": "Save current prompts as a style. If you add the token {prompt} to the text, the style use that as placeholder for your prompt when you use the style in the future.", "Create style": "Save current prompts as a style. If you add the token {prompt} to the text, the style use that as placeholder for your prompt when you use the style in the future.",
"Checkpoint name": "Loads weights from checkpoint before making images. You can either use hash or a part of filename (as seen in settings) for checkpoint name. Recommended to use with Y axis for less switching.",
} }
function gradioApp(){ function gradioApp(){
...@@ -74,6 +76,90 @@ function gradioApp(){ ...@@ -74,6 +76,90 @@ function gradioApp(){
global_progressbar = null global_progressbar = null
function closeModal() {
gradioApp().getElementById("lightboxModal").style.display = "none";
}
function showModal(event) {
var source = event.target || event.srcElement;
gradioApp().getElementById("modalImage").src = source.src
var lb = gradioApp().getElementById("lightboxModal")
lb.style.display = "block";
lb.focus()
event.stopPropagation()
}
function negmod(n, m) {
return ((n % m) + m) % m;
}
function modalImageSwitch(offset){
var galleryButtons = gradioApp().querySelectorAll(".gallery-item.transition-all")
if(galleryButtons.length>1){
var currentButton = gradioApp().querySelector(".gallery-item.transition-all.\\!ring-2")
var result = -1
galleryButtons.forEach(function(v, i){ if(v==currentButton) { result = i } })
if(result != -1){
nextButton = galleryButtons[negmod((result+offset),galleryButtons.length)]
nextButton.click()
gradioApp().getElementById("modalImage").src = nextButton.children[0].src
setTimeout( function(){gradioApp().getElementById("lightboxModal").focus()},10)
}
}
}
function modalNextImage(event){
modalImageSwitch(1)
event.stopPropagation()
}
function modalPrevImage(event){
modalImageSwitch(-1)
event.stopPropagation()
}
function modalKeyHandler(event){
switch (event.key) {
case "ArrowLeft":
modalPrevImage(event)
break;
case "ArrowRight":
modalNextImage(event)
break;
}
}
function showGalleryImage(){
setTimeout(function() {
fullImg_preview = gradioApp().querySelectorAll('img.w-full.object-contain')
if(fullImg_preview != null){
fullImg_preview.forEach(function function_name(e) {
if(e && e.parentElement.tagName == 'DIV'){
e.style.cursor='pointer'
e.addEventListener('click', function (evt) {
showModal(evt)
},true);
}
});
}
}, 100);
}
function galleryImageHandler(e){
if(e && e.parentElement.tagName == 'BUTTON'){
e.onclick = showGalleryImage;
}
}
function addTitles(root){ function addTitles(root){
root.querySelectorAll('span, button, select').forEach(function(span){ root.querySelectorAll('span, button, select').forEach(function(span){
tooltip = titles[span.textContent]; tooltip = titles[span.textContent];
...@@ -115,13 +201,18 @@ function addTitles(root){ ...@@ -115,13 +201,18 @@ function addTitles(root){
img2img_preview.style.width = img2img_gallery.clientWidth + "px" img2img_preview.style.width = img2img_gallery.clientWidth + "px"
img2img_preview.style.height = img2img_gallery.clientHeight + "px" img2img_preview.style.height = img2img_gallery.clientHeight + "px"
} }
window.setTimeout(requestProgress, 500) window.setTimeout(requestProgress, 500)
}); });
mutationObserver.observe( progressbar, { childList:true, subtree:true }) mutationObserver.observe( progressbar, { childList:true, subtree:true })
} }
fullImg_preview = gradioApp().querySelectorAll('img.w-full')
if(fullImg_preview != null){
fullImg_preview.forEach(galleryImageHandler);
}
} }
document.addEventListener("DOMContentLoaded", function() { document.addEventListener("DOMContentLoaded", function() {
...@@ -129,6 +220,49 @@ document.addEventListener("DOMContentLoaded", function() { ...@@ -129,6 +220,49 @@ document.addEventListener("DOMContentLoaded", function() {
addTitles(gradioApp()); addTitles(gradioApp());
}); });
mutationObserver.observe( gradioApp(), { childList:true, subtree:true }) mutationObserver.observe( gradioApp(), { childList:true, subtree:true })
const modalFragment = document.createDocumentFragment();
const modal = document.createElement('div')
modal.onclick = closeModal;
const modalClose = document.createElement('span')
modalClose.className = 'modalClose cursor';
modalClose.innerHTML = '&times;'
modalClose.onclick = closeModal;
modal.id = "lightboxModal";
modal.tabIndex=0
modal.addEventListener('keydown', modalKeyHandler, true)
modal.appendChild(modalClose)
const modalImage = document.createElement('img')
modalImage.id = 'modalImage';
modalImage.onclick = closeModal;
modalImage.tabIndex=0
modalImage.addEventListener('keydown', modalKeyHandler, true)
modal.appendChild(modalImage)
const modalPrev = document.createElement('a')
modalPrev.className = 'modalPrev';
modalPrev.innerHTML = '&#10094;'
modalPrev.tabIndex=0
modalPrev.addEventListener('click',modalPrevImage,true);
modalPrev.addEventListener('keydown', modalKeyHandler, true)
modal.appendChild(modalPrev)
const modalNext = document.createElement('a')
modalNext.className = 'modalNext';
modalNext.innerHTML = '&#10095;'
modalNext.tabIndex=0
modalNext.addEventListener('click',modalNextImage,true);
modalNext.addEventListener('keydown', modalKeyHandler, true)
modal.appendChild(modalNext)
gradioApp().getRootNode().appendChild(modal)
document.body.appendChild(modalFragment);
}); });
function selected_gallery_index(){ function selected_gallery_index(){
...@@ -180,6 +314,15 @@ function submit(){ ...@@ -180,6 +314,15 @@ function submit(){
for(var i=0;i<arguments.length;i++){ for(var i=0;i<arguments.length;i++){
res.push(arguments[i]) res.push(arguments[i])
} }
// As it is currently, txt2img and img2img send back the previous output args (txt2img_gallery, generation_info, html_info) whenever you generate a new image.
// This can lead to uploading a huge gallery of previously generated images, which leads to an unnecessary delay between submitting and beginning to generate.
// I don't know why gradio is seding outputs along with inputs, but we can prevent sending the image gallery here, which seems to be an issue for some.
// If gradio at some point stops sending outputs, this may break something
if(Array.isArray(res[res.length - 3])){
res[res.length - 3] = null
}
return res return res
} }
......
...@@ -59,7 +59,7 @@ def find_noise_for_image(p, cond, uncond, cfg_scale, steps): ...@@ -59,7 +59,7 @@ def find_noise_for_image(p, cond, uncond, cfg_scale, steps):
return x / x.std() return x / x.std()
Cached = namedtuple("Cached", ["noise", "cfg_scale", "steps", "latent", "original_prompt"]) Cached = namedtuple("Cached", ["noise", "cfg_scale", "steps", "latent", "original_prompt", "original_negative_prompt"])
class Script(scripts.Script): class Script(scripts.Script):
...@@ -74,34 +74,45 @@ class Script(scripts.Script): ...@@ -74,34 +74,45 @@ class Script(scripts.Script):
def ui(self, is_img2img): def ui(self, is_img2img):
original_prompt = gr.Textbox(label="Original prompt", lines=1) original_prompt = gr.Textbox(label="Original prompt", lines=1)
original_negative_prompt = gr.Textbox(label="Original negative prompt", lines=1)
cfg = gr.Slider(label="Decode CFG scale", minimum=0.0, maximum=15.0, step=0.1, value=1.0) cfg = gr.Slider(label="Decode CFG scale", minimum=0.0, maximum=15.0, step=0.1, value=1.0)
st = gr.Slider(label="Decode steps", minimum=1, maximum=150, step=1, value=50) st = gr.Slider(label="Decode steps", minimum=1, maximum=150, step=1, value=50)
randomness = gr.Slider(label="randomness", minimum=0.0, maximum=1.0, step=0.01, value=0.0)
return [original_prompt, original_negative_prompt, cfg, st, randomness]
return [original_prompt, cfg, st] def run(self, p, original_prompt, original_negative_prompt, cfg, st, randomness):
def run(self, p, original_prompt, cfg, st):
p.batch_size = 1 p.batch_size = 1
p.batch_count = 1 p.batch_count = 1
def sample_extra(x, conditioning, unconditional_conditioning): def sample_extra(x, conditioning, unconditional_conditioning):
lat = (p.init_latent.cpu().numpy() * 10).astype(int) 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 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
same_everything = same_params and self.cache.latent.shape == lat.shape and np.abs(self.cache.latent-lat).sum() < 100 same_everything = same_params and self.cache.latent.shape == lat.shape and np.abs(self.cache.latent-lat).sum() < 100
if same_everything: if same_everything:
noise = self.cache.noise rec_noise = self.cache.noise
else: else:
shared.state.job_count += 1 shared.state.job_count += 1
cond = p.sd_model.get_learned_conditioning(p.batch_size * [original_prompt]) cond = p.sd_model.get_learned_conditioning(p.batch_size * [original_prompt])
uncond = p.sd_model.get_learned_conditioning(p.batch_size * [""]) uncond = p.sd_model.get_learned_conditioning(p.batch_size * [original_negative_prompt])
noise = find_noise_for_image(p, cond, uncond, cfg, st) rec_noise = find_noise_for_image(p, cond, uncond, cfg, st)
self.cache = Cached(noise, cfg, st, lat, original_prompt) self.cache = Cached(rec_noise, cfg, st, lat, original_prompt, original_negative_prompt)
rand_noise = processing.create_random_tensors(p.init_latent.shape[1:], [p.seed + x + 1 for x in range(p.init_latent.shape[0])])
combined_noise = ((1 - randomness) * rec_noise + randomness * rand_noise) / ((randomness**2 + (1-randomness)**2) ** 0.5)
sampler = samplers[p.sampler_index].constructor(p.sd_model) sampler = samplers[p.sampler_index].constructor(p.sd_model)
samples_ddim = sampler.sample(p, noise, conditioning, unconditional_conditioning) sigmas = sampler.model_wrap.get_sigmas(p.steps)
return samples_ddim
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)
p.sample = sample_extra p.sample = sample_extra
......
This diff is collapsed.
...@@ -13,28 +13,42 @@ from modules.shared import opts, cmd_opts, state ...@@ -13,28 +13,42 @@ from modules.shared import opts, cmd_opts, state
class Script(scripts.Script): class Script(scripts.Script):
def title(self): def title(self):
return "Prompts from file" return "Prompts from file or textbox"
def ui(self, is_img2img): def ui(self, is_img2img):
# This checkbox would look nicer as two tabs, but there are two problems:
# 1) There is a bug in Gradio 3.3 that prevents visibility from working on Tabs
# 2) Even with Gradio 3.3.1, returning a control (like Tabs) that can't be used as input
# causes a AttributeError: 'Tabs' object has no attribute 'preprocess' assert,
# due to the way Script assumes all controls returned can be used as inputs.
# Therefore, there's no good way to use grouping components right now,
# so we will use a checkbox! :)
checkbox_txt = gr.Checkbox(label="Show Textbox", value=False)
file = gr.File(label="File with inputs", type='bytes') file = gr.File(label="File with inputs", type='bytes')
prompt_txt = gr.TextArea(label="Prompts")
return [file] checkbox_txt.change(fn=lambda x: [gr.File.update(visible = not x), gr.TextArea.update(visible = x)], inputs=[checkbox_txt], outputs=[file, prompt_txt])
return [checkbox_txt, file, prompt_txt]
def run(self, p, data: bytes):
lines = [x.strip() for x in data.decode('utf8', errors='ignore').split("\n")] def run(self, p, checkbox_txt, data: bytes, prompt_txt: str):
if (checkbox_txt):
lines = [x.strip() for x in prompt_txt.splitlines()]
else:
lines = [x.strip() for x in data.decode('utf8', errors='ignore').split("\n")]
lines = [x for x in lines if len(x) > 0] lines = [x for x in lines if len(x) > 0]
batch_count = math.ceil(len(lines) / p.batch_size) img_count = len(lines) * p.n_iter
print(f"Will process {len(lines) * p.n_iter} images in {batch_count * p.n_iter} batches.") batch_count = math.ceil(img_count / p.batch_size)
loop_count = math.ceil(batch_count / p.n_iter)
print(f"Will process {img_count} images in {batch_count} batches.")
p.do_not_save_grid = True p.do_not_save_grid = True
state.job_count = batch_count state.job_count = batch_count
images = [] images = []
for batch_no in range(batch_count): for loop_no in range(loop_count):
state.job = f"{batch_no + 1} out of {batch_count * p.n_iter}" state.job = f"{loop_no + 1} out of {loop_count}"
p.prompt = lines[batch_no*p.batch_size:(batch_no+1)*p.batch_size] * p.n_iter p.prompt = lines[loop_no*p.batch_size:(loop_no+1)*p.batch_size] * p.n_iter
proc = process_images(p) proc = process_images(p)
images += proc.images images += proc.images
......
...@@ -10,7 +10,9 @@ import gradio as gr ...@@ -10,7 +10,9 @@ import gradio as gr
from modules import images from modules import images
from modules.processing import process_images, Processed from modules.processing import process_images, Processed
from modules.shared import opts, cmd_opts, state from modules.shared import opts, cmd_opts, state
import modules.shared as shared
import modules.sd_samplers import modules.sd_samplers
import modules.sd_models
import re import re
...@@ -41,6 +43,15 @@ def apply_sampler(p, x, xs): ...@@ -41,6 +43,15 @@ def apply_sampler(p, x, xs):
p.sampler_index = sampler_index p.sampler_index = sampler_index
def apply_checkpoint(p, x, xs):
applicable = [info for info in modules.sd_models.checkpoints_list.values() if x in info.title]
assert len(applicable) > 0, f'Checkpoint {x} for found'
info = applicable[0]
modules.sd_models.reload_model_weights(shared.sd_model, info)
def format_value_add_label(p, opt, x): def format_value_add_label(p, opt, x):
if type(x) == float: if type(x) == float:
x = round(x, 8) x = round(x, 8)
...@@ -74,15 +85,16 @@ axis_options = [ ...@@ -74,15 +85,16 @@ axis_options = [
AxisOption("CFG Scale", float, apply_field("cfg_scale"), format_value_add_label), AxisOption("CFG Scale", float, apply_field("cfg_scale"), format_value_add_label),
AxisOption("Prompt S/R", str, apply_prompt, format_value), AxisOption("Prompt S/R", str, apply_prompt, format_value),
AxisOption("Sampler", str, apply_sampler, format_value), AxisOption("Sampler", str, apply_sampler, format_value),
AxisOption("Checkpoint name", str, apply_checkpoint, format_value),
AxisOptionImg2Img("Denoising", float, apply_field("denoising_strength"), format_value_add_label), # as it is now all AxisOptionImg2Img items must go after AxisOption ones AxisOptionImg2Img("Denoising", float, apply_field("denoising_strength"), format_value_add_label), # as it is now all AxisOptionImg2Img items must go after AxisOption ones
] ]
def draw_xy_grid(p, xs, ys, x_label, y_label, cell, draw_legend): def draw_xy_grid(p, xs, ys, x_labels, y_labels, cell, draw_legend):
res = [] res = []
ver_texts = [[images.GridAnnotation(y_label(y))] for y in ys] ver_texts = [[images.GridAnnotation(y)] for y in y_labels]
hor_texts = [[images.GridAnnotation(x_label(x))] for x in xs] hor_texts = [[images.GridAnnotation(x)] for x in x_labels]
first_pocessed = None first_pocessed = None
...@@ -206,8 +218,8 @@ class Script(scripts.Script): ...@@ -206,8 +218,8 @@ class Script(scripts.Script):
p, p,
xs=xs, xs=xs,
ys=ys, ys=ys,
x_label=lambda x: x_opt.format_value(p, x_opt, x), x_labels=[x_opt.format_value(p, x_opt, x) for x in xs],
y_label=lambda y: y_opt.format_value(p, y_opt, y), y_labels=[y_opt.format_value(p, y_opt, y) for y in ys],
cell=cell, cell=cell,
draw_legend=draw_legend draw_legend=draw_legend
) )
...@@ -215,4 +227,7 @@ class Script(scripts.Script): ...@@ -215,4 +227,7 @@ class Script(scripts.Script):
if opts.grid_save: if opts.grid_save:
images.save_image(processed.images[0], p.outpath_grids, "xy_grid", prompt=p.prompt, seed=processed.seed, grid=True, p=p) images.save_image(processed.images[0], p.outpath_grids, "xy_grid", prompt=p.prompt, seed=processed.seed, grid=True, p=p)
# restore checkpoint in case it was changed by axes
modules.sd_models.reload_model_weights(shared.sd_model)
return processed return processed
.output-html p {margin: 0 0.5em;} .output-html p {margin: 0 0.5em;}
.performance { font-size: 0.85em; color: #444; }
.performance {
font-size: 0.85em;
color: #444;
display: flex;
justify-content: space-between;
white-space: nowrap;
}
.performance .time {
margin-right: 0;
}
.performance .vram {
margin-left: 0;
text-align: right;
}
#generate{ #generate{
min-height: 4.5em; min-height: 4.5em;
} }
#txt2img_gallery, #img2img_gallery{ @media screen and (min-width: 2500px) {
min-height: 768px; #txt2img_gallery, #img2img_gallery {
min-height: 768px;
}
} }
#txt2img_gallery img, #img2img_gallery img{ #txt2img_gallery img, #img2img_gallery img{
object-fit: scale-down; object-fit: scale-down;
} }
.justify-center.overflow-x-scroll {
justify-content: left;
}
.justify-center.overflow-x-scroll button:first-of-type {
margin-left: auto;
}
.justify-center.overflow-x-scroll button:last-of-type {
margin-right: auto;
}
#subseed_show{ #subseed_show{
min-width: 6em; min-width: 6em;
max-width: 6em; max-width: 6em;
...@@ -151,6 +182,12 @@ input[type="range"]{ ...@@ -151,6 +182,12 @@ input[type="range"]{
#txt2img_negative_prompt, #img2img_negative_prompt{ #txt2img_negative_prompt, #img2img_negative_prompt{
} }
#progressbar{
position: absolute;
z-index: 1000;
right: 0;
}
.progressDiv{ .progressDiv{
width: 100%; width: 100%;
height: 30px; height: 30px;
...@@ -174,3 +211,66 @@ input[type="range"]{ ...@@ -174,3 +211,66 @@ input[type="range"]{
border-radius: 8px; border-radius: 8px;
} }
#lightboxModal{
display: none;
position: fixed;
z-index: 900;
padding-top: 100px;
left: 0;
top: 0;
width: 100%;
height: 100%;
overflow: auto;
background-color: rgba(20, 20, 20, 0.95);
}
.modalClose {
color: white;
position: absolute;
top: 10px;
right: 25px;
font-size: 35px;
font-weight: bold;
}
.modalClose:hover,
.modalClose:focus {
color: #999;
text-decoration: none;
cursor: pointer;
}
#modalImage {
display: block;
margin-left: auto;
margin-right: auto;
margin-top: auto;
width: auto;
}
.modalPrev,
.modalNext {
cursor: pointer;
position: absolute;
top: 50%;
width: auto;
padding: 16px;
margin-top: -50px;
color: white;
font-weight: bold;
font-size: 20px;
transition: 0.6s ease;
border-radius: 0 3px 3px 0;
user-select: none;
-webkit-user-select: none;
}
.modalNext {
right: 0;
border-radius: 3px 0 0 3px;
}
.modalPrev:hover,
.modalNext:hover {
background-color: rgba(0, 0, 0, 0.8);
}
...@@ -3,13 +3,8 @@ import threading ...@@ -3,13 +3,8 @@ import threading
from modules.paths import script_path from modules.paths import script_path
import torch
from omegaconf import OmegaConf
import signal import signal
from ldm.util import instantiate_from_config
from modules.shared import opts, cmd_opts, state from modules.shared import opts, cmd_opts, state
import modules.shared as shared import modules.shared as shared
import modules.ui import modules.ui
...@@ -24,6 +19,7 @@ import modules.extras ...@@ -24,6 +19,7 @@ import modules.extras
import modules.lowvram import modules.lowvram
import modules.txt2img import modules.txt2img
import modules.img2img import modules.img2img
import modules.sd_models
modules.codeformer_model.setup_codeformer() modules.codeformer_model.setup_codeformer()
...@@ -33,29 +29,17 @@ shared.face_restorers.append(modules.face_restoration.FaceRestoration()) ...@@ -33,29 +29,17 @@ shared.face_restorers.append(modules.face_restoration.FaceRestoration())
esrgan.load_models(cmd_opts.esrgan_models_path) esrgan.load_models(cmd_opts.esrgan_models_path)
realesrgan.setup_realesrgan() realesrgan.setup_realesrgan()
queue_lock = threading.Lock()
def load_model_from_config(config, ckpt, verbose=False):
print(f"Loading model [{shared.sd_model_hash}] from {ckpt}")
pl_sd = torch.load(ckpt, map_location="cpu")
if "global_step" in pl_sd:
print(f"Global Step: {pl_sd['global_step']}")
sd = pl_sd["state_dict"]
model = instantiate_from_config(config.model) def wrap_queued_call(func):
m, u = model.load_state_dict(sd, strict=False) def f(*args, **kwargs):
if len(m) > 0 and verbose: with queue_lock:
print("missing keys:") res = func(*args, **kwargs)
print(m)
if len(u) > 0 and verbose:
print("unexpected keys:")
print(u)
if cmd_opts.opt_channelslast:
model = model.to(memory_format=torch.channels_last)
model.eval()
return model
return res
queue_lock = threading.Lock() return f
def wrap_gradio_gpu_call(func): def wrap_gradio_gpu_call(func):
...@@ -80,33 +64,8 @@ def wrap_gradio_gpu_call(func): ...@@ -80,33 +64,8 @@ def wrap_gradio_gpu_call(func):
modules.scripts.load_scripts(os.path.join(script_path, "scripts")) modules.scripts.load_scripts(os.path.join(script_path, "scripts"))
try: shared.sd_model = modules.sd_models.load_model()
# this silences the annoying "Some weights of the model checkpoint were not used when initializing..." message at start. shared.opts.onchange("sd_model_checkpoint", wrap_queued_call(lambda: modules.sd_models.reload_model_weights(shared.sd_model)))
from transformers import logging
logging.set_verbosity_error()
except Exception:
pass
with open(cmd_opts.ckpt, "rb") as file:
import hashlib
m = hashlib.sha256()
file.seek(0x100000)
m.update(file.read(0x10000))
shared.sd_model_hash = m.hexdigest()[0:8]
sd_config = OmegaConf.load(cmd_opts.config)
shared.sd_model = load_model_from_config(sd_config, cmd_opts.ckpt)
shared.sd_model = (shared.sd_model if cmd_opts.no_half else shared.sd_model.half())
if cmd_opts.lowvram or cmd_opts.medvram:
modules.lowvram.setup_for_low_vram(shared.sd_model, cmd_opts.medvram)
else:
shared.sd_model = shared.sd_model.to(shared.device)
modules.sd_hijack.model_hijack.hijack(shared.sd_model)
def webui(): def webui():
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment