lllyasviel/stable-diffusion-webui-forge
1
0
Fork 0
mirror of https://github.com/lllyasviel/stable-diffusion-webui-forge.git synced 2026-04-21 23:09:19 +00:00
Code Issues Packages Projects Releases Wiki Activity

move hypertile, svd, z123 to legacy extensions

Browse Source 
https://github.com/lllyasviel/forge-legacy-extensions
This commit is contained in:
layerdiffusion
2024-07-26 14:28:47 -07:00
parent bc22a930e5
commit 1d185584a7
3 changed files with 0 additions and 272 deletions
Download Patch File Download Diff File Expand all files Collapse all files

50
extensions-builtin/sd_forge_hypertile/scripts/forge_hypertile.py
View File

@@ -1,50 +0,0 @@
import gradio as gr
from modules import scripts
from ldm_patched.contrib.external_hypertile import HyperTile
opHyperTile = HyperTile()
class HyperTileForForge(scripts.Script):
sorting_priority = 13
def title(self):
return "HyperTile Integrated"
def show(self, is_img2img):
return scripts.AlwaysVisible
def ui(self, *args, **kwargs):
with gr.Accordion(open=False, label=self.title()):
enabled = gr.Checkbox(label='Enabled', value=False)
tile_size = gr.Slider(label='Tile Size', minimum=1, maximum=2048, step=1, value=256)
swap_size = gr.Slider(label='Swap Size', minimum=1, maximum=128, step=1, value=2)
max_depth = gr.Slider(label='Max Depth', minimum=0, maximum=10, step=1, value=0)
scale_depth = gr.Checkbox(label='Scale Depth', value=False)
return enabled, tile_size, swap_size, max_depth, scale_depth
def process_before_every_sampling(self, p, *script_args, **kwargs):
enabled, tile_size, swap_size, max_depth, scale_depth = script_args
tile_size, swap_size, max_depth = int(tile_size), int(swap_size), int(max_depth)
if not enabled:
return
unet = p.sd_model.forge_objects.unet
unet = opHyperTile.patch(unet, tile_size, swap_size, max_depth, scale_depth)[0]
p.sd_model.forge_objects.unet = unet
p.extra_generation_params.update(dict(
HyperTile_enabled=enabled,
HyperTile_tile_size=tile_size,
HyperTile_swap_size=swap_size,
HyperTile_max_depth=max_depth,
HyperTile_scale_depth=scale_depth,
))
return

121
extensions-builtin/sd_forge_svd/scripts/forge_svd.py
View File

@@ -1,121 +0,0 @@
import torch
import gradio as gr
import os
import pathlib
import modules.infotext_utils as parameters_copypaste
from modules import script_callbacks
from modules.paths import models_path
from modules.ui_common import ToolButton, refresh_symbol
from modules.ui_components import ResizeHandleRow
from modules import shared
from modules_forge.forge_util import numpy_to_pytorch, pytorch_to_numpy, write_images_to_mp4
from ldm_patched.modules.sd import load_checkpoint_guess_config
from ldm_patched.contrib.external_video_model import VideoLinearCFGGuidance, SVD_img2vid_Conditioning
from ldm_patched.contrib.external import KSampler, VAEDecode
opVideoLinearCFGGuidance = VideoLinearCFGGuidance()
opSVD_img2vid_Conditioning = SVD_img2vid_Conditioning()
opKSampler = KSampler()
opVAEDecode = VAEDecode()
svd_root = os.path.join(models_path, 'svd')
os.makedirs(svd_root, exist_ok=True)
svd_filenames = []
def update_svd_filenames():
global svd_filenames
svd_filenames = [
pathlib.Path(x).name for x in
shared.walk_files(svd_root, allowed_extensions=[".pt", ".ckpt", ".safetensors"])
]
return svd_filenames
@torch.inference_mode()
@torch.no_grad()
def predict(filename, width, height, video_frames, motion_bucket_id, fps, augmentation_level,
sampling_seed, sampling_steps, sampling_cfg, sampling_sampler_name, sampling_scheduler,
sampling_denoise, guidance_min_cfg, input_image):
filename = os.path.join(svd_root, filename)
model_raw, _, vae, clip_vision = \
load_checkpoint_guess_config(filename, output_vae=True, output_clip=False, output_clipvision=True)
model = opVideoLinearCFGGuidance.patch(model_raw, guidance_min_cfg)[0]
init_image = numpy_to_pytorch(input_image)
positive, negative, latent_image = opSVD_img2vid_Conditioning.encode(
clip_vision, init_image, vae, width, height, video_frames, motion_bucket_id, fps, augmentation_level)
output_latent = opKSampler.sample(model, sampling_seed, sampling_steps, sampling_cfg,
sampling_sampler_name, sampling_scheduler, positive,
negative, latent_image, sampling_denoise)[0]
output_pixels = opVAEDecode.decode(vae, output_latent)[0]
outputs = pytorch_to_numpy(output_pixels)
video_filename = write_images_to_mp4(outputs, fps=fps)
return outputs, video_filename
def on_ui_tabs():
with gr.Blocks() as svd_block:
with ResizeHandleRow():
with gr.Column():
input_image = gr.Image(label='Input Image', source='upload', type='numpy', height=400)
with gr.Row():
filename = gr.Dropdown(label="SVD Checkpoint Filename",
choices=svd_filenames,
value=svd_filenames[0] if len(svd_filenames) > 0 else None)
refresh_button = ToolButton(value=refresh_symbol, tooltip="Refresh")
refresh_button.click(
fn=lambda: gr.update(choices=update_svd_filenames()),
inputs=[], outputs=filename)
width = gr.Slider(label='Width', minimum=16, maximum=8192, step=8, value=1024)
height = gr.Slider(label='Height', minimum=16, maximum=8192, step=8, value=576)
video_frames = gr.Slider(label='Video Frames', minimum=1, maximum=4096, step=1, value=14)
motion_bucket_id = gr.Slider(label='Motion Bucket Id', minimum=1, maximum=1023, step=1, value=127)
fps = gr.Slider(label='Fps', minimum=1, maximum=1024, step=1, value=6)
augmentation_level = gr.Slider(label='Augmentation Level', minimum=0.0, maximum=10.0, step=0.01,
value=0.0)
sampling_steps = gr.Slider(label='Sampling Steps', minimum=1, maximum=200, step=1, value=20)
sampling_cfg = gr.Slider(label='CFG Scale', minimum=0.0, maximum=50.0, step=0.1, value=2.5)
sampling_denoise = gr.Slider(label='Sampling Denoise', minimum=0.0, maximum=1.0, step=0.01, value=1.0)
guidance_min_cfg = gr.Slider(label='Guidance Min Cfg', minimum=0.0, maximum=100.0, step=0.5, value=1.0)
sampling_sampler_name = gr.Radio(label='Sampler Name',
choices=['euler', 'euler_ancestral', 'heun', 'heunpp2', 'dpm_2',
'dpm_2_ancestral', 'lms', 'dpm_fast', 'dpm_adaptive',
'dpmpp_2s_ancestral', 'dpmpp_sde', 'dpmpp_sde_gpu',
'dpmpp_2m', 'dpmpp_2m_sde', 'dpmpp_2m_sde_gpu',
'dpmpp_3m_sde', 'dpmpp_3m_sde_gpu', 'ddpm', 'lcm', 'ddim',
'uni_pc', 'uni_pc_bh2'], value='euler')
sampling_scheduler = gr.Radio(label='Scheduler',
choices=['normal', 'karras', 'exponential', 'sgm_uniform', 'simple',
'ddim_uniform'], value='karras')
sampling_seed = gr.Number(label='Seed', value=12345, precision=0)
generate_button = gr.Button(value="Generate")
ctrls = [filename, width, height, video_frames, motion_bucket_id, fps, augmentation_level,
sampling_seed, sampling_steps, sampling_cfg, sampling_sampler_name, sampling_scheduler,
sampling_denoise, guidance_min_cfg, input_image]
with gr.Column():
output_video = gr.Video(autoplay=True)
output_gallery = gr.Gallery(label='Gallery', show_label=False, object_fit='contain',
visible=True, height=1024, columns=4)
generate_button.click(predict, inputs=ctrls, outputs=[output_gallery, output_video])
PasteField = parameters_copypaste.PasteField
paste_fields = [
PasteField(width, "Size-1", api="width"),
PasteField(height, "Size-2", api="height"),
]
parameters_copypaste.add_paste_fields("svd", init_img=input_image, fields=paste_fields)
return [(svd_block, "SVD", "svd")]
update_svd_filenames()
script_callbacks.on_ui_tabs(on_ui_tabs)

101
extensions-builtin/sd_forge_z123/scripts/forge_z123.py
View File

@@ -1,101 +0,0 @@
import torch
import gradio as gr
import os
import pathlib
from modules import script_callbacks
from modules.paths import models_path
from modules.ui_common import ToolButton, refresh_symbol
from modules.ui_components import ResizeHandleRow
from modules import shared
from modules_forge.forge_util import numpy_to_pytorch, pytorch_to_numpy
from ldm_patched.modules.sd import load_checkpoint_guess_config
from ldm_patched.contrib.external_stable3d import StableZero123_Conditioning
from ldm_patched.contrib.external import KSampler, VAEDecode
opStableZero123_Conditioning = StableZero123_Conditioning()
opKSampler = KSampler()
opVAEDecode = VAEDecode()
model_root = os.path.join(models_path, 'z123')
os.makedirs(model_root, exist_ok=True)
model_filenames = []
def update_model_filenames():
global model_filenames
model_filenames = [
pathlib.Path(x).name for x in
shared.walk_files(model_root, allowed_extensions=[".pt", ".ckpt", ".safetensors"])
]
return model_filenames
@torch.inference_mode()
@torch.no_grad()
def predict(filename, width, height, batch_size, elevation, azimuth,
sampling_seed, sampling_steps, sampling_cfg, sampling_sampler_name, sampling_scheduler, sampling_denoise, input_image):
filename = os.path.join(model_root, filename)
model, _, vae, clip_vision = \
load_checkpoint_guess_config(filename, output_vae=True, output_clip=False, output_clipvision=True)
init_image = numpy_to_pytorch(input_image)
positive, negative, latent_image = opStableZero123_Conditioning.encode(
clip_vision, init_image, vae, width, height, batch_size, elevation, azimuth)
output_latent = opKSampler.sample(model, sampling_seed, sampling_steps, sampling_cfg,
sampling_sampler_name, sampling_scheduler, positive,
negative, latent_image, sampling_denoise)[0]
output_pixels = opVAEDecode.decode(vae, output_latent)[0]
outputs = pytorch_to_numpy(output_pixels)
return outputs
def on_ui_tabs():
with gr.Blocks() as model_block:
with ResizeHandleRow():
with gr.Column():
input_image = gr.Image(label='Input Image', source='upload', type='numpy', height=400)
with gr.Row():
filename = gr.Dropdown(label="Zero123 Checkpoint Filename",
choices=model_filenames,
value=model_filenames[0] if len(model_filenames) > 0 else None)
refresh_button = ToolButton(value=refresh_symbol, tooltip="Refresh")
refresh_button.click(
fn=lambda: gr.update(choices=update_model_filenames),
inputs=[], outputs=filename)
width = gr.Slider(label='Width', minimum=16, maximum=8192, step=8, value=256)
height = gr.Slider(label='Height', minimum=16, maximum=8192, step=8, value=256)
batch_size = gr.Slider(label='Batch Size', minimum=1, maximum=4096, step=1, value=4)
elevation = gr.Slider(label='Elevation', minimum=-180.0, maximum=180.0, step=0.001, value=10.0)
azimuth = gr.Slider(label='Azimuth', minimum=-180.0, maximum=180.0, step=0.001, value=142.0)
sampling_denoise = gr.Slider(label='Sampling Denoise', minimum=0.0, maximum=1.0, step=0.01, value=1.0)
sampling_steps = gr.Slider(label='Sampling Steps', minimum=1, maximum=10000, step=1, value=20)
sampling_cfg = gr.Slider(label='CFG Scale', minimum=0.0, maximum=100.0, step=0.1, value=5.0)
sampling_sampler_name = gr.Radio(label='Sampler Name',
choices=['euler', 'euler_ancestral', 'heun', 'heunpp2', 'dpm_2',
'dpm_2_ancestral', 'lms', 'dpm_fast', 'dpm_adaptive',
'dpmpp_2s_ancestral', 'dpmpp_sde', 'dpmpp_sde_gpu',
'dpmpp_2m', 'dpmpp_2m_sde', 'dpmpp_2m_sde_gpu',
'dpmpp_3m_sde', 'dpmpp_3m_sde_gpu', 'ddpm', 'lcm', 'ddim',
'uni_pc', 'uni_pc_bh2'], value='euler')
sampling_scheduler = gr.Radio(label='Sampling Scheduler',
choices=['normal', 'karras', 'exponential', 'sgm_uniform', 'simple',
'ddim_uniform'], value='sgm_uniform')
sampling_seed = gr.Number(label='Seed', value=12345, precision=0)
generate_button = gr.Button(value="Generate")
ctrls = [filename, width, height, batch_size, elevation, azimuth, sampling_seed, sampling_steps, sampling_cfg, sampling_sampler_name, sampling_scheduler, sampling_denoise, input_image]
with gr.Column():
output_gallery = gr.Gallery(label='Gallery', show_label=False, object_fit='contain',
visible=True, height=1024, columns=4)
generate_button.click(predict, inputs=ctrls, outputs=[output_gallery])
return [(model_block, "Z123", "z123")]
update_model_filenames()
script_callbacks.on_ui_tabs(on_ui_tabs)