Spaces:
Running
on
Zero
Running
on
Zero
Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,443 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import requests
|
| 3 |
+
import time
|
| 4 |
+
import io
|
| 5 |
+
import torch
|
| 6 |
+
from PIL import Image
|
| 7 |
+
import cv2
|
| 8 |
+
import numpy as np
|
| 9 |
+
from diffusers import StableDiffusionControlNetImg2ImgPipeline, ControlNetModel, DDIMScheduler
|
| 10 |
+
from diffusers.models import AutoencoderKL
|
| 11 |
+
from RealESRGAN import RealESRGAN
|
| 12 |
+
import gradio as gr
|
| 13 |
+
import subprocess
|
| 14 |
+
from tqdm import tqdm
|
| 15 |
+
import shutil
|
| 16 |
+
import uuid
|
| 17 |
+
import json
|
| 18 |
+
import threading
|
| 19 |
+
|
| 20 |
+
# Constants
|
| 21 |
+
USE_TORCH_COMPILE = False
|
| 22 |
+
ENABLE_CPU_OFFLOAD = os.getenv("ENABLE_CPU_OFFLOAD", "0") == "1"
|
| 23 |
+
|
| 24 |
+
# Ensure CUDA is available
|
| 25 |
+
if not torch.cuda.is_available():
|
| 26 |
+
raise RuntimeError("CUDA is not available. This script requires a CUDA-capable GPU.")
|
| 27 |
+
|
| 28 |
+
device = torch.device("cuda")
|
| 29 |
+
print(f"Using device: {device}")
|
| 30 |
+
|
| 31 |
+
# Replace the global abort_status with an Event
|
| 32 |
+
abort_event = threading.Event()
|
| 33 |
+
|
| 34 |
+
css = """
|
| 35 |
+
.gradio-container {
|
| 36 |
+
max-width: 100% !important;
|
| 37 |
+
padding: 20px !important;
|
| 38 |
+
}
|
| 39 |
+
#component-0 {
|
| 40 |
+
height: auto !important;
|
| 41 |
+
overflow: visible !important;
|
| 42 |
+
}
|
| 43 |
+
"""
|
| 44 |
+
|
| 45 |
+
def abort_job():
|
| 46 |
+
if abort_event.is_set():
|
| 47 |
+
return "Job is already being aborted."
|
| 48 |
+
abort_event.set()
|
| 49 |
+
return "Aborting job... Processing will stop after the current frame."
|
| 50 |
+
|
| 51 |
+
def check_ffmpeg():
|
| 52 |
+
try:
|
| 53 |
+
subprocess.run(["ffmpeg", "-version"], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, check=True)
|
| 54 |
+
return True
|
| 55 |
+
except (subprocess.CalledProcessError, FileNotFoundError):
|
| 56 |
+
return False
|
| 57 |
+
|
| 58 |
+
def download_file(url, folder_path, filename):
|
| 59 |
+
if not os.path.exists(folder_path):
|
| 60 |
+
os.makedirs(folder_path)
|
| 61 |
+
file_path = os.path.join(folder_path, filename)
|
| 62 |
+
|
| 63 |
+
if os.path.isfile(file_path):
|
| 64 |
+
print(f"File already exists: {file_path}")
|
| 65 |
+
else:
|
| 66 |
+
response = requests.get(url, stream=True)
|
| 67 |
+
if response.status_code == 200:
|
| 68 |
+
with open(file_path, 'wb') as file:
|
| 69 |
+
for chunk in response.iter_content(chunk_size=1024):
|
| 70 |
+
file.write(chunk)
|
| 71 |
+
print(f"File successfully downloaded and saved: {file_path}")
|
| 72 |
+
else:
|
| 73 |
+
print(f"Error downloading the file. Status code: {response.status_code}")
|
| 74 |
+
|
| 75 |
+
def download_models():
|
| 76 |
+
models = {
|
| 77 |
+
"MODEL": ("https://huggingface.co/dantea1118/juggernaut_reborn/resolve/main/juggernaut_reborn.safetensors?download=true", "models/models/Stable-diffusion", "juggernaut_reborn.safetensors"),
|
| 78 |
+
"UPSCALER_X2": ("https://huggingface.co/ai-forever/Real-ESRGAN/resolve/main/RealESRGAN_x2.pth?download=true", "models/upscalers/", "RealESRGAN_x2.pth"),
|
| 79 |
+
"UPSCALER_X4": ("https://huggingface.co/ai-forever/Real-ESRGAN/resolve/main/RealESRGAN_x4.pth?download=true", "models/upscalers/", "RealESRGAN_x4.pth"),
|
| 80 |
+
"NEGATIVE_1": ("https://huggingface.co/philz1337x/embeddings/resolve/main/verybadimagenegative_v1.3.pt?download=true", "models/embeddings", "verybadimagenegative_v1.3.pt"),
|
| 81 |
+
"NEGATIVE_2": ("https://huggingface.co/datasets/AddictiveFuture/sd-negative-embeddings/resolve/main/JuggernautNegative-neg.pt?download=true", "models/embeddings", "JuggernautNegative-neg.pt"),
|
| 82 |
+
"LORA_1": ("https://huggingface.co/philz1337x/loras/resolve/main/SDXLrender_v2.0.safetensors?download=true", "models/Lora", "SDXLrender_v2.0.safetensors"),
|
| 83 |
+
"LORA_2": ("https://huggingface.co/philz1337x/loras/resolve/main/more_details.safetensors?download=true", "models/Lora", "more_details.safetensors"),
|
| 84 |
+
"CONTROLNET": ("https://huggingface.co/lllyasviel/ControlNet-v1-1/resolve/main/control_v11f1e_sd15_tile.pth?download=true", "models/ControlNet", "control_v11f1e_sd15_tile.pth"),
|
| 85 |
+
"VAE": ("https://huggingface.co/stabilityai/sd-vae-ft-mse-original/resolve/main/vae-ft-mse-840000-ema-pruned.safetensors?download=true", "models/VAE", "vae-ft-mse-840000-ema-pruned.safetensors"),
|
| 86 |
+
}
|
| 87 |
+
|
| 88 |
+
for model, (url, folder, filename) in models.items():
|
| 89 |
+
download_file(url, folder, filename)
|
| 90 |
+
|
| 91 |
+
def timer_func(func):
|
| 92 |
+
def wrapper(*args, **kwargs):
|
| 93 |
+
start_time = time.time()
|
| 94 |
+
result = func(*args, **kwargs)
|
| 95 |
+
end_time = time.time()
|
| 96 |
+
print(f"{func.__name__} took {end_time - start_time:.2f} seconds")
|
| 97 |
+
return result
|
| 98 |
+
return wrapper
|
| 99 |
+
|
| 100 |
+
class ModelManager:
|
| 101 |
+
def __init__(self):
|
| 102 |
+
self.pipe = None
|
| 103 |
+
self.realesrgan_x2 = None
|
| 104 |
+
self.realesrgan_x4 = None
|
| 105 |
+
|
| 106 |
+
def load_models(self, progress=gr.Progress()):
|
| 107 |
+
if self.pipe is None:
|
| 108 |
+
progress(0, desc="Loading Stable Diffusion pipeline...")
|
| 109 |
+
self.pipe = self.setup_pipeline()
|
| 110 |
+
self.pipe.to(device)
|
| 111 |
+
if USE_TORCH_COMPILE:
|
| 112 |
+
progress(0.5, desc="Compiling the model...")
|
| 113 |
+
self.pipe.unet = torch.compile(self.pipe.unet, mode="reduce-overhead", fullgraph=True)
|
| 114 |
+
|
| 115 |
+
if self.realesrgan_x2 is None:
|
| 116 |
+
progress(0.7, desc="Loading RealESRGAN x2 model...")
|
| 117 |
+
self.realesrgan_x2 = RealESRGAN(device, scale=2)
|
| 118 |
+
self.realesrgan_x2.load_weights('models/upscalers/RealESRGAN_x2.pth', download=False)
|
| 119 |
+
|
| 120 |
+
if self.realesrgan_x4 is None:
|
| 121 |
+
progress(0.9, desc="Loading RealESRGAN x4 model...")
|
| 122 |
+
self.realesrgan_x4 = RealESRGAN(device, scale=4)
|
| 123 |
+
self.realesrgan_x4.load_weights('models/upscalers/RealESRGAN_x4.pth', download=False)
|
| 124 |
+
|
| 125 |
+
progress(1.0, desc="All models loaded successfully")
|
| 126 |
+
|
| 127 |
+
def setup_pipeline(self):
|
| 128 |
+
controlnet = ControlNetModel.from_single_file(
|
| 129 |
+
"models/ControlNet/control_v11f1e_sd15_tile.pth", torch_dtype=torch.float16
|
| 130 |
+
)
|
| 131 |
+
model_path = "models/models/Stable-diffusion/juggernaut_reborn.safetensors"
|
| 132 |
+
pipe = StableDiffusionControlNetImg2ImgPipeline.from_single_file(
|
| 133 |
+
model_path,
|
| 134 |
+
controlnet=controlnet,
|
| 135 |
+
torch_dtype=torch.float16,
|
| 136 |
+
use_safetensors=True,
|
| 137 |
+
safety_checker=None
|
| 138 |
+
)
|
| 139 |
+
vae = AutoencoderKL.from_single_file(
|
| 140 |
+
"models/VAE/vae-ft-mse-840000-ema-pruned.safetensors",
|
| 141 |
+
torch_dtype=torch.float16
|
| 142 |
+
)
|
| 143 |
+
pipe.vae = vae
|
| 144 |
+
pipe.load_textual_inversion("models/embeddings/verybadimagenegative_v1.3.pt")
|
| 145 |
+
pipe.load_textual_inversion("models/embeddings/JuggernautNegative-neg.pt")
|
| 146 |
+
pipe.load_lora_weights("models/Lora/SDXLrender_v2.0.safetensors")
|
| 147 |
+
pipe.fuse_lora(lora_scale=0.5)
|
| 148 |
+
pipe.load_lora_weights("models/Lora/more_details.safetensors")
|
| 149 |
+
pipe.fuse_lora(lora_scale=1.)
|
| 150 |
+
pipe.scheduler = DDIMScheduler.from_config(pipe.scheduler.config)
|
| 151 |
+
pipe.enable_freeu(s1=0.9, s2=0.2, b1=1.3, b2=1.4)
|
| 152 |
+
return pipe
|
| 153 |
+
|
| 154 |
+
@timer_func
|
| 155 |
+
def process_image(self, input_image, resolution, num_inference_steps, strength, hdr, guidance_scale):
|
| 156 |
+
condition_image = self.prepare_image(input_image, resolution, hdr)
|
| 157 |
+
|
| 158 |
+
prompt = "masterpiece, best quality, highres"
|
| 159 |
+
negative_prompt = "low quality, normal quality, ugly, blurry, blur, lowres, bad anatomy, bad hands, cropped, worst quality, verybadimagenegative_v1.3, JuggernautNegative-neg"
|
| 160 |
+
|
| 161 |
+
options = {
|
| 162 |
+
"prompt": prompt,
|
| 163 |
+
"negative_prompt": negative_prompt,
|
| 164 |
+
"image": condition_image,
|
| 165 |
+
"control_image": condition_image,
|
| 166 |
+
"width": condition_image.size[0],
|
| 167 |
+
"height": condition_image.size[1],
|
| 168 |
+
"strength": strength,
|
| 169 |
+
"num_inference_steps": num_inference_steps,
|
| 170 |
+
"guidance_scale": guidance_scale,
|
| 171 |
+
"generator": torch.Generator(device=device).manual_seed(0),
|
| 172 |
+
}
|
| 173 |
+
|
| 174 |
+
print("Running inference...")
|
| 175 |
+
result = self.pipe(**options).images[0]
|
| 176 |
+
print("Image processing completed successfully")
|
| 177 |
+
|
| 178 |
+
return result
|
| 179 |
+
|
| 180 |
+
def prepare_image(self, input_image, resolution, hdr):
|
| 181 |
+
condition_image = self.resize_and_upscale(input_image, resolution)
|
| 182 |
+
condition_image = self.create_hdr_effect(condition_image, hdr)
|
| 183 |
+
return condition_image
|
| 184 |
+
|
| 185 |
+
@timer_func
|
| 186 |
+
def resize_and_upscale(self, input_image, resolution):
|
| 187 |
+
scale = 2 if resolution <= 2048 else 4
|
| 188 |
+
|
| 189 |
+
if isinstance(input_image, str):
|
| 190 |
+
input_image = Image.open(input_image).convert("RGB")
|
| 191 |
+
elif isinstance(input_image, io.IOBase):
|
| 192 |
+
input_image = Image.open(input_image).convert("RGB")
|
| 193 |
+
elif isinstance(input_image, Image.Image):
|
| 194 |
+
input_image = input_image.convert("RGB")
|
| 195 |
+
elif isinstance(input_image, np.ndarray):
|
| 196 |
+
input_image = Image.fromarray(input_image).convert("RGB")
|
| 197 |
+
else:
|
| 198 |
+
raise ValueError(f"Unsupported input type for input_image: {type(input_image)}")
|
| 199 |
+
|
| 200 |
+
W, H = input_image.size
|
| 201 |
+
k = float(resolution) / min(H, W)
|
| 202 |
+
H = int(round(H * k / 64.0)) * 64
|
| 203 |
+
W = int(round(W * k / 64.0)) * 64
|
| 204 |
+
img = input_image.resize((W, H), resample=Image.LANCZOS)
|
| 205 |
+
|
| 206 |
+
if scale == 2:
|
| 207 |
+
img = self.realesrgan_x2.predict(img)
|
| 208 |
+
else:
|
| 209 |
+
img = self.realesrgan_x4.predict(img)
|
| 210 |
+
|
| 211 |
+
return img
|
| 212 |
+
|
| 213 |
+
@timer_func
|
| 214 |
+
def create_hdr_effect(self, original_image, hdr):
|
| 215 |
+
if hdr == 0:
|
| 216 |
+
return original_image
|
| 217 |
+
cv_original = cv2.cvtColor(np.array(original_image), cv2.COLOR_RGB2BGR)
|
| 218 |
+
factors = [1.0 - 0.9 * hdr, 1.0 - 0.7 * hdr, 1.0 - 0.45 * hdr,
|
| 219 |
+
1.0 - 0.25 * hdr, 1.0, 1.0 + 0.2 * hdr,
|
| 220 |
+
1.0 + 0.4 * hdr, 1.0 + 0.6 * hdr, 1.0 + 0.8 * hdr]
|
| 221 |
+
images = [cv2.convertScaleAbs(cv_original, alpha=factor) for factor in factors]
|
| 222 |
+
merge_mertens = cv2.createMergeMertens()
|
| 223 |
+
hdr_image = merge_mertens.process(images)
|
| 224 |
+
hdr_image_8bit = np.clip(hdr_image * 255, 0, 255).astype('uint8')
|
| 225 |
+
hdr_result = Image.fromarray(cv2.cvtColor(hdr_image_8bit, cv2.COLOR_BGR2RGB))
|
| 226 |
+
|
| 227 |
+
return hdr_result
|
| 228 |
+
|
| 229 |
+
model_manager = ModelManager()
|
| 230 |
+
|
| 231 |
+
def extract_frames(video_path, output_folder):
|
| 232 |
+
os.makedirs(output_folder, exist_ok=True)
|
| 233 |
+
command = [
|
| 234 |
+
'ffmpeg',
|
| 235 |
+
'-i', video_path,
|
| 236 |
+
'-vf', 'fps=30',
|
| 237 |
+
f'{output_folder}/frame_%06d.png'
|
| 238 |
+
]
|
| 239 |
+
subprocess.run(command, check=True)
|
| 240 |
+
|
| 241 |
+
def frames_to_video(input_folder, output_path, fps, original_video_path):
|
| 242 |
+
# First, create the video from frames without audio
|
| 243 |
+
temp_output_path = output_path + "_temp.mp4"
|
| 244 |
+
video_command = [
|
| 245 |
+
'ffmpeg',
|
| 246 |
+
'-framerate', str(fps),
|
| 247 |
+
'-i', f'{input_folder}/frame_%06d.png',
|
| 248 |
+
'-c:v', 'libx264',
|
| 249 |
+
'-pix_fmt', 'yuv420p',
|
| 250 |
+
temp_output_path
|
| 251 |
+
]
|
| 252 |
+
subprocess.run(video_command, check=True)
|
| 253 |
+
|
| 254 |
+
# Then, copy the audio from the original video and add it to the new video
|
| 255 |
+
final_command = [
|
| 256 |
+
'ffmpeg',
|
| 257 |
+
'-i', temp_output_path,
|
| 258 |
+
'-i', original_video_path,
|
| 259 |
+
'-c:v', 'copy',
|
| 260 |
+
'-c:a', 'aac',
|
| 261 |
+
'-map', '0:v:0',
|
| 262 |
+
'-map', '1:a:0?',
|
| 263 |
+
'-shortest',
|
| 264 |
+
output_path
|
| 265 |
+
]
|
| 266 |
+
subprocess.run(final_command, check=True)
|
| 267 |
+
|
| 268 |
+
# Remove the temporary file
|
| 269 |
+
os.remove(temp_output_path)
|
| 270 |
+
|
| 271 |
+
@timer_func
|
| 272 |
+
def process_video(input_video, resolution, num_inference_steps, strength, hdr, guidance_scale, max_frames=None, frame_interval=1, preserve_frames=False, progress=gr.Progress()):
|
| 273 |
+
abort_event.clear() # Clear the abort flag at the start of a new job
|
| 274 |
+
print("Starting video processing...")
|
| 275 |
+
model_manager.load_models(progress) # Ensure models are loaded
|
| 276 |
+
|
| 277 |
+
# Create a new job folder
|
| 278 |
+
job_id = str(uuid.uuid4())
|
| 279 |
+
job_folder = os.path.join("jobs", job_id)
|
| 280 |
+
os.makedirs(job_folder, exist_ok=True)
|
| 281 |
+
|
| 282 |
+
# Save job config
|
| 283 |
+
config = {
|
| 284 |
+
"resolution": resolution,
|
| 285 |
+
"num_inference_steps": num_inference_steps,
|
| 286 |
+
"strength": strength,
|
| 287 |
+
"hdr": hdr,
|
| 288 |
+
"guidance_scale": guidance_scale,
|
| 289 |
+
"max_frames": max_frames,
|
| 290 |
+
"frame_interval": frame_interval,
|
| 291 |
+
"preserve_frames": preserve_frames
|
| 292 |
+
}
|
| 293 |
+
with open(os.path.join(job_folder, "config.json"), "w") as f:
|
| 294 |
+
json.dump(config, f)
|
| 295 |
+
|
| 296 |
+
# If input_video is a file object or has a 'name' attribute, use its name
|
| 297 |
+
if isinstance(input_video, io.IOBase) or hasattr(input_video, 'name'):
|
| 298 |
+
input_video = input_video.name
|
| 299 |
+
|
| 300 |
+
# Set up folders
|
| 301 |
+
frames_folder = os.path.join(job_folder, "video_frames")
|
| 302 |
+
processed_frames_folder = os.path.join(job_folder, "processed_frames")
|
| 303 |
+
os.makedirs(frames_folder, exist_ok=True)
|
| 304 |
+
os.makedirs(processed_frames_folder, exist_ok=True)
|
| 305 |
+
|
| 306 |
+
# Extract frames
|
| 307 |
+
progress(0.1, desc="Extracting frames...")
|
| 308 |
+
extract_frames(input_video, frames_folder)
|
| 309 |
+
|
| 310 |
+
# Process selected frames
|
| 311 |
+
frame_files = sorted(os.listdir(frames_folder))
|
| 312 |
+
total_frames = len(frame_files)
|
| 313 |
+
frames_to_process = min(max_frames, total_frames) if max_frames else total_frames
|
| 314 |
+
|
| 315 |
+
try:
|
| 316 |
+
progress(0.2, desc="Processing frames...")
|
| 317 |
+
for i, frame_file in enumerate(tqdm(frame_files[:frames_to_process], desc="Processing frames")):
|
| 318 |
+
if abort_event.is_set():
|
| 319 |
+
print("Job aborted. Stopping processing of new frames.")
|
| 320 |
+
break
|
| 321 |
+
|
| 322 |
+
output_frame_path = os.path.join(processed_frames_folder, frame_file)
|
| 323 |
+
if not preserve_frames or not os.path.exists(output_frame_path):
|
| 324 |
+
if i % frame_interval == 0:
|
| 325 |
+
# Process this frame
|
| 326 |
+
input_image = Image.open(os.path.join(frames_folder, frame_file))
|
| 327 |
+
processed_image = model_manager.process_image(input_image, resolution, num_inference_steps, strength, hdr, guidance_scale)
|
| 328 |
+
processed_image.save(output_frame_path)
|
| 329 |
+
else:
|
| 330 |
+
# Copy the previous processed frame or the original frame
|
| 331 |
+
prev_frame = f"frame_{int(frame_file.split('_')[1].split('.')[0]) - 1:06d}.png"
|
| 332 |
+
prev_frame_path = os.path.join(processed_frames_folder, prev_frame)
|
| 333 |
+
if os.path.exists(prev_frame_path):
|
| 334 |
+
shutil.copy2(prev_frame_path, output_frame_path)
|
| 335 |
+
else:
|
| 336 |
+
shutil.copy2(os.path.join(frames_folder, frame_file), output_frame_path)
|
| 337 |
+
progress((0.2 + 0.7 * (i + 1) / frames_to_process), desc=f"Processing frame {i+1}/{frames_to_process}")
|
| 338 |
+
|
| 339 |
+
# Always attempt to reassemble video
|
| 340 |
+
progress(0.9, desc="Reassembling video...")
|
| 341 |
+
input_filename = os.path.splitext(os.path.basename(input_video))[0]
|
| 342 |
+
output_video = os.path.join(job_folder, f"{input_filename}_upscaled.mp4")
|
| 343 |
+
frames_to_video(processed_frames_folder, output_video, 30, input_video)
|
| 344 |
+
|
| 345 |
+
if abort_event.is_set():
|
| 346 |
+
progress(1.0, desc="Video processing aborted, but partial result saved")
|
| 347 |
+
print("Video processing aborted, but partial result saved")
|
| 348 |
+
else:
|
| 349 |
+
progress(1.0, desc="Video processing completed successfully")
|
| 350 |
+
print("Video processing completed successfully")
|
| 351 |
+
|
| 352 |
+
return output_video
|
| 353 |
+
|
| 354 |
+
except Exception as e:
|
| 355 |
+
print(f"An error occurred during processing: {str(e)}")
|
| 356 |
+
progress(1.0, desc=f"Error: {str(e)}")
|
| 357 |
+
return None
|
| 358 |
+
|
| 359 |
+
def gradio_process_media(input_media, resolution, num_inference_steps, strength, hdr, guidance_scale, max_frames, frame_interval, preserve_frames, progress=gr.Progress()):
|
| 360 |
+
abort_event.clear() # Clear the abort flag at the start of a new job
|
| 361 |
+
if input_media is None:
|
| 362 |
+
return None, "No input media provided."
|
| 363 |
+
|
| 364 |
+
print(f"Input media type: {type(input_media)}")
|
| 365 |
+
|
| 366 |
+
# Get the file path
|
| 367 |
+
if isinstance(input_media, str):
|
| 368 |
+
file_path = input_media
|
| 369 |
+
elif isinstance(input_media, io.IOBase):
|
| 370 |
+
file_path = input_media.name
|
| 371 |
+
elif hasattr(input_media, 'name'):
|
| 372 |
+
file_path = input_media.name
|
| 373 |
+
else:
|
| 374 |
+
raise ValueError(f"Unsupported input type: {type(input_media)}")
|
| 375 |
+
|
| 376 |
+
print(f"File path: {file_path}")
|
| 377 |
+
|
| 378 |
+
# Ensure models are loaded
|
| 379 |
+
model_manager.load_models(progress)
|
| 380 |
+
|
| 381 |
+
# Check if the file is a video
|
| 382 |
+
video_extensions = ('.mp4', '.avi', '.mov', '.mkv')
|
| 383 |
+
if file_path.lower().endswith(video_extensions):
|
| 384 |
+
print("Processing video...")
|
| 385 |
+
result = process_video(file_path, resolution, num_inference_steps, strength, hdr, guidance_scale, max_frames, frame_interval, preserve_frames, progress)
|
| 386 |
+
if result:
|
| 387 |
+
return result, "Video processing completed successfully."
|
| 388 |
+
else:
|
| 389 |
+
return None, "Error occurred during video processing."
|
| 390 |
+
else:
|
| 391 |
+
print("Processing image...")
|
| 392 |
+
result = model_manager.process_image(file_path, resolution, num_inference_steps, strength, hdr, guidance_scale)
|
| 393 |
+
if result:
|
| 394 |
+
# Save the processed image
|
| 395 |
+
output_path = os.path.join("processed_images", f"processed_{os.path.basename(file_path)}")
|
| 396 |
+
os.makedirs(os.path.dirname(output_path), exist_ok=True)
|
| 397 |
+
result.save(output_path)
|
| 398 |
+
return output_path, "Image processing completed successfully."
|
| 399 |
+
else:
|
| 400 |
+
return None, "Error occurred during image processing."
|
| 401 |
+
|
| 402 |
+
# Update the Gradio interface
|
| 403 |
+
with gr.Blocks(css=css, theme=gr.themes.Default(primary_hue="blue")) as iface:
|
| 404 |
+
gr.Markdown(
|
| 405 |
+
"""
|
| 406 |
+
# SimpleSlowVideoUpscaler
|
| 407 |
+
|
| 408 |
+
Built by [Hrishi](https://twitter.com/hrishioa) and Claude
|
| 409 |
+
|
| 410 |
+
This project is based on [gokaygokay/Tile-Upscaler](https://huggingface.co/spaces/gokaygokay/Tile-Upscaler), which in turn is inspired by ideas from [@philz1337x/clarity-upscaler](https://github.com/philz1337x/clarity-upscaler) and [@BatouResearch/controlnet-tile-upscale](https://github.com/BatouResearch/controlnet-tile-upscale).
|
| 411 |
+
|
| 412 |
+
If you find this project useful, please consider [starring it on GitHub](https://github.com/hrishioa/SimpleSlowVideoUpscaler)!
|
| 413 |
+
"""
|
| 414 |
+
)
|
| 415 |
+
|
| 416 |
+
with gr.Row():
|
| 417 |
+
with gr.Column(scale=2):
|
| 418 |
+
input_media = gr.File(label="Input Media (Image or Video)")
|
| 419 |
+
resolution = gr.Slider(256, 2048, 512, step=256, label="Resolution")
|
| 420 |
+
num_inference_steps = gr.Slider(1, 50, 10, step=1, label="Number of Inference Steps")
|
| 421 |
+
strength = gr.Slider(0, 1, 0.3, step=0.01, label="Strength")
|
| 422 |
+
hdr = gr.Slider(0, 1, 0, step=0.1, label="HDR Effect")
|
| 423 |
+
guidance_scale = gr.Slider(0, 20, 5, step=0.5, label="Guidance Scale")
|
| 424 |
+
max_frames = gr.Number(label="Max Frames to Process (leave empty for full video)", precision=0)
|
| 425 |
+
frame_interval = gr.Slider(1, 30, 1, step=1, label="Frame Interval (process every nth frame)")
|
| 426 |
+
preserve_frames = gr.Checkbox(label="Preserve Existing Processed Frames", value=True)
|
| 427 |
+
|
| 428 |
+
with gr.Column(scale=1):
|
| 429 |
+
submit_button = gr.Button("Process Media")
|
| 430 |
+
abort_button = gr.Button("Abort Job")
|
| 431 |
+
output = gr.File(label="Processed Media")
|
| 432 |
+
status = gr.Markdown("Ready to process media.")
|
| 433 |
+
|
| 434 |
+
submit_button.click(
|
| 435 |
+
gradio_process_media,
|
| 436 |
+
inputs=[input_media, resolution, num_inference_steps, strength, hdr, guidance_scale, max_frames, frame_interval, preserve_frames],
|
| 437 |
+
outputs=[output, status]
|
| 438 |
+
)
|
| 439 |
+
|
| 440 |
+
abort_button.click(abort_job, inputs=[], outputs=status)
|
| 441 |
+
|
| 442 |
+
# Launch the Gradio app
|
| 443 |
+
iface.launch()
|