Update app.py

app.py

@@ -1,10 +1,6 @@
 import os
-import sys
-import uuid
-import shutil
-import time
-import gradio as gr
 import torch
+import gradio as gr
 from diffusers import StableVideoDiffusionPipeline
 from PIL import Image
 import numpy as np
@@ -12,127 +8,72 @@ import cv2
 import tempfile
 from diffusers.utils import export_to_video
 
-class WanAnimateApp:
-    def __init__(self):
-        model_name = "stabilityai/stable-video-diffusion-img2vid-xt"
-        dtype = torch.float16 if torch.cuda.is_available() else torch.float32
-        self.pipe = StableVideoDiffusionPipeline.from_pretrained(
-            model_name,
-            torch_dtype=dtype,
+pipe = None
+def load():
+    global pipe
+    if pipe is None:
+        pipe = StableVideoDiffusionPipeline.from_pretrained(
+            "stabilityai/stable-video-diffusion-img2vid-xt",
+            torch_dtype=torch.float16,
             variant="fp16"
         )
-        device = "cuda" if torch.cuda.is_available() else "cpu"
-        self.pipe.to(device)
-        gr.Info(f"Модель на {device.upper()}. Если CPU — переключись на GPU в Settings!")
-
-    def predict(self, ref_img, video, model_id, model, progress=gr.Progress()):
-        if ref_img is None or video is None:
-            return None, "Upload both image and video."
-
-        progress(0, desc="Подготовка...")
-        ref_image = Image.fromarray(ref_img).convert("RGB").resize((576, 320))
-
-        cap = cv2.VideoCapture(video)
-        frame_count = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
-        cap.release()
-        motion_hint = f" with dynamic motion from {frame_count} frames"
-
-        num_frames = 25 if model == "wan-pro" else 14
-        num_steps = 25 if model == "wan-pro" else 15
-
-        noise_aug_strength = 0.02
-        if model_id == "wan2.2-animate-mix":
-            noise_aug_strength = 0.1
-
-        generator = torch.Generator(device="cuda" if torch.cuda.is_available() else "cpu").manual_seed(42)
-        start_time = time.time()
-        output = self.pipe(
-            ref_image,
-            num_inference_steps=num_steps,
-            num_frames=num_frames,
-            generator=generator,
-            decode_chunk_size=2,
-            noise_aug_strength=noise_aug_strength,
-            callback_on_step_end=lambda step, timestep, latents: progress((step + 1) / num_steps, desc=f"Шаг {step + 1}/{num_steps}. Время: {int(time.time() - start_time)} сек")
-        ).frames[0]
-
-        with tempfile.NamedTemporaryFile(suffix=".mp4", delete=False) as temp_video:
-            export_to_video(output, temp_video.name, fps=7)
-
-        return temp_video.name, "SUCCEEDED" + motion_hint
-
-def start_app():
-    os.makedirs("/tmp/gradio", exist_ok=True)
-
-    app = WanAnimateApp()
-    with gr.Blocks(title="Wan2.2-Animate (Local No API)") as demo:
-        gr.HTML("""
-            Wan2.2-Animate: Unified Character Animation and Replacement with Holistic Replication
-            Local version without API (SVD Proxy)
-            Tongyi Lab, Alibaba
-            📄Paper 💻GitHub 🤗HF Model
-        """)
-
-        with gr.Accordion("Usage Instructions (инструкции)", open=False):
-            gr.HTML("""
-                ‼️Usage (использования) Wan-Animate supports two modes:
-
-                * Move Mode: animate the character in input image with movements from the input video
-
-                * Mix Mode: replace the character in input video with the character in input video
-
-                Wan-Animate supports two modes:
-
-                * Move Mode: Use the movements extracted from the input video to drive the character in the input image
-
-                * Mix Mode: Use the character in the input image to replace the character in the input video
-
-                Currently, the following restrictions apply to inputs:
-
-                * Video file size: Less than 200MB
-
-                * Video resolution: The shorter side must be greater than 200, and the longer side must be less than 2048
-
-                * Video duration: 2s to 30s
-
-                * Video aspect ratio: 1:3 to 3:1
-
-                * Video formats: mp4, avi, mov
-
-                * Image file size: Less than 5MB
-
-                * Image resolution: The shorter side must be greater than 200, and the longer side must be less than 4096
-
-                * Image formats: jpg, png, jpeg, webp, bmp
-
-                Current, the inference quality has two variants. You can use our open-source code for more flexible configuration.
-
-                * wan-pro: 25fps, 720p
-
-                * wan-std: 15fps, 720p
-            """)
-
-        with gr.Row():
-            with gr.Column():
-                ref_img = gr.Image(label="Reference Image (изображение)", type="numpy", sources=["upload"])
-                video = gr.Video(label="Template Video (шаблонное видео)", sources=["upload"])
-                with gr.Row():
-                    model_id = gr.Dropdown(label="Mode (режим)", choices=["wan2.2-animate-move", "wan2.2-animate-mix"], value="wan2.2-animate-move")
-                    model = gr.Dropdown(label="Inference Quality (качество)", choices=["wan-pro", "wan-std"], value="wan-pro")
-                run_button = gr.Button("Generate Video (генерировать)")
-
-            with gr.Column():
-                output_video = gr.Video(label="Output Video (результат)")
-                output_status = gr.Textbox(label="Status (статус)")
-
-        run_button.click(
-            fn=app.predict,
-            inputs=[ref_img, video, model_id, model],
-            outputs=[output_video, output_status]
-        )
-
-    demo.queue(default_concurrency_limit=1)
-    demo.launch(server_name="0.0.0.0", server_port=7860)
-
-if __name__ == "__main__":
-    start_app()
+        pipe.to("cuda")
+        gr.Info("Модель на GPU — генерация 30–60 сек")
+    return pipe
+
+def run(ref_img, video, mode, quality, prog=gr.Progress()):
+    pipe = load()
+    prog(0, desc="Подготовка...")
+    img = Image.fromarray(ref_img).convert("RGB").resize((576, 320))
+
+    cap = cv2.VideoCapture(video)
+    n = int(cap.get(7)); cap.release()
+    hint = f" ({n} кадров)"
+
+    steps = 25 if quality == "wan-pro" else 15
+    frames = 25 if quality == "wan-pro" else 14
+    noise = 0.1 if mode == "wan2.2-animate-mix" else 0.02
+
+    def cb(step, *_): 
+        prog((step+1)/steps, desc=f"Шаг {step+1}/{steps}")
+    prog(0.1, desc="Генерация...")
+
+    out = pipe(
+        img,
+        num_inference_steps=steps,
+        num_frames=frames,
+        decode_chunk_size=2,
+        noise_aug_strength=noise,
+        callback_on_step_end=cb
+    ).frames[0]
+
+    tmp = tempfile.NamedTemporaryFile(delete=False, suffix=".mp4")
+    export_to_video(out, tmp.name, fps=7)
+    return tmp.name, "Готово!" + hint
+
+with gr.Blocks() as demo:
+    gr.Markdown("# Wan2.2-Animate (GPU)")
+    with gr.Accordion("Инструкция", open=False):
+        gr.Markdown("Загрузи фото + видео → выбери режим → жми Generate")
+
+    with gr.Row():
+        with gr.Column():
+            img = gr.Image(label="Фото", type="numpy")
+            vid = gr.Video(label="Видео")
+            with gr.Row():
+                mode = gr.Dropdown(["wan2.2-animate-move", "wan2.2-animate-mix"], 
+                                  label="Режим", value="wan2.2-animate-move")
+                qual = gr.Dropdown(["wan-pro", "wan-std"], label="Качество", value="wan-pro")
+            btn = gr.Button("Generate Video")
+        with gr.Column():
+            out = gr.Video(label="Результат")
+            stat = gr.Textbox(label="Статус")
+
+    btn.click(run, [img, vid, mode, qual], [out, stat])
+
+demo.queue(max_size=2).launch(
+    server_name="0.0.0.0",
+    server_port=7860,
+    share=True,           # ← ФИКС 1
+    enable_queue=True
+)