Create app.py

app.py

@@ -0,0 +1,72 @@
+import gradio as gr
+from diffusers import DiffusionPipeline
+import torch
+import numpy as np
+from PIL import Image
+import os
+
+# Load the model (we'll use caching to improve performance)
+def load_model():
+    model_id = "cerspense/zeroscope_v2_576w"
+    pipe = DiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16)
+    pipe = pipe.to("cuda")
+    return pipe
+
+# Generate video from text
+def generate_video(prompt, num_frames=24, num_inference_steps=50):
+    # Load model (with caching)
+    if not hasattr(generate_video, "pipe"):
+        generate_video.pipe = load_model()
+    
+    pipe = generate_video.pipe
+    
+    # Generate video
+    video_frames = pipe(prompt, num_frames=num_frames, num_inference_steps=num_inference_steps).frames
+    
+    # Convert frames to video file
+    output_path = "output.mp4"
+    frame_rate = 8  # frames per second
+    
+    # Save as GIF (simpler implementation)
+    gif_path = "output.gif"
+    video_frames[0].save(
+        gif_path,
+        save_all=True,
+        append_images=video_frames[1:],
+        duration=1000//frame_rate,
+        loop=0
+    )
+    
+    return gif_path
+
+# Gradio interface
+with gr.Blocks() as demo:
+    gr.Markdown("# 🎥 Text-to-Video Generator")
+    gr.Markdown("Generate short video clips from text prompts using Zeroscope model")
+    
+    with gr.Row():
+        with gr.Column():
+            prompt = gr.Textbox(label="Enter your prompt", placeholder="A robot dancing on the moon")
+            frames = gr.Slider(minimum=8, maximum=48, value=24, step=8, label="Number of frames")
+            steps = gr.Slider(minimum=20, maximum=100, value=50, step=5, label="Inference steps")
+            submit = gr.Button("Generate Video")
+        
+        with gr.Column():
+            output = gr.Image(label="Generated Video", format="gif")
+    
+    examples = gr.Examples(
+        examples=[
+            ["A spaceship flying through a nebula"],
+            ["A cat wearing sunglasses surfing on a wave"],
+            ["A futuristic city with flying cars at sunset"]
+        ],
+        inputs=prompt
+    )
+    
+    submit.click(
+        fn=generate_video,
+        inputs=[prompt, frames, steps],
+        outputs=output
+    )
+
+demo.launch()