Update app.py

app.py

@@ -1,28 +1,43 @@
-import gradio as gr
-from transformers import pipeline
 import torch
+from PIL import Image
+import imageio
+from diffusers import StableVideoDiffusionPipeline
+import gradio as gr
+
+# Load the pipeline
+pipe = StableVideoDiffusionPipeline.from_pretrained(
+    "stabilityai/stable-video-diffusion-img2vid-xt", torch_dtype=torch.float16, variant="fp16"
+)
+pipe.enable_model_cpu_offload()
+
+def generate_video(image, seed=42, fps=7):
+    # Resize the image
+    image = image.resize((1024, 576))
+
+    # Set the generator seed
+    generator = torch.manual_seed(seed)
+
+    # Generate the frames
+    frames = pipe(image, decode_chunk_size=8, generator=generator).frames[0]
 
-# Load the model and pipeline
-model_id = "stabilityai/stable-video-diffusion-img2vid-xt"
-pipe = pipeline("text-to-video-generation", model=model_id, torch_dtype=torch.float16, device="cuda")
+    # Export the frames to a video
+    output_path = "generated.mp4"
+    imageio.mimwrite(output_path, frames, fps=fps)
 
-def generate_video(image, prompt):
-    # Generate the video from the image and prompt
-    video = pipe(prompt, image, num_inference_steps=50, guidance_scale=7.5)
-    return video
+    return output_path
 
 # Create the Gradio interface
 iface = gr.Interface(
     fn=generate_video,
     inputs=[
-        gr.Image(type="pil", label="Input Image"),
-        gr.Textbox(lines=2, placeholder="Enter a prompt...", label="Prompt")
+        gr.Image(type="pil", label="Upload Image"),
+        gr.Number(label="Seed", value=42),
+        gr.Number(label="FPS", value=7)
     ],
     outputs=gr.Video(label="Generated Video"),
-    title="Stable Video Diffusion img2vid-xt",
-    description="Generate a video from an image using the stabilityai/stable-video-diffusion-img2vid-xt model."
+    title="Stable Video Diffusion",
+    description="Generate a video from an uploaded image using Stable Video Diffusion."
 )
 
 # Launch the interface
-if __name__ == "__main__":
-    iface.launch()
+iface.launch()