app.py

import gradio as gr
import torch
from diffusers import I2VGenXLPipeline
from diffusers.utils import export_to_gif, load_image
import spaces

# Initialize the pipeline
pipeline = I2VGenXLPipeline.from_pretrained("ali-vilab/i2vgen-xl", torch_dtype=torch.float16, variant="fp16")
pipeline.enable_model_cpu_offload()

@spaces.GPU(duration=240)
def generate_gif(image, prompt, negative_prompt, num_inference_steps, guidance_scale, seed):
    # Load the image
    image = load_image(image).convert("RGB")

    # Set the generator seed
    generator = torch.manual_seed(seed)

    # Generate the frames
    frames = pipeline(
        prompt=prompt,
        image=image,
        num_inference_steps=num_inference_steps,
        negative_prompt=negative_prompt,
        guidance_scale=guidance_scale,
        generator=generator
    ).frames[0]

    # Export to GIF
    gif_path = "i2v.gif"
    export_to_gif(frames, gif_path)

    return gif_path

# Create the Gradio interface
iface = gr.Interface(
    fn=generate_gif,
    inputs=[
        gr.Image(type="filepath", label="Input Image"),
        gr.Textbox(lines=2, placeholder="Enter your prompt here...", label="Prompt"),
        gr.Textbox(lines=2, placeholder="Enter your negative prompt here...", label="Negative Prompt"),
        gr.Slider(1, 100, step=1, value=50, label="Number of Inference Steps"),
        gr.Slider(1, 20, step=0.1, value=9.0, label="Guidance Scale"),
        gr.Number(label="Seed", value=8888)
    ],
    outputs=gr.Video(label="Generated GIF"),
    title="I2VGen-XL GIF Generator",
    description="Generate a GIF from an image and a prompt using the I2VGen-XL model."
)

# Launch the interface
iface.launch()