File size: 1,033 Bytes
b935e3d f404d24 36bfcb2 b935e3d 5dc1a58 36bfcb2 ec9e5f3 5dc1a58 ec9e5f3 5dc1a58 f404d24 ec9e5f3 ed0ae5f 5dc1a58 ec9e5f3 5dc1a58 ec9e5f3 ed0ae5f f404d24 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 |
import gradio as gr
from transformers import pipeline
from spaces import GPU # Import the GPU decorator for ZeroGPU
# Use @GPU to allocate a GPU when running the video generation function
@GPU
def generate_video(prompt, image):
# Load the Kandinsky video model
video_model = pipeline("video-generation", model="ai-forever/KandinskyVideo_1_1", device=0)
# Generate video based on the prompt and image
video = video_model(prompt=prompt, init_image=image)
return video # Ensure this returns the video file or frames
# Create the Gradio interface with text and image inputs
interface = gr.Interface(
fn=generate_video,
inputs=[
gr.Textbox(label="Enter your prompt here"),
gr.Image(label="Upload an initial image") # Image upload input
],
outputs="video",
title="AI Video Generator",
description="This app generates a short video based on your input prompt and initial image using Kandinsky 1.1.",
theme="dark"
)
if __name__ == "__main__":
interface.launch()
|