README-HF.md

FramePack - Image-to-Video Generation

An AI application that converts static images into dynamic videos. Upload a character image, add a motion description, and generate smooth videos!

How to Use

1. Upload a character image
2. Enter a prompt describing the desired motion (e.g., "The girl dances gracefully")
3. Adjust video length and other optional parameters
4. Click the "Start Generation" button
5. Wait for the video to generate (the process is progressive, continuously extending the video length)

Example Prompts

• "The girl dances gracefully, with clear movements, full of charm."
• "The man dances energetically, leaping mid-air with fluid arm swings and quick footwork."
• "A character doing some simple body movements."

Technical Features

• Based on Hunyuan Video and FramePack architecture
• Supports low-memory GPU operation
• Can generate videos up to 120 seconds long
• Uses TeaCache technology to accelerate the generation process

Notes

• Video generation is done in reverse order, with the ending motion generated before the starting motion
• The first use requires downloading the model (approximately 30GB), please be patient
• If you encounter an out-of-memory error, increase the value of "GPU inference reserved memory"

---

Original Project: FramePack GitHub