README.md

metadata

title: HOMA
emoji: 🎬
colorFrom: pink
colorTo: blue
sdk: gradio
sdk_version: 6.15.1
app_file: app.py
short_description: Human-object interaction video from image+audio+text
python_version: '3.10'
startup_duration_timeout: 1h

HunyuanVideo-HOMA

This Space demonstrates HOMA (Human-Object interaction in Multimodal driven human Animation), a model that generates realistic human-object interaction videos from a reference person image, an object image, speech audio, and a text prompt.

Based on the paper: HOMA: Towards Generic Human-Object Interaction in Multimodal Driven Human Animation with Weak Conditions (SIGGRAPH Asia 2025).

How to use

1. Upload a reference person image (the character to animate)
2. Upload an object image (the object to interact with)
3. Upload speech audio (driving audio for lip-sync)
4. Enter a text prompt describing the desired interaction
5. Click Generate Video

The model will produce a video of the person interacting with the object, with lip movements synchronized to the audio.