HunyuanCustom is a multimodal video customization framework by Tencent Hunyuan, aimed at generating customized videos featuring particular subjects (people, characters) under flexible conditions, while maintaining subject/identity consistency. It supports conditioning via image, audio, video, and text, and can perform subject replacement in videos, generate avatars speaking given audio, or combine multiple subject images. The architecture builds on HunyuanVideo, with added modules for identity reinforcement and modality-specific condition injection. Text-image fusion module based on LLaVA for improved multimodal understanding. Applicable to single- and multi-subject scenarios, video editing/replacement, singing avatars etc.
Features
- Supports multimodal conditions: text, image, audio, and video input modalities
- Identity / subject consistency modules (e.g. image ID enhancement, temporal concatenation) across frames
- Text-image fusion module based on LLaVA for improved multimodal understanding
- AudioNet module for hierarchical alignment of audio conditions
- Video-driven condition injection via patchify-based feature alignment network
- Applicable to single- and multi-subject scenarios, video editing / replacement, singing avatars etc.
Project Samples
