HunyuanCustom Brings Single-Image Video Deepfakes, With Audio and Lip Sync
SMRTR summary
Tencent has released HunyuanCustom, a new version of its Hunyuan Video model for customized video generation. The system can create deepfake-style videos from a single image, potentially making Hunyuan LoRA models obsolete. HunyuanCustom offers improved identity consistency, audio-driven animation, and video editing capabilities. It outperforms many competing systems in tests, including commercial options like Kling. The release includes two model versions and open-source code, though high GPU requirements may limit immediate accessibility for many users.
SMRTR provides this summary for quick context. The original article belongs to Unite AI.
Read the original article