GitHub - MeiGen-AI/InfiniteTalk: ​​Unlimited-length talking video generation​​ that supports image-to-video and video-to-video generation

InfiniteTalk is a model for unlimited-length talking video generation, supporting both image-to-video and video-to-video synthesis driven by audio. It offers accurate lip synchronization along with aligned head movements, body posture, and facial expressions, enabling stable, identity-preserving long video dubbing. The project continuously evolves with updates like LongCat-Video-Avatar v1.5, improving lip sync accuracy and supporting diverse styles, and provides community integration and tools for efficient inference. Installation and usage guidelines are provided along with model weights and demos for easy experimentation.