ID-Animator is a zero-shot human video generation method that can generate personalized videos based on a single reference facial image without further training. This technology inherits the existing diffusion-based video generation framework and incorporates a face adapter to encode identity-related embeddings. Through this approach, ID-Animator maintains the details of the subject's identity during the video generation process while improving training efficiency.