Character.AI launches AvatarFX video generation model: bring still image characters to life with speech.

Character.AI has launched the AvatarFX model, which enables characters in static images to “talk.” By uploading an image and selecting a voice, users can generate animated characters that speak and express emotions. The model is based on a state-of-the-art DiT-based diffusion video generation technology, offering high fidelity and temporal consistency. It remains stable even in multi-character, long-sequence dialogues.