Hallo3 is an open-source portrait animation model by Fudan Vision Lab using Diffusion Transformer Networks to generate realistic talking head videos from photos and audio.
