Dance Synthesis

Dance synthesis focuses on generating realistic and expressive 3D dance movements, often conditioned on music and sometimes lyrics or even onomatopoeia. Current research emphasizes developing generative models, particularly diffusion models and transformers, to achieve high-fidelity motion, temporal synchronization with audio, and diverse, controllable outputs, including multi-dancer scenarios and camera movement synthesis. This field is advancing both the understanding of human movement and the creation of realistic virtual characters for applications in entertainment, gaming, and virtual reality.