Posts

Showing posts with the label latent-diffusion-model

Disentangled Motion Representation: Encoding Full-Body Avatars into Discrete Latent Spaces