Uni4D

uni4dTitle: Uni4D: Unifying Visual Foundation Models for 4D Modeling from a Single Video; Authors: David Yifan Yao, Albert J. Zhai, Shenlong Wang, Abstract summary: We introduce Uni4D, aUni4D addresses key challenges in existing methods, effectively modeling long-term motion representations in a latent space while preserving geometric information in 4D space.