Uni4D: Unifying Visual Foundation Models for 4D Modeling from

uni4dUni4D demonstrates superior visual quality, temporal consistency, and geometric accuracy across a variety of challenging scenes across various datasets. Quantative evaluation throughUni4D addresses key challenges in existing methods, effectively modeling long-term motion representations in a latent space while preserving geometric information in 4D space.