CLOTH3D: Clothed 3D Humans
This provides a foundational resource for researchers in computer vision and graphics working on 3D human modeling and animation.
The authors tackled the lack of large-scale synthetic datasets for 3D clothed humans by creating CLOTH3D, which includes diverse garments and realistic dynamics, and they developed a generative model that produces 3D garments on SMPL models for any pose and shape.
This work presents CLOTH3D, the first big scale synthetic dataset of 3D clothed human sequences. CLOTH3D contains a large variability on garment type, topology, shape, size, tightness and fabric. Clothes are simulated on top of thousands of different pose sequences and body shapes, generating realistic cloth dynamics. We provide the dataset with a generative model for cloth generation. We propose a Conditional Variational Auto-Encoder (CVAE) based on graph convolutions (GCVAE) to learn garment latent spaces. This allows for realistic generation of 3D garments on top of SMPL model for any pose and shape.