1 paper
Ziheng Ouyang, Zeqiang Lai, Jiarui Chen +7
Recent single-stage 3D generative models commonly adopt VecSet representations, encoding 3D shapes as unordered sets of latent tokens. However, compared with two-stage methods that…