Search papers, labs, and topics across Lattice.
1
0
3
13
Training 3D avatar diffusion models on millions of in-the-wild videos is now possible, thanks to a clever 3D tokenization and visibility-aware training strategy that overcomes partial observability.
