FACE: A Face-based Autoregressive Representation for High-Fidelity and Efficient Mesh Generation
Hanxiao Wang, Yuan-Chen Guo, Ying-Tian Liu, Zi-Xin Zou, Biao Zhang, Weize Quan, Ding Liang, Yan-Pei Cao, Dong-Ming Yan · 2026
Read FACE: A Face-based Autoregressive Representation for High-Fidelity and Efficient Mesh Generation by Hanxiao Wang, Yuan-Chen Guo, Ying-Tian Liu, Zi-Xin Zou, Biao Zhang, Weize Quan, Ding Liang, Yan-Pei Cao, Dong-Ming Yan (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-01515.txt.
Description
Building a 3D mesh one point at a time makes long lists that cost a lot to generate. Hanxiao Wang, Yuan-Chen Guo, Ying-Tian Liu, and colleagues rethink that order in this 2026 work. They argue the flaw is the unit: a mesh is made of faces, not lone points.
Their system, FACE, treats each triangle face as a single token for the generator. That one choice cuts the list length by a factor of nine and beats prior compression by half. A strong encoder reads point clouds into a compact code that the face decoder then expands. Tests show top rebuild quality on standard sets, and a diffusion model on the same code turns one image into a detailed mesh. The result lowers the cost of quality 3D content.
Catalog record
| CREATOR | Hanxiao Wang, Yuan-Chen Guo, Ying-Tian Liu, Zi-Xin Zou, Biao Zhang, Weize Quan, Ding Liang, Yan-Pei Cao, Dong-Ming Yan |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 33035 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.01515