LATO.2

<div align="center"> <h1>LATO.2: Factorized 3D Mesh Generation with Vertex and Topology Flow</h1> <p> <a href="https://github.com/LoHhhha">Hang Long</a>^1,2,* &nbsp; <a href="https://tianhaozhao668.github.io">Tianhao Zhao</a>^1,2,* &nbsp; <a href="https://github.com/MayMhappy">Junkai Lin</a>^1,2 &nbsp; <a href="https://youjiazhang.github.io">Youjia Zhang</a>^1,2 &nbsp; Huipeng Guo¹<br> <a href="https://github.com/PENGUINLIONG">Rendong Liang</a>² &nbsp; <a href="https://bluestyle97.github.io">Jiale Xu</a>² &nbsp; Jozef Hladký³ &nbsp; <a href="https://niessnerlab.org">Matthias Nießner</a>⁴ &nbsp; <a href="https://yuanming.taichi.graphics">Yuanming Hu</a>² &nbsp; <a href="https://weiyang-hust.github.io">Wei Yang</a>^1,† </p> <p> ¹Huazhong University of Science and Technology &nbsp;·&nbsp; ²Meshy AI &nbsp;·&nbsp; ³Independent Researcher &nbsp;·&nbsp; ⁴Technical University of Munich </p> <p>^*Equal contribution &nbsp;·&nbsp; ^†Corresponding author</p> <p><em>This work was completed during internships at Meshy AI.</em></p> <p> <a href="https://arxiv.org/abs/2607.10623"> <img src="https://img.shields.io/badge/arXiv-b31b1b.svg?logo=arXiv&logoColor=white" alt="arXiv"> </a> <a href="https://github.com/LoHhhha/LATO.2"> <img src="https://img.shields.io/badge/GitHub-%23121011.svg?logo=github&logoColor=white" alt="GitHub"> </a> </p> </div>

Abstract

Flow matching over carefully designed latent representations has recently emerged as a powerful paradigm for topology-aware mesh generation. Existing approaches, however, model vertices and connectivity jointly in a joint latent space, entangling continuous vertex geometry with discrete combinatorial structure; this complicates flow learning and manifests as drifting vertices and broken surfaces. We present LATO.2, a factorized flow matching framework that decomposes mesh generation into a vertex flow followed by a connectivity flow conditioned on the realized vertices, with both stages anchored to a shared coarse voxel scaffold. Dedicated VAEs underpin the two stages, recovering vertices at sub-voxel precision and embedding discrete connectivity into a continuous latent space. We demonstrate two advantages unique to this factorization: (i) part-wise generation, in which the scaffold is partitioned and each part synthesized at full latent capacity, yielding substantially higher-resolution meshes than a monolithic latent permits; and (ii) topology-adaptive editing, in which manipulating first-stage vertices induces the corresponding connectivity without re-optimization. Experiments show that LATO.2 surpasses state-of-the-art topology-aware mesh generators in geometric fidelity and connectivity quality.