LATO: 3D Mesh Flow Matching with Structured TOpology Preserving LAtents

TL;DR

LATO introduces a topology-preserving latent representation for 3D meshes, enabling scalable, efficient, and accurate mesh synthesis through flow matching without heuristic meshing or isosurface extraction.

Contribution

It proposes a novel voxel-based latent representation and a two-stage flow matching process for topology-aware 3D mesh generation, improving efficiency and mesh quality.

Findings

Generates complex 3D meshes with accurate topology.

Achieves high inference efficiency compared to prior methods.

Successfully synthesizes meshes without heuristic meshing.

Abstract

In this paper, we introduce LATO, a novel topology-preserving latent representation that enables scalable, flow matching-based synthesis of explicit 3D meshes. LATO represents a mesh as a Vertex Displacement Field (VDF) anchored on surface, incorporating a sparse voxel Variational Autoencoder (VAE) to compress this explicit signal into a structured, topology-aware voxel latent. To decapsulate the mesh, the VAE decoder progressively subdivides and prunes latent voxels to instantiate precise vertex locations. In the end, a dedicated connection head queries the voxel latent to predict edge connectivity between vertex pairs directly, allowing mesh topology to be recovered without isosurface extraction or heuristic meshing. For generative modeling, LATO adopts a two-stage flow matching process, first synthesizing the structure voxels and subsequently refining the voxel-wise topology features. Compared to prior isosurface/triangle-based diffusion models and autoregressive generation approaches, LATO generates meshes with complex geometry, well-formed topology while being highly efficient in inference.