Meshy T2: Fast Native Mesh Generation with Flow Matching

2026-07-28

AI Summary

Existing autoregressive methods for 3D mesh generation suffer from slow inference and error accumulation, making them ill-suited for real-time creation of high-quality meshes with artist-friendly topology. This work proposes the first native mesh generation framework based on flow matching: it employs a vertex-set mesh VAE to encode meshes into continuous latent representations and introduces a two-stage cascaded flow matching model that enables single-pass, parallel decoding from image to complete mesh. The approach eliminates the need for vertex quantization and welding, supports interactive generation (median time of 6 seconds), explicit control over face count, and multi-part asset modeling. It significantly outperforms current autoregressive baselines in both geometric fidelity and generation efficiency, achieving speedups of over an order of magnitude.

Abstract

Polygonal meshes are the standard surface representation of modern 3D pipelines, and generating high-quality meshes with artist-style topology is essential for film, gaming, and interactive 3D applications. Mainstream approaches serialize a mesh into a token sequence and decode it autoregressively, which is slow at inference and sensitive to error accumulation, making them impractical for interactive asset creation. We present Meshy T2, a fast native mesh generation framework built on flow matching. At its core is a vertex-set mesh VAE that encodes a mesh into one continuous latent token per vertex and decodes vertices, edge connectivity, and face winding order in a single pass, preserving high-precision geometry and artist-authored topology without vertex quantization or welding. Generation proceeds as a coarse-to-fine cascade of two flow-matching models: an image-conditioned voxel flow first sketches the overall shape as a coarse occupancy scaffold, and a mesh flow then populates the scaffold with per-vertex latent tokens, conditioned on the image, the scaffold, and a requested vertex budget. This design delivers three practical capabilities: interactive generation speed through parallel flow-based synthesis; effective face-count control through the requested vertex budget; and native support for multi-part assets, whose components emerge directly from the generated connectivity. In our experiments, Meshy T2 achieves state-of-the-art geometric fidelity and completes end-to-end image-to-mesh generation within a median of 6 seconds, over an order of magnitude faster than autoregressive baselines. Code and weights will be available at https://github.com/meshy-dev/meshy-t2.

Problem

• mesh generation
• autoregressive decoding
• interactive 3D
• artist-style topology
• inference speed

Innovation

• flow matching
• native mesh generation
• vertex-set VAE
• coarse-to-fine cascade
• interactive 3D generation