MeshWeaver: Sparse-Voxel-Guided Surface Weaving for Autoregressive Mesh Generation

2026-06-03

AI Summary

Existing autoregressive mesh generation methods struggle to scale to high-face-count models due to inefficient tokenization and excessively long sequences, while also lacking explicit local geometric guidance. This work proposes a “surface weaving” mechanism that formulates mesh generation as a vertex-by-vertex prediction process. It introduces a multi-level sparse voxel encoder to provide rich geometric context, integrating voxel-feature cross-attention and structural scaffold constraints to incorporate local geometric priors into the autoregressive framework for the first time at the vertex level. The approach substantially improves both generation efficiency and geometric fidelity, enabling synthesis of meshes with up to 16K faces and achieving an 18% state-of-the-art compression rate.

Abstract

Autoregressive mesh generation has gained attention by tokenizing meshes into sequences and training models in a language-modeling fashion. However, existing approaches suffer from two fundamental limitations: (i) low tokenization efficiency, which yields long token sequences and prevents scaling to high-poly meshes, and (ii) absence of geometry-aware guidance, as generation is conditioned only on global shape embeddings rather than local surface cues. We introduce MeshWeaver, an autoregressive framework that treats mesh generation as a surface weaving process by directly predicting the next vertex instead of independent coordinates. At its core is a multi-level sparse-voxel encoder that injects geometric context into the generative process in three complementary ways: providing voxel features as vertex representations, guiding token prediction via cross-attention to voxel features, and serving as a structural scaffold that constrains generation around the input surface. Our hierarchical design enables coarse-to-fine vertex prediction in a single decoding step, while tightly coupling the generative model with 3D geometry. Extensive experiments demonstrate that MeshWeaver achieves a state-of-the-art compression ratio of 18%, can generate meshes with up to 16K faces, and significantly improves geometric fidelity over prior approaches.

Problem

• autoregressive mesh generation
• tokenization efficiency
• geometry-aware guidance
• high-poly meshes
• local surface cues

Innovation

• autoregressive mesh generation
• sparse-voxel encoding
• surface weaving
• geometry-aware guidance
• coarse-to-fine prediction