FACE: A Face-based Autoregressive Representation for High-Fidelity and Efficient Mesh Generation

TL;DR

FACE introduces a face-level autoregressive framework for 3D mesh generation, significantly reducing sequence length and computational costs while maintaining high fidelity and quality, enabling efficient high-quality mesh synthesis.

Contribution

The paper proposes FACE, a face-based autoregressive autoencoder that generates meshes at the face level, achieving unprecedented compression and efficiency without sacrificing quality.

Findings

Reduces sequence length by a factor of nine.

Halves the previous state-of-the-art compression ratio.

Achieves state-of-the-art reconstruction quality on benchmarks.

Abstract

Autoregressive models for 3D mesh generation suffer from a fundamental limitation: they flatten meshes into long vertex-coordinate sequences. This results in prohibitive computational costs, hindering the efficient synthesis of high-fidelity geometry. We argue this bottleneck stems from operating at the wrong semantic level. We introduce FACE, a novel Autoregressive Autoencoder (ARAE) framework that reconceptualizes the task by generating meshes at the face level. Our one-face-one-token strategy treats each triangle face, the fundamental building block of a mesh, as a single, unified token. This simple yet powerful design reduces the sequence length by a factor of nine, leading to an unprecedented compression ratio of 0.11, halving the previous state-of-the-art. This dramatic efficiency gain does not compromise quality; by pairing our face-level decoder with a powerful VecSet encoder, FACE achieves state-of-the-art reconstruction quality on standard benchmarks. The versatility of the learned latent space is further demonstrated by training a latent diffusion model that achieves high-fidelity, single-image-to-mesh generation. FACE provides a simple, scalable, and powerful paradigm that lowers the barrier to high-quality structured 3D content creation.