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

2026-03-02

AI Summary

Existing autoregressive 3D mesh generation methods suffer from excessive sequence lengths and high computational costs due to flattening meshes into vertex sequences. To address this, this work proposes FACE, a novel framework that performs autoregressive modeling at the face (triangle) level using a “one-face-one-token” strategy, reducing sequence length by a factor of nine (compression ratio of 0.11) and substantially improving generation efficiency. The approach integrates a face-level autoregressive autoencoder (ARAE), a VecSet encoder, and a latent diffusion model. Evaluated on standard benchmarks, FACE achieves state-of-the-art reconstruction quality and successfully enables high-fidelity single-image-to-3D-mesh generation.

Technology Category

• Machine Learning: Deep Generative Models & Autoencoders
• Computer Vision: Diffusion Models for Vision
• Natural Language Processing: Generation

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

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.

Problem

• 3D mesh generation
• autoregressive models
• computational efficiency
• high-fidelity geometry
• sequence length

Innovation

• face-level autoregressive modeling
• mesh generation
• one-face-one-token
• latent diffusion model
• 3D reconstruction