HiFi-Mesh: High-Fidelity Efficient 3D Mesh Generation via Compact Autoregressive Dependence

TL;DR

This paper introduces LANE, a novel autoregressive model for 3D mesh generation that significantly improves sequence length and inference speed, enabling high-quality, detailed 3D mesh synthesis.

Contribution

The paper proposes LANE with compact autoregressive dependencies and AdaGraph for efficient inference, advancing 3D mesh generation capabilities.

Findings

6x increase in maximum sequence length

Faster inference with AdaGraph strategy

Improved structural detail and geometric consistency

Abstract

High-fidelity 3D meshes can be tokenized into one-dimension (1D) sequences and directly modeled using autoregressive approaches for faces and vertices. However, existing methods suffer from insufficient resource utilization, resulting in slow inference and the ability to handle only small-scale sequences, which severely constrains the expressible structural details. We introduce the Latent Autoregressive Network (LANE), which incorporates compact autoregressive dependencies in the generation process, achieving a $6\times$ improvement in maximum generatable sequence length compared to existing methods. To further accelerate inference, we propose the Adaptive Computation Graph Reconfiguration (AdaGraph) strategy, which effectively overcomes the efficiency bottleneck of traditional serial inference through spatiotemporal decoupling in the generation process. Experimental validation demonstrates that LANE achieves superior performance across generation speed, structural detail, and geometric consistency, providing an effective solution for high-quality 3D mesh generation.