Learned Localized Mesh Refinement

2026-09-25

AI Summary

This study addresses the inefficiency of computational resource allocation in existing adaptive mesh refinement methods, which lack semantic awareness and region-specific control. We propose a neural adaptive triangular mesh refinement approach based on a locally conditioned autoregressive architecture. A key contribution is a novel mesh tokenizer that enables multi-trajectory upsampling generation, facilitating precise local manipulation of both topology and geometry. Experimental results demonstrate that the method preserves global structural integrity while introducing geometric details to user-selected regions. Furthermore, it supports view-dependent refinement and physics-aware optimization, overcoming the limitations of global generation approaches that lack regional controllability. This work significantly enhances the flexibility and efficiency of mesh generation.

Abstract

We present a neural method for adaptive triangle mesh refinement, in which an autoregressive model adds geometric detail to selected regions of an input mesh while leaving the rest unchanged, a key capability for efficiently allocating mesh budget. Existing upsampling methods struggle to achieve this. Classical subdivision schemes refine triangulation without semantic awareness of the underlying shape or the ability to recover geometric details missing from a coarse input. Recent neural mesh models generate shapes globally, sacrificing region-specific control. We propose a novel tokenizer that yields combinatorially many valid upsampling trajectories from a single mesh. Trained on such data, our locally-conditioned autoregressive architecture allows for direct manipulation of topology and geometry within target regions of an input mesh. We validate our method against state-of-the-art approaches and demonstrate its ability to perform adaptive upsampling with region-selective control, a capability absent from existing approaches. This unlocks inference-time view-dependent refinement, physics-aware region refinement, and coarse-shape conditioned novel mesh synthesis. We provide code at https://github.com/seanxzhan/learned-localized-mesh-refinement/.

Problem

• Mesh Refinement
• Adaptive Upsampling
• Local Control
• Triangle Mesh
• Geometric Detail

Innovation

• Mesh Refinement
• Autoregressive Model
• Adaptive Upsampling
• Tokenizer
• Region-selective Control