FreeMesh: Boosting Mesh Generation with Coordinates Merging

2025-05-19

AI Summary

Existing autoregressive mesh generation methods lack a training-free, interpretable evaluation metric for mesh tokenizers, hindering compression efficiency optimization. To address this, we propose Per-Token Mesh Entropy (PTME), the first theoretical framework enabling training-free quantification of tokenizer performance. We further introduce a plug-and-play coordinate-merging technique that reconstructs tokenization structures via sequence reordering and lossless fusion of high-frequency coordinate patterns. Our approach combines information-theoretic entropy analysis with statistical modeling of coordinate distributions, ensuring compatibility with mainstream mesh tokenizers—including MeshXL, MeshAnything V2, and Edgerunner. Experiments demonstrate significant compression ratio improvements across multiple tokenizers, while PTME exhibits strong correlation with actual generation quality. This work establishes a novel, interpretable, and optimization-friendly paradigm for mesh serialization.

Technology Category

• Natural Language Processing: Generation
• Machine Learning: Learning on the Edge & Model Compression
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

The next-coordinate prediction paradigm has emerged as the de facto standard in current auto-regressive mesh generation methods. Despite their effectiveness, there is no efficient measurement for the various tokenizers that serialize meshes into sequences. In this paper, we introduce a new metric Per-Token-Mesh-Entropy (PTME) to evaluate the existing mesh tokenizers theoretically without any training. Building upon PTME, we propose a plug-and-play tokenization technique called coordinate merging. It further improves the compression ratios of existing tokenizers by rearranging and merging the most frequent patterns of coordinates. Through experiments on various tokenization methods like MeshXL, MeshAnything V2, and Edgerunner, we further validate the performance of our method. We hope that the proposed PTME and coordinate merging can enhance the existing mesh tokenizers and guide the further development of native mesh generation.

Problem

• Evaluating mesh tokenizers without training using PTME metric
• Improving compression ratios via coordinate merging technique
• Validating method performance across existing tokenization approaches

Innovation

• Introduces Per-Token-Mesh-Entropy (PTME) metric
• Proposes plug-and-play coordinate merging technique
• Improves compression ratios of mesh tokenizers