FreeMesh: Boosting Mesh Generation with Coordinates Merging

Jian Liu, Haohan Weng, Biwen Lei, Xianghui Yang, Zibo Zhao, Zhuo Chen, Song Guo, Tao Han, Chunchao Guo
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:39789-39802, 2025.

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.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-liu25bz, title = {{F}ree{M}esh: Boosting Mesh Generation with Coordinates Merging}, author = {Liu, Jian and Weng, Haohan and Lei, Biwen and Yang, Xianghui and Zhao, Zibo and Chen, Zhuo and Guo, Song and Han, Tao and Guo, Chunchao}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {39789--39802}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/liu25bz/liu25bz.pdf}, url = {https://proceedings.mlr.press/v267/liu25bz.html}, 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.} }
Endnote
%0 Conference Paper %T FreeMesh: Boosting Mesh Generation with Coordinates Merging %A Jian Liu %A Haohan Weng %A Biwen Lei %A Xianghui Yang %A Zibo Zhao %A Zhuo Chen %A Song Guo %A Tao Han %A Chunchao Guo %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-liu25bz %I PMLR %P 39789--39802 %U https://proceedings.mlr.press/v267/liu25bz.html %V 267 %X 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.
APA
Liu, J., Weng, H., Lei, B., Yang, X., Zhao, Z., Chen, Z., Guo, S., Han, T. & Guo, C.. (2025). FreeMesh: Boosting Mesh Generation with Coordinates Merging. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:39789-39802 Available from https://proceedings.mlr.press/v267/liu25bz.html.

Related Material