DeepMesh: Auto-Regressive Artist-mesh Creation with Reinforcement Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhao, Ruowen, Ye, Junliang, Wang, Zhengyi, Liu, Guangce, Chen, Yiwen, Wang, Yikai, Zhu, Jun
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915205007540224
author Zhao, Ruowen
Ye, Junliang
Wang, Zhengyi
Liu, Guangce
Chen, Yiwen
Wang, Yikai
Zhu, Jun
author_facet Zhao, Ruowen
Ye, Junliang
Wang, Zhengyi
Liu, Guangce
Chen, Yiwen
Wang, Yikai
Zhu, Jun
contents Triangle meshes play a crucial role in 3D applications for efficient manipulation and rendering. While auto-regressive methods generate structured meshes by predicting discrete vertex tokens, they are often constrained by limited face counts and mesh incompleteness. To address these challenges, we propose DeepMesh, a framework that optimizes mesh generation through two key innovations: (1) an efficient pre-training strategy incorporating a novel tokenization algorithm, along with improvements in data curation and processing, and (2) the introduction of Reinforcement Learning (RL) into 3D mesh generation to achieve human preference alignment via Direct Preference Optimization (DPO). We design a scoring standard that combines human evaluation with 3D metrics to collect preference pairs for DPO, ensuring both visual appeal and geometric accuracy. Conditioned on point clouds and images, DeepMesh generates meshes with intricate details and precise topology, outperforming state-of-the-art methods in both precision and quality. Project page: https://zhaorw02.github.io/DeepMesh/
format Preprint
id arxiv_https___arxiv_org_abs_2503_15265
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle DeepMesh: Auto-Regressive Artist-mesh Creation with Reinforcement Learning
Zhao, Ruowen
Ye, Junliang
Wang, Zhengyi
Liu, Guangce
Chen, Yiwen
Wang, Yikai
Zhu, Jun
Computer Vision and Pattern Recognition
Triangle meshes play a crucial role in 3D applications for efficient manipulation and rendering. While auto-regressive methods generate structured meshes by predicting discrete vertex tokens, they are often constrained by limited face counts and mesh incompleteness. To address these challenges, we propose DeepMesh, a framework that optimizes mesh generation through two key innovations: (1) an efficient pre-training strategy incorporating a novel tokenization algorithm, along with improvements in data curation and processing, and (2) the introduction of Reinforcement Learning (RL) into 3D mesh generation to achieve human preference alignment via Direct Preference Optimization (DPO). We design a scoring standard that combines human evaluation with 3D metrics to collect preference pairs for DPO, ensuring both visual appeal and geometric accuracy. Conditioned on point clouds and images, DeepMesh generates meshes with intricate details and precise topology, outperforming state-of-the-art methods in both precision and quality. Project page: https://zhaorw02.github.io/DeepMesh/
title DeepMesh: Auto-Regressive Artist-mesh Creation with Reinforcement Learning
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2503.15265