What type of study is this?

This is a Experimental Study study.

October 8, 2025Open Access

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

Key Points

DeepMesh generates high-quality triangle meshes, achieving superior precision and detail compared to prior methods.
The framework employs innovations such as an efficient pre-training strategy and a novel tokenization algorithm.
Incorporating reinforcement learning via Direct Preference Optimization aligns generated meshes with human preferences.
A robust scoring standard combines human evaluation and 3D metrics to ensure both aesthetic and geometric accuracy.

Abstract

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/

Read Full Paperexternally

Mark Helpful

Bookmark

Relay

View Full Paper