DeepMesh: Auto-Regressive Artist-Mesh Creation
With Reinforcement Learning

Ruowen Zhao^1,2*, Junliang Ye^1,2*, Zhengyi Wang^1,2*,
Guangce Liu², Yiwen Chen³, Yikai Wang¹, Jun Zhu^1,2†
*Equal Contribution.
^†Corresponding authors.
¹Tsinghua University, ²ShengShu,
³S-Lab, Nanyang Technological University,

                   

All of the meshes above are generated by DeepMesh. DeepMesh can generate high-quality meshes conditioned on the given point cloud by auto-regressive transformer.

Release

• [3/20] 🔥🔥We released the pretrained weight of DeepMesh (0.5 B).

Contents

• Release
• Installation
• Usage
• Important Notes
• Todo
• Acknowledgement
• BibTeX

Installation

Our environment has been tested on Ubuntu 22, CUDA 11.8 with A100, A800 and A6000.

1. Clone our repo and create conda environment

git clone https://github.com/zhaorw02/DeepMesh.git && cd DeepMesh
conda env create -f environment.yaml
conda activate deepmesh

2. Install the pretrained model weight

pip install -U "huggingface_hub[cli]"
huggingface-cli login
huggingface-cli download zzzrw/DeepMesh --local-dir ./

Usage

Command line inference

# Note: if you want to use your own point cloud, please make sure the normal is included.
# The file format should be a .npy file with shape (N, 6), where N is the number of points. The first 3 columns are the coordinates, and the last 3 columns are the normal.

# Generate all obj/ply in your folder
CUDA_VISIBLE_DEVICES=0 torchrun --nproc-per-node=1 --master_port=12345 sample.py \
    --model_path "your_model_path" \
    --steps 90000 \
    --input_path examples \
    --output_path mesh_output \
    --repeat_num 4 \
    --temperature 0.5 \

# Generate the specified obj/ply in your folder
CUDA_VISIBLE_DEVICES=0 torchrun --nproc-per-node=1 --master_port=22345.py \
    --model_path "your_model_path" \
    --steps 90000 \
    --input_path examples \
    --output_path mesh_output \
    --repeat_num 4 \
    --uid_list "wand1.obj,wand2.obj,wand3.ply" \
    --temperature 0.5 \

# Or
bash sample.sh

Important Notes

• Please refer to our project_page for more examples.

Todo

• Release of pre-training code ( truncted sliding training ).
• Release of post-training code ( DPO ).
• Release of larger model ( 1b version ).

Acknowledgement

Our code is based on these wonderful repos:

• BPT
• LLaMA-Mesh
• SMDM
• Meshanything V2
• Michelangelo

BibTeX

@article{zhao2025deepmesh,
  title={DeepMesh: Auto-Regressive Artist-mesh Creation with Reinforcement Learning},
  author={Zhao, Ruowen and Ye, Junliang and Wang, Zhengyi and Liu, Guangce and Chen, Yiwen and Wang, Yikai and Zhu, Jun},
  journal={arXiv preprint arXiv:2503.15265},
  year={2025}
}