Yiwen Chen1,2*,
Tong He2†,
Di Huang2,
Weicai Ye2,
Sijin Chen3,
Jiaxiang Tang4
Xin Chen5,
Zhongang Cai6,
Lei Yang6,
Gang Yu7,
Guosheng Lin1†,
Chi Zhang8†
*Work done during a research internship at Shanghai AI Lab.
†Corresponding authors.
1S-Lab, Nanyang Technological University,
2Shanghai AI Lab,
3Fudan University,
4Peking University,
5University of Chinese Academy of Sciences,
6SenseTime Research,
7Stepfun,
8Westlake University
- [6/17] 🔥🔥 We released the 350m version of MeshAnything.
Our environment has been tested on Ubuntu 22, CUDA 11.8 with A100, A800 and A6000.
- Clone our repo and create conda environment
git clone https://github.com/buaacyw/MeshAnything.git && cd MeshAnything
conda create -n MeshAnything python==3.10.13
conda activate MeshAnything
pip install torch==2.1.1 torchvision==0.16.1 torchaudio==2.1.1 --index-url https://download.pytorch.org/whl/cu118
pip install -r requirements.txt
pip install flash-attn --no-build-isolation
or
pip install git+https://github.com/buaacyw/MeshAnything.gitAnd directly use in your code as
import MeshAnything
python app.py
# folder input
python main.py --input_dir examples --out_dir mesh_output --input_type mesh
# single file input
python main.py --input_path examples/wand.ply --out_dir mesh_output --input_type mesh
# Preprocess with Marching Cubes first
python main.py --input_dir examples --out_dir mesh_output --input_type mesh --mc
# 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.
# inference for folder
python main.py --input_dir pc_examples --out_dir pc_output --input_type pc_normal
# inference for single file
python main.py --input_path pc_examples/mouse.npy --out_dir pc_output --input_type pc_normal
- It takes about 7GB and 30s to generate a mesh on an A6000 GPU.
- The input mesh will be normalized to a unit bounding box. The up vector of the input mesh should be +Y for better results.
- Limited by computational resources, MeshAnything is trained on meshes with fewer than 800 faces and cannot generate meshes with more than 800 faces. The shape of the input mesh should be sharp enough; otherwise, it will be challenging to represent it with only 800 faces. Thus, feed-forward 3D generation methods may often produce bad results due to insufficient shape quality. We suggest using results from 3D reconstruction, scanning and SDS-based method (like DreamCraft3D) as the input of MeshAnything.
- Please refer to https://huggingface.co/spaces/Yiwen-ntu/MeshAnything/tree/main/examples for more examples.
The repo is still being under construction, thanks for your patience.
- Release of training code.
- Release of larger model.
Our code is based on these wonderful repos:
@misc{chen2024meshanything,
title={MeshAnything: Artist-Created Mesh Generation with Autoregressive Transformers},
author={Yiwen Chen and Tong He and Di Huang and Weicai Ye and Sijin Chen and Jiaxiang Tang and Xin Chen and Zhongang Cai and Lei Yang and Gang Yu and Guosheng Lin and Chi Zhang},
year={2024},
eprint={2406.10163},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent