TeMO: Towards Text-Driven 3D Stylization for Multi-Object Meshes

Xuying Zhang, Bowen Yin, Yuming Chen, Zheng Lin, Yuheng Li, Qibin Hou, Ming-Ming Cheng

Paper Link, accepted by CVPR 2024

TODO

Recently, I have been occupied by other projects and cannot finish this repository in time. I apologize for any inconvenience caused.
If you are in hurry, please add my wechat via zhangxuying1004 or email me via [email protected]. Thanks for your patience.

• Release of model code.
• Release of training and testing code.

🛠️ Dependencies and Installation

conda env create --file temo.yml
conda activate temo

Note1: The below installation will fail if run on something other than a CUDA GPU machine.
Note2: Installing clip by referring to link.
Note3: Installing kaolin by referring to link. If saying something like nvcc not found, you may need to set your CUDA_HOME environment variable to the 11.3 folder i.e. export CUDA_HOME=/usr/local/cuda-11.3, then retuning the installation.
Note4: Installing Open3D by pip install open3d==0.14.1.

🤖 Training and Validation

1. Train: call the below shell scripts to generate example styles:

# a fire dragon and an ice dragon
./scripts/run_dual_dragon.sh
# A garfield cat and a brown horse
./scripts/run_cat_horse.sh
# ...

3. Validate: call the below shell scripts to generate gifs:

# a fire dragon and an ice dragon
./scripts/test_dual_dragon.sh
# A garfield cat and a brown horse
./scripts/test_cat_horse.sh
# ...

📖 Citation

If our work gives some inspiration to your research, please star this project and cite us. Thank you!

@article{zhang2023temo,
  title={TeMO: Towards Text-Driven 3D Stylization for Multi-Object Meshes},
  author={Zhang, Xuying and Yin, Bo-Wen and Chen, Yuming and Lin, Zheng and Li, Yunheng and Hou, Qibin and Cheng, Ming-Ming},
  journal={arXiv preprint arXiv:2312.04248},
  year={2023}
}

🤝 Acknowledgement

This repo is mainly built based on TANGO, Text2Mesh, and X-Mesh. Thanks for their great work!