TANGO: Text-driven Photorealistic and Robust 3D Stylization via Lighting Decomposition

Yongwei Chen, Rui Chen, Jiabao Lei, Yabin Zhang, Kui Jia

NeurIPS 2022 Spotlight

Paper | Project Page

News

:trophy: [Nov 11th 2022] TANGO was selected in NeurIPS spotlight papers.

Installation

Note: You can directly pull the image we uploaded to AliCloud

docker pull registry.cn-shenzhen.aliyuncs.com/baopin/t2m:1.7

System Requirements

• Python >=3.7 and <=3.9
• CUDA 11
• Nvidia GPU with 12 GB ram at least
• Open3d >=0.14.1
• the package of clip (https://github.com/openai/CLIP)

Train

Call the below shell scripts to generate example styles.

# shoe made of gold
./demo/run_shoe_gold.sh
# vase made of wicker 
./demo/run_vase_wicker.sh
# car made of wood
./demo/run_car_wood_origin.sh
# ...

The outputs will be saved to results/demo

Validate

Call the below shell scripts to generate gif.

# shoe made of gold
./demo/test_shoe_gold.sh
# vase made of wicker 
./demo/test_vase_wicker.sh
# car made of wood
./demo/test_car_wood_origin.sh
# ...

Citation

@article{chen2022tango,
  title={Tango: Text-driven photorealistic and robust 3d stylization via lighting decomposition},
  author={Chen, Yongwei and Chen, Rui and Lei, Jiabao and Zhang, Yabin and Jia, Kui},
  journal={Advances in Neural Information Processing Systems},
  volume={35},
  pages={30923--30936},
  year={2022}
}