SAniHead [Paper][Video][Poster]

SAniHead: Sketching Animal-like 3D Character Heads Using a View-surface Collaborative Mesh Generative Network
Dong Du, Xiaoguang Han, Hongbo Fu, Feiyang Wu, Yizhou Yu, Shuguang Cui, Ligang Liu
In IEEE Transactions on Visualization and Computer Graphics (TVCG), 2020.

---

Install

sanihead_end2end

This implementation has been tested on Ubuntu 18.04, using Python 2.7.12, TensorFlow-gpu 1.13.2, TFLearn 0.5.0, CUDA 10.0, etc.

To train the code, please install several external libraries.

1. Chamfer Distance
   Please make and replace the "cd_dist_so.so" file for the calculation of Chamfer distance. The source code and introduction can be found from Pixel2Mesh.

2. DIRT
   Comparied with the source code of DIRT, we add the orthogonal projection function for our mesh rendering. The modified code can be found in dirt. Please refer to DIRT to install it.

The dataset can be found in this link. We also provide pre-trained models for the fine-tuning or testing. You can download them and put them in the directory of "sanihead_end2end".

ui

We design the user interface for real-time sketch-based animal-head modeling. It is built on Ubuntu 18.04, using Qt Creator 5.9.5. The ependencies are also given. To run the demo, a GPU with more than 8 GB memory is required.

---

Citation

If you find our work helpful, please consider citing

@article{du2020sanihead,
  title={SAniHead: Sketching Animal-like 3D Character Heads Using a View-surface Collaborative Mesh Generative Network},
  author={Du, Dong and Han, Xiaoguang and Fu, Hongbo and Wu, Feiyang and Yu, Yizhou and Cui, Shuguang and Liu, Ligang},
  journal={IEEE Transactions on Visualization and Computer Graphics},
  year={2020},
  publisher={IEEE}
}

---

License

SAniHead is relased under the MIT License. See the LICENSE file for more details.