1 - Zhejiang University

2 - Tencent

3 - ZJU-Tencent Game and Intelligent Graphics Innovation Technology Joint Lab

Our three-stage network architecture: The static stage aims to learn a static cloth skinning model. The skeleton sequence stage and the mesh sequence stage are trained to add coarse dynamic component and wrinkle dynamic component. The serialized skeleton information and the serialized mesh information are used to add dynamics.

Abstract

We propose a three-stage network that utilizes a skinning-based model to accurately predict dynamic cloth deformation. Our approach decomposes cloth deformation into three distinct components: static, coarse dynamic, and wrinkle dynamic components. To capture these components, we train our three-stage network accordingly. In the first stage, the static component is predicted by constructing a static skinning model that incorporates learned joint increments and skinning weight increments. Then, in the second stage, the coarse dynamic component is added to the static skinning model by incorporating serialized skeleton information. Finally, in the third stage, the mesh sequence stage refines the prediction by incorporating the wrinkle dynamic component using serialized mesh information. We have implemented our network and used it in a Unity game scene, enabling real-time prediction of cloth dynamics. Our implementation achieves impressive prediction speeds of approximately 3.6ms using an NVIDIA GeForce RTX 3090 GPU and 9.6ms on on an Intel i7-7700 CPU. Compared to SOTA methods, our network excels in accurately capturing fine dynamic cloth deformations.

Results

The running results of our three-stage network in a Unity game scene. Taking into account the movements of characters, our network can predict cloth dynamic deformations in real-time.  

Contents

Yudi Li, Min Tang, Yun Yang, Ruofeng Tong, Bailin An, Shuangcai Yang, Yao Li, Qilong Kou, D-Cloth: Skinning-based Cloth Dynamic Prediction with a Three-stage Network, Computer Graphics Forum, 42(7): e14937, (Proceedings of Pacific Graphics 2023), 2023.

   @article{ncloth22,
      author = {Li, Yudi and Tang, Min and Yang, Yun and Tong, Ruofeng and An, Bailin and Yang, Shuangcai and Li, Yao and Kou, Qilong},
      title = {{D-Cloth}: Skinning-based Cloth Dynamic Prediction with a Three-stage Network},
      journal = {Computer Graphics Forum},
      volume = {42},
      number = {7},
      pages = {e14937},
      url = {https://onlinelibrary.wiley.com/doi/abs/10.1111/cgf.14937},
      year = {2023}
      }