Cloth Simulation has been used widely in many industries such as fashion (design, online retail, etc.) and entertainment (games, movies, etc. ). However, realistic and fast cloth simulation is still a vastly open problem. For instance, realistic garment simulation is very difficult due to the low speed of high-fidelity simulation methods. This project aims to combine state-of-the-art theoretical research in cloth simulation, computer graphics and machine learning to target this problem. It looks into using new methods in physics-inspired machine learning for high-fidelity fabrics models. It investigates how physics and machine learning models can be combined to automatically simulate fabrics with desired properties, such as to mimic real fabrics.
Fabric Physics and Deep Learning
Project Description
Resources
1. Deshan Gong, Zhanxing Zhu, Andrew J. Bulpitt and He Wang, Fine-grained differentiable physics: a yarn-level model for fabrics, International Conference on Learning Representation 2022
Bibtex
@InProceedings{Gong_finegrained_2022,
author="Gong, Deshan
and Zhu, Zhanxing
and Andrew, Bulpitt.
and Wang, He",
title="Fine-grained differentiable physics: a yarn-level model for fabrics",
booktitle="International Conference on Learning Representations",
year="2022",
}