PGN-Cloth: Physics-based graph network model for 3D cloth animation

Tao Peng, Wenjie Wu, Junping Liu, Li Li, Jiazhe Miao, Xinrong Hu, Ruhan He, Lijun Li

Abstract: Highlights•Physics-based deep learning can constrain model training to greatly improve training speed.•The introduction of physical formula constraints can generate more reasonable results.•The fabric structure is well suited to the use of mesh representation and suitable for learning motion features using graph neural networks.•World-Edge model can handle collision problems well.