Topology-independent 3D garment fitting for virtual clothing

Mingmin Zhang, Ling Lin, Zhigeng Pan, Nan Xiang

Abstract: Using computer-aided design system to design an elegant 3D garment for a virtual human is often tedious and labor-intensive. Moreover, the garment is usually designed for a reference human model and generally not fitted to other individuals, which largely reduces the reusability of existing 3D garments. In this paper, we introduce proxy mesh to fit 3D garment to another human model whose topology or shape is different from the garment’s reference human model. Firstly, a proxy mesh is generated for the reference human model and the specified human model respectively. Secondly, the garment is parameterized based on the proxy mesh of the reference model and an independent dataset is obtained. Thirdly, the dataset is decoded to the proxy mesh of the other human model and a roughly fitted garment is gained. Lastly, local shape constrains are enforced to the fitted garment and garment-body penetrations are resolved to get a well fitted garment. Our approach is efficient, simple to implement and is potential to be applied to existing applications such as virtual try-on and virtual clothing design.