Data-driven Garment Pattern Estimation from 3D Geometries

Abstract

Three-dimensional scanning technology recently becomes widely available to the public. However, it is difficult to simulate clothing deformation from the scanned people because scanned data lacks information required for the clothing simulation. In this paper, we present a technique to estimate clothing patterns from a scanned person in cloth. Our technique uses image-based deep learning to estimate the type of pattern on the projected image. The key contribution is converting image-based inference into three-dimensional clothing pattern estimation. We evaluate our technique by applying our technique to an actual scan.