Subspace Clothing Simulation Using Adaptive Bases

摘要

We present a new approach to clothing simulation using lowdimensionalrnlinear subspaces with temporally adaptive bases. Ourrnmethod exploits full-space simulation training data in order tornconstruct a pool of low-dimensional bases distributed across posernspace. For this purpose, we interpret the simulation data as offsetsrnfrom a kinematic deformation model that captures the global shapernof clothing due to body pose. During subspace simulation, we selectrnlow-dimensional sets of basis vectors according to the currentrnpose of the character and the state of its clothing. Thanks to thisrnadaptive basis selection scheme, our method is able to reproducerndiverse and detailed folding patterns with only a few basis vectors.rnOur experiments demonstrate the feasibility of subspace clothingrnsimulation and indicate its potential in terms of quality and computationalrnefficiency.