Spatial-Temporal Editing for Dynamic Hair Data

Abstract

Hair plays a unique role in depicting a person's character. Currently, most hair simulation techniques require a lot of computation time, or rely on complex capture settings. Editing and reusing of existing hair model data are very important topics in computer graphics. In this paper, we present a spatialtemporal editing technique for dynamic hair data. This method can generate a longer or even infinite length sequence of hair motion according to its motion trend from a short input. Firstly, we build spatial-temporal neighborhood information about input hair data. We then initialize the output according to the input exemplar and output constraints, and optimize the output through iterative search and assignment steps. To make the method be more efficient, we select a sparse part of the hair as the guide hair to simplify the model, and interpolate a full set of hair after the synthesis. Results show that our method can deal with a variety of hairstyles and different way of motions.