Reduction of Animated Models for Embedded Devices

Abstract

We present a new supplementary method for reduction of animated 3D polygonal models. The method is applicable mainly in animation of human faces and it is based on intelligent merging of visemes represented by key polygonal meshes. It is useful for devices with limited CPU and memory resources like mobile phones or other embedded devices. Using this approach we can reduce operation memory needs and time to load the model from storage. We describe the algorithm for viseme merging and we prove that our method is optimal for selected metrics. Finally we validate method performance on an example and compare with the case when only traditional methods for 3D models reduction are used.