Geometry Videos: A New Representation for 3D Animations

Abstract

We present the “Geometry Video,” a new data structure to encode animated meshes. Being able to encode animated meshes in a generic source-independent format allows people to share experiences. Changing the viewpoint allows more interaction than the fixed view supported by 2D video. Geometry videos are based on the “Geometry Image” mesh representation introduced by Gu et al. 4. Our novel data structure provides a way to treat an animated mesh as a video sequence (i.e., 3D image) and is well suited for network streaming. This representation also offers the possibility of applying and adapting existing mature video processing and compression techniques (such as MPEG encoding) to animated meshes. This paper describes an algorithm to generate geometry videos from animated meshes. The main insight of this paper, is that Geometry Videos re-sample and re-organize the geometry information, in such a way, that it becomes very compressible. They provide a unified and intuitive method for level-of-detail control, both in terms of mesh resolution (by scaling the two spatial dimensions) and of frame rate (by scaling the temporal dimension). Geometry Videos have a very uniform and regular structure. Their resource and computational requirements can be calculated exactly, hence making them also suitable for applications requiring level of service guarantees.