Efficient Representation of Layered Depth Images for Real-time Volumetric Tests

Abstract

Representing Layered Depth Images (LDI) as 3D texture can be used to approximate complex, arbitrarily shaped volumes on graphics hardware. Based on this concept, a number of real-time applications such as collision detection or 3D clipping against multiple volumes can be implemented efficiently using programmable hardware. One major drawback of this image-based representation is the high video memory consumption. To compensate that disadvantage, this paper presents a concept and associated algorithms that enable a lossless, efficient LDI representation which is especially designed for the usage within shader programs. The concept comprises the application of a viewpoint selection, a cropping, and a compression algorithm. We evaluated our algorithm with different test cases and show possible use cases.