Shadow Mapping Based on Dual Depth Layers

Abstract

Shadow maps are a widely used means for the generation of shadows although they exhibit aliasing artifacts and problems of numerical precision. In this paper we extend the concept of a single shadow map by introducing dual shadow maps, which are based on the two depth layers that are closest to the light source. Our shadow algorithm takes into account these two depth values and computes an adaptive depth bias to achieve a robust determination of shadowed regions. Dual depth mapping only modifies the construction of the shadow map and can therefore be combined with other extensions such as filtering, perspective shadow maps, or adaptive shadow maps. Our approach can be mapped to graphics hardware for interactive applications and can also be used in high-quality software renderers.