Irregular Adaptive Shadow Maps

Abstract

Shadow mapping is a common approach to visualizing shadows in real-time graphics applications such as games. Given the performant nature of these applications, often limited resources are reserved for this task despite shadows providing a strong and important visual cue in computer generated graphics. With this work we introduce a new shadow mapping algorithm that aims for optimal resource usage while being comparable in quality to other popular approaches. A tile based partitioning scheme is provided to facilitate dynamic customizability and allow for adaptive distribution of resources at run time which leads to more efficient use of memory resources. Detailed results for memory and execution costs are presented and comparisons to commonly used previous work are made.