Managing Simulation Level-of-Detail with the LOD Trader

Abstract

We present the "LOD Trader", an algorithm for optimizing the perceptual quality of a simulation while respecting computational resource constraints by selectively reducing the level of detail with which less important entities are simulated. Our approach to LOD selection is better suited to simulation LOD than conventional graphical LOD techniques have been. We motivate the need for simulation-specific methods of LOD selection and describe a viewer-centric and perceptually-based measure of simulation quality. We then describe the algorithm itself, which maximizes simulation quality while respecting resource constraints. Finally, we describe a case study in which the LOD Trader produced superior results to conventional LOD selection. The information in this paper is a shortened version of [Sunshine-Hill 2011].