Contents 1 Introduction
Level-of-detail rendering techniques aim to manage intelligently the complexity of rendered scenes at render time in order to regulate frame rates while maximizing the perceptual benefit of the rendering to the user. Successful methods are typically based on the use of hierarchical scene descriptions that combine the elegance of a recursive hierarchical spatial or geometric decomposition of the scene with the ability to provide simplified drawable representations for groups of related subobjects. The use of such techniques for active frame rate control has created the need for intelligent hierarchically aware level-of-detail optimization algorithms whose task is to select at render-time between the multiple drawable scene representations that may be extracted from hierarchical scene descriptions. Their aim in this selection is the maximization of the perceptual benefit of the resulting rendering, subject to a constraint on its total rendering cost, so as to maintain regular frame rates while making the best use of available resources. Although several such algorithms have been proposed, there has been little in the way of robust analytical investigation of the hierarchical level-of-detail optimization problem. In this paper, we present a semantic aid to this investigation in the form of the level-of-detail graph, a graphical representation of the level-of-detail state spaces generated by hierarchical scene descriptions in which any number of simplified representations may be provided for groups of related objects. We demonstrate by example in Section 6 how these graphs have enabled us to gain a better understanding of the meaning of shared object representations and to derive new analytical results concerning level-of-detail optimization algorithms.
