Interactive Scenegraph Performance Analysis, Diagnosis, and Enhancement

Abstract

In recent years scenegraphs have emerged as the most preferred way to represent and interact with large models in the field of computer graphics. The performance of any scenegraph application depends critically on the underlying structure of the data. Since there are many sources of 3D models and methods for constructing scenegraph data structures, poorly structured input is common. This work presents the design and implementation of an extensible performance monitoring system for a popular open source scenegraph, OpenSG. Several methods for interrogating the various factors which can affect the performance of a scenegraph are presented. A software implementation of these methods provides a tool that enables interactive exploration and diagnosis of a scenegraph to identify potential performance bottlenecks. The tool also implements some optimization techniques that can be applied to improve the structure of the scenegraph and thus the performance of the application.