Developing systems for interactive visualization of arbitrarily large datasets is a fundamental challenge in computer graphics. We propose to build a visualization system capable of processing arbitrarily large polygonal datasets at interactive frame rates and on a wide range of hardware configurations. The system will adjust the quality of the images and the frame rate depending on the performance of each stage of the visualization pipeline. On high-end devices, the system will deliver high-resolution images at real-time frame rates, while on lower-end devices (e.g., PDAs), the system will deliver smaller images at slower (albeit interactive) frame rates. To achieve this scalable performance over a spectrum of devices, we will develop techniques that effectively decouple the visualization requirements from the available hardware resources.
Interactive visualization of large datasets has applications in many areas. Examples of such datasets include computer-aided design (CAD) models for manufacturing (e.g., Boeing's 777), models of real-world environments for disaster simulation and analysis, and results of scientific simulations (e.g., environmental modeling). Because most of these data are represented as collections of triangles, visualizing polygonal datasets is a particularly important problem. Using this technology, end-users will be able to interact with complex visualizations of large 3D data from the office or on the field.
