William V. Baxter III, Avneesh Sud, Naga K. Govindaraju, and Dinesh Manocha
Double Eagle Tanker: This 4 GB environment consists of more than 82 million triangles and 127 thousand objects. Our algorithm can render it 11-50 frames per second on an SGI system with two IR2 graphics pipelines and three 300 MHz R12000 CPUs.
GigaWalk is a system for interactive walkthrough of complex, gigabyte-sized environments. It combines occlusion culling and levels-of-detail and uses two graphics pipelines with one or more processors. We use a unified scene graph representation for multiple acceleration techniques, and we present novel algorithms for clustering geometry spatially, computing a scene graph hierarchy, performing conservative occlusion culling, and performing load-balancing between graphics pipelines and processors. GigaWalk has been used to render CAD environments composed of tens of millions of polygons at interactive rates on an SGI Onyx system with two Infinite Reality rendering pipelines. Overall, our system's combination of levels-of-detail and occlusion culling techniques results in significant improvements in frame-rate over view-frustum culling or either single technique alone.
Coal-Fired Power Plant: This 1.7 GB environment consists of over 13 million triangles and 1200 objects. GigaWalk can display it 12-37 frames per second on an SGI Onyx workstation using two IR2 graphics pipelines and three 300 MHz R12000 CPUs.
Paper (preprint) William V. Baxter III, Avneesh Sud, Naga K. Govindaraju, and Dinesh Manocha, "GigaWalk: Interactive Walkthrough of Complex Environments", Eurographics Rendering Workshop, 2002.
Videos/ Contains videos demonstrating the GigaWalk's interactive display.
Images/DE/ Images of the Double Eagle, showing different parts of the model and its complexity.
Images/DE/Tutorial/ Images of the Double Eagle tutorial from the video.
Images/DE/Clusters/ Results of our clustering algorithm on the Double Eagle.
Images/PP/ Images of the Power Plant, showing different parts of the model and its complexity.
Images/PP/Clusters/ Results of our clustering algorithm on the Power Plant.
The HTML has links to resized JPEG files. Full size PNG files are available as the same filename in the appropriate directory. See the README file in the directory for details about the PNG files.
CB #3175, Department of Computer Science
University of North Carolina
Chapel Hill, NC 27599-3175
(919) 962-1749
geom@cs.unc.edu