QTViewer - Software for Three-Dimensional Visualization of Large Models
The recent emergence of high-resolution interferometric synthetic-aperture radar (IFSAR) and lidar has offered the potential for producing high-resolution digital topography automatically at near real-time rates. The significant advantage of IFSAR is an all-weather collection capability. The main advantage of lidar is a generally higher data quality. The recent availability of these data collection systems has opened up multiple possibilities for exploiting high-resolution digital topography. However, the previous software available for exploitation had significant limitations. Either the 3D data was converted to 2D products that could be employed by conventional GIS workstation software, or small subsets (less that ~1 million samples) of the data were selected for visualization. Larger data sets could be selected for visualization but at a cost of significant render times or expensive hardware facilities.
Because of the large-model problem in computer graphics, high-resolution 3D data had to be substantially thinned out to remove the geometry-processing bottleneck for virtual-reality simulations. To recover complexity in the scenes, texture maps were manually inserted over the geometry. For example, a tree would be represented as an artist's conception using a texture and transparency map over a flat rectangular plane. This whole process of manually transforming data so that virtual-reality simulations can handle simplified geometry takes so much manual effort and cost that it constitutes a very large portion of the total cost for today's simulators.
Two recent technology developments have opened up the potential for significantly better exploitation. The first is the emergence of cost-effective and very high power 3D chips and boards. This emergence was driven by the demands of the 3D video game market and resulted in a 3D-chip performance growth rate that significantly exceeds that for conventional microprocessors. The result is the wide availability of modest-cost PC plug-in boards that can render tens of millions of triangles per second. The second development is the application of quad-tree data structures to visualization algorithms. Quad-tree data structures are an efficient method for encoding 3D data in a manner that permits rapid visualization. When combined with recent 3D chips and boards, near-real-time visualization of large digital topographic data sets can be performed on PC-based systems. This represents a significant paradigm shift for GIS because 3D data sets can now be used as easily as 2D data sets.
The Johns Hopkins Applied Physics Laboratory has developed software using these new technology developments for visualizing very large 3D models in real time with standard PC hardware. Using cost-effective 3D chips developed primarily for the video-game market, real-time display of large models is achieved using improved data structures (quad-tree-based) for data storage and retrieval. 3D visualization can now provide improved utility because the previous need for high computational capabilities (i.e. graphics supercomputers) is now eliminated. High-quality performance can be obtained with laptop computers for a fraction of the previous cost. The implication of this work is a paradigm shift for geo-spatial information systems -- 3D data can now be as easy to use as 2D data.
For more info, go to http://www.jhuapl.edu/qtviewer.
*For licensing information, please contact Applied Imagery: www.appliedimagery.com, firstname.lastname@example.org.CONTACT:
Mr. K. Chao
Phone: (443) 778-7927