Embed Size (px)
Citation preview
Dr. Oliver B. R. Strimpel, Curator The Computer Museum 300 Congress Street Boston, MA 02210
Dear Dr. Strimpel,
Lawrence Livermore National Laboratory
April 2, 1984
Enclosed are three "Grand Canyon" and three "San Francisco Bay Area" images, as well as a brief description of the method that was employed to generate them. I hope you find them sufficiently interesting to include some of them in your exhibits. I certainly would be flattered by that decision.
As soon as I receive the extra CRA Y-CHAN NELS magazines, I will mail a copy to you. I would also appreciate it if you could send to me some information about The Computer Museum.
PPW /dgs Enclosures
With best wishes,
II&t/~ Patrick P. Weidhaas Atmospheric and Geophysical
Sciences Division
An Equal Opportunity Employer· University of California· P.D. Box 808 Livermore, California 94550 • Telephone (415) 422-1100 • Twx 910-386-8339 UCLLL LVMR
February 24, 1984
Information regarding the accompanying computelJ{enerated images
by Patrick P. Weidhaas
The accompanying images are representations of gridded elevation data in a computer file extracted from a digital data base covering the entire continental U.S. This large data base ('" 7 gigabytes) contains land surface elevations at a resolution of 65 meters.
In 1983 two programmers at LLNL, Craig Upson and Patrick Weidhaas, developed a computer program that would produce three-dimensional color images of any subset of the digital terrain data base as viewed from any point of observation. The major difficulty in three-dimensional graphics is the elimination of so-called "hidden surfaces." As one looks sideways into the terrain, distant hills may be obscured ("hidden") by other hills or mountains that are located closer to the observer. The technique employed by Upson and Weidhaas to deal with this problem is known as "visible surface ray tracing." It basically determines all points on the terrain surface that are visible, and does not waste any time on invisible (hidden) surfaces.
Conceptually, one should think of the observer as a point (the "eye") from which many straight lines ("view rays") emanate. Many of these view rays will never hit the terrain surface; they represent the sky. Many more, however, will intersect the terrain; the point of intersection closest to the eye is also the termination of that view ray, and it is clearly a point on the surface that is visible to the observer. Obviously, a computer program can only handle finitely many view rays, and in fact for each picture element ("pixel") on a raster screen (which conceptually is placed between the observer and the terrain), one view ray is generated and drawn from the eye through the pixel and traced until it intersects the terrain.
The color of each pixel is determined by the elevation of the associated point on the terrain surface. In addition, the brightness of a pixel is varied according to the angle which a small surface element surrounding the given point makes with the direction to an imaginary sun. This last feature then provides surface shading which is crucial in producing a realistic depth-effect.
The images show an area of the Gl'and Canyon viewed from the North Rim towards the South Rim. As mentioned above, surface colors are assigned by elevation. In addition, the effect of atmospheric haze has been modelled by Changing the colors by distance, arriving at a bluish color for distant hills. The pixel resolution used here is 2048 X 1700, in other words almost 3.5 million view rays (one per pixel) had to be traced. The program produces three ouput files, one containing the intensities of "red" for all pixels, one for "green," and one for "blue." These three "color-files" are then projected sequentially on a DICOMED D48 film recorder, and recorded on film (either 4 X 5 negative, or 35 mm slide film). After the film has been exposed to all three color projections, it contains the complete color image.
The program, named "TERRA," was written for a CRA Y-l computer. To generate one image (i.e., 3 color files) at a resolution of 2048 X 1700 required approximately 1.5 minutes of CPU time on the CRAY-l. This time can be drastically reduced by using a lower pixel resolution. For example, an image at a 500 X 400 resolution takes about 8 seconds of CPU time.
Becomes a Reality ...
Visualizing terrain for real-time emergency assessment Th e Atmospheric Release Advisory Capability at Lawrence Livermore National Laboratory relies heavily on CRAY computers to provide a real-time assessment of possib le consequences of an atmospheric release of radioactive material. Th e distribution and contamination of such a release is a fun ction of the wind velocity and direction near the earth's surface. This in turn is highly dependent on the local terrain. To attempt to predict the diffusion of released material therefore requires the land surface elevation data of the affected geographic area. This 7 -g igabyte data base was provided by the Defense Mapping Agency.
What if the meteorologist study ing the diffusion of released rad ioact ive materials could see the topography of the affected area? This visual presentation might provide va luable c lues about the atmospheric phenomena affecting the transportation and diffusion of the re leased material.
Patrick We idhaas and Cra ig Upson of the University of Ca lifornia Lawrence Livermore National Laboratory have made this imagined capabi lity a reality with the "TERR A" program written for C RAY computers. The program uses a modified ray-tracing algorith m to allow the presentation of any location in the continental U.S. from any ang le and any elevat ion.
Imagine what else you cou ld do with a CRAYl
Figures: Top - The San Francisco Bay Area. © Patrick Weidhaas, La wrence Livermore Nationa l Laboratory, 1983. Bottom - The Grand Canyon in Winter. © Patrick Weidhaas , Lawrence Livermore National Laboratory, 1984.
Acknowledgement: This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under contract No. W-7405-Eng-4B.
N/aking the Imagined a Realitv ••• Making the imagined a reality has become commonplace using CRAY computers. Previously insoluble problems in the aerospace, petroleum and automotive industries, in science, engineering, and graphics are being solved today using the power and flexibility of CRA Y computer systems. In each of these disciplines, the CRA Y is used to simulate a real-world process with a computational model in less time and at less cost.
To support these applications, a wide range of graphic software systems is offered for CRAY computers. Oeviceindependent line drawing systems like 01-3000 from Precision Visuals, Inc., TEMPLATE from Megatek, Inc., and OISSPLA from ISSCO, Inc., are being used now on many Cray computers. Systems
for CAD/CAM and pre- and postprocessing like PATRAN from POA Engineering and MOVIE.BYU from Brigham Young University support a variety of engineering design activities. In those cases where photographic quality scene generation is the objective, the designers, artists, scientists, and movie-makers are turning to Cray systems to do what could not otherwise be done.
If your application or graphics task requires extraordinary computer power ... if the problems you can do are much smaller than the problems you would like to do . . . if you need a general purpose powerhouse to run a variety of simulation, engineering or scientific codes ... you need a CRAY!
I=IESEAI=ICH, INC.
Corporate Headquarters 608 Second Avenue South Minneapolis, MN 55402 612/333-5889
Domestic sales offices
Albuquerque, New Mexico Atlanta, Georgia Beltsville, Maryland Boston, Massachusetts Boulder, Colorado Chicago, Illinois Dallas, Texas Houston, Texas Laurel, Maryland Los Angeles, California Minneapolis, Minnesota Pittsburgh, Pennsylvania Pleasanton, California Seattle, Washington Tulsa, Oklahoma
International subsidiaries
Cray Research (UK) Limited Bracknell, Berkshire United Kingdom
Cray Research GmbH Munich, West Germany
Cray Research France, SA Paris, France
Cray Research Japan, Limited Tokyo, Japan
Cray Canada Inc. Toronto, Canada
., Copyright 1984. Cray Research. Inc.
AP-Ol0l
April 6 1874
Mr Patrick Weidhaas Atmospheric and Geophysical Sciences Division Lawrence Livermore National Laboratory PO Box 808 Livermore CA 94550
Dear Mr Weidhaas
Thank you for your letter of 2 April and the slides. The slides are most impressive. Did you by any chance make up some more nearly vertical views looking down over the Bay area or the Grand Canyon? It might be interesting to show how the computer-generated view can alter the angle of view as well as the position of the viewpoint.
Incidentally, Mike Kobrick at JPL showed me a view of Death Valley generated from the digital topographic data, but the colouring was taken from Landsat and Seasat imagery. Comparison with a real photograph reveals quite a good match. However his pictures cover a smaller region than yours and so the 65-metre pixels are quite visible in the foreground.
I think I may well wish to include some of your slides in the exhibit, especially if one could compare a view with, say, a straightforward vertical view coloured with the same scheme. This would then show how much more information the new perspective conveys to the eye without adding any more data - in fact with less owing to the hidden portions.
I enclose a copy of the latest edition of The Computer Museum Report and the outline of my exhibit.
I look forward to your comments and suggestions. Thank you for your interest.
Yours sincerely
Dr Oliver Strimpel Curator
enclosures
