Images Worth Thousands Of Bits Of Data

Computer artists transform equations into dramatic simulations at the Ilinois Supercomputer Center. On a hot day in Illinois, a storm is brewing. Above cornfields and a dusty road, a cloud dramatically billows and grows. To the experienced eye of veteran meteorologist Robert Wilhelmson, the gathering tempest looks like a potential tornado. But don't run for the storm cellar. The cloud is only 12 inches high and the sky is just the deep blue background of a computer screen. The whole scene exists only as an elaborate series of equations and numbers that Wilhelmson has fed into a Cray supercomputer at the University of Illinois' National Center for Supercomputer Applications. Once there was a starlet startup called Digital Productions. Formed in Hollywood in the wake of an early 1980s entertainment industry love affair with supercomputer graphics, Digital was one of half a dozen companies that flourished, then failed, as movie-makers belatedly discovered the cost and complexity of silicon animation. Ironically, the National Center for Supercomputer Applications: innovative "visualization" centers owes its existence, in large part, to the woes of this Hollywood firm. Before the bottom fell out of the industry in in the mid-1980s, Digital Productions and the other companies worked their Crays overtime pumping out commercials, animated openings and logos for the television networks, and computer-generated movies like the 1984 "Last Starfighter." But at a cost of over $3,000 per second of film, these flashy graphics soon became just an expensive fad. Company after company folded; by 1986, only one remained. Spying an opportunity to make his supercomputer center unique, NCSA director Larry Smarr snared animator Craig Upson (who had learned enough about fluid dynamics to create a computer animation of Jupiter's churning clouds for the movie "2010.") and manager Nancy St. John from Digital Productions. The two were hired to do no more than set up a scientific animation program. But by 1987 they had convinced Smarr to create a full-fledged production group patterned after the commercial programs they had left. Two years later, the visualization project has become the nation's largest. Although Upson and St. John have since left, another Digital Productions alumnus, Stephan Fangmeier, now heads the group. --G.C.A. Wilhelmson's storm is a graphic simulation created from a mathematical model, one of dozens that the Illinois supercomputer center has produced in the last year with near-cinematic realism. Like the four other National Science Foundation centers, NCSA has state-of-the-art graphics machines and several supercomputers (in NCSA's case, a Cray X-MP and a Cray 2). But what makes the Illinois facility unique is that it offers more than just machines; it also boasts over a dozen artists, animators, and graphics specialists. Although these aesthetic experts admit that they often don't know much about the science, they are able to transform researchers' arcane equations and numbers into stunning information-packed images of everything from cosmic strings to slime molds. So far, their efforts have paid off with international recognition, awards, and perhaps the best compliment of all - a spiraling demand for their services. "NCSA has the most extensive and really innovative visualization program in the country," says William Wulf, NSF director of computing and information science. NCSA's visualization teams are part of a major revolution in computing, a pioneering effort to reach beyond the digital precision of computer data into the fuzzy but vastly more comprehensible world of perception. Since the center's inception in 1985, its graphics experts have helped scores of scientists translate their numbers into stunning visuals - compressing up to nine independent variables into a single animated display that can be as simple as the image of a single rotating molecule, or as complex as an animated sequence showing the collision of two neutron stars. By any measure, NCSA's experts are an unlikely assemblage. The wizard behind Wilhelmson's storm, for example, is Stefan Fangmeier - an animator who two years ago was making commercials and scientific demos in Hollywood. Another member of the group is Donna Cox, an art professor at the university who works with, among others, Illinois botanist David Onstad to create images of an insect infestation in a corn field. Then there's Matthew Arrott, who until recently was in Los Angeles creating high-tech television commercials. Now he spends his time working on projects like Harvard chemist Paul Bash's simulations of quantum-level chemical reactions. Most of the graphics experts are newcomers to science. In fact, many of the program's founders came to Illinois after the collapse of the Hollywood computer animation industry in 1986 (see story, page 17), a commercial heritage that can still be found in the team's film-industry flavor (scientists are "clients," and projects are "storyboarded"). Not surprisingly, the secret of their success is in making movies - animations of the parts of the world or universe that are either too small or large, too fast or slow, or just too difficult to see any other way. From a slow-motion animation of the complex chemical reactions in an explosion, to a vastly speeded-up simulation of two stars colliding, "we're taking the invisible and making it visible," says Cox. "We can make pictures of phenomena like cosmic strings and quantum physics that will never be directly experienced." The group makes little effort to hide its advertising roots. Much of its most impressive work is designed to showcase the potential of supercomputer graphics. In fact, says Fangmeier, "What we're really doing is making scientific commercials." NCSA has 2,500 users, but its staff can only work with several dozen scientists a year. Rather than expand infinitely to satisfy the rising demand for its services, NCSA provides examples of top-notch scientific animation and writes sophisticated graphics software for popular workstations and PCs. The intent? A proletarian vision: to bring state-of-the-art graphics to the scientific masses. Explains Fangmeier: "If you can give scientists an education and some tools, you're giving them the means to do it themselves." Of course, given the proliferation of powerful new graphics machines, from the Macintosh II to high-end workstations like those made by Apollo and Silicon Graphics, almost any scientist can visualize data. But where NCSA's teams enter is in using art and animation techniques to add extra dimensions, color, perspective, and motion, in order to wring the maximum information out of the data. That often means condensing a billion numbers into a glossy three-minute computer animation. But those three minutes are solid gold; scientists often learn more about their data in those few moments than in months of previous study. Illinois engineer Robert Haber discovered this the hard way. After using nearly a hundred hours of supercomputer time to simulate cracks in metal, he found that the basic mechanisms of stress propagation still remained stubbornly hidden in the data. A simulation he ran in 1986 had produced reams of numbers, but its traditional XY graphs of stress and energy were frustratingly opaque. "We weren't learning much," Haber admits. He suspected that complex motion in metals occurs when stress waves hit the crack, but even portraying the region in two-dimensional color failed to reveal it. The problem seemed to be an obvious target for NCSA's animated 3-D graphics, which were by then already earning an enviable reputation in the scientific community. Still, it wasn't an easy task. Although a slab of aluminum is a relatively simple object, compared to the multidimensional mathematical abstractions of some scientists, it took months for Haber and a small team at NCSA to select the views most likely to provide new insight. They eventually produced a videotape of the metal under stress. The time was well spent. When Haber first viewed the tape, he could at last see waves of stress hitting the crack, enlarging it, and being reflected outward at thousands of meters per second, like ripples in a pond. Now, nearly two years later, Haber has assembled more sophisticated machines that allow him to interact with the computer as it runs the simulations. He has even set up a new group at NCSA to refine techniques to vastly speed up the visualization process, with the goal of eventually running complex interactive animations in "real time" - like playing a video game instead of watching a movie. Just as NCSA technology helped Haber to finally see the basic mechanisms of stress propagation, other groups of graphics specialists team up with scientists to tackle problems in all areas of science. Some of the most challenging or unusual graphics projects at the center find their way to artist Cox's Renaissance Team. Cox and her small group (often just herself, a programmer, and a scientist - hence the name; a multidisciplinary Renaissance Team in a field too complex for traditional Renaissance Men) specialize in milking information out of dense data with an innovative use of color and an aptitude for coming up with fanciful geometric shapes to represent abstract physical quantities. One such project involved what Cox and Kodak engineer Richard Ellson call "glyphs." As a scientist working for one of NCSA's industrial partners, Ellson had finished a supercomputer simulation last year that approximated the complex flow of molten plastic into a thin mold. He hoped it would help other engineers at Kodak understand the changing pressures inside molds so they could design new molds that wouldn't split while they are filled with the high-pressure plastic. But, although the computer produced perfectly acceptable-looking numbers, they couldn't be checked against reality. Ellson's solution: an artist's "sleight of hand." Donna Cox created an innovative technique that clearly displayed a record nine distinct variables simultaneously changing in an animated videotape. To pack variables to such a density, Cox invented a unique 3-D wedge shape, the glyph (from hieroglyphics, the Egyptian pictographs), to represent each computed portion of the flowing plastic. The shape, color (the blue side of the spectrum for pressure and the red for temperature), and orientation of the wedge indicate the state of the flowing material at particular points. The finished videotape shows the plastic (in the form of an army of small wedges) marching into the mold, swiveling, changing direction and color, and eventually settling and hardening in a series of complex steps. As a result of techniques refined in the work of Cox and Ellson, Kodak engineers can now design new molds more quickly and accurately. For Ellson, seeing his work in pictures meant relief. "It let me know that the simulation was still making sense, even as we moved into thinner and thinner molds," he says. With the work of Cox, Haber, Arrott and others at NCSA, an increasing number of scientists are discovering what Illinois physicist Klaus Schulten means when he calls supercomputers "intelligent microscopes." As the sophistication of machines and software grows, the list of potential applications will encompass the breadth of science. For many in the next generation of scientists, speculates Schulten, the supercomputer may become the only lab they know.

Images Worth Thousands Of Bits Of Data

Interested in reading more?

Become a Member of