Visualization and 3D Graphics

We have already alluded to how fundamental limits are reached with regard to information absorption. That is, we need our new information sources more finely digested and presented in ways that we can understand and understand more rapidly. Passive data-display devices, such as the Orb, are useful. Heads-up eyewear and immersion devices are useful. Advances in 3D graphics and displays can bring an even deeper presentation of information.

The fundamental building block of computerized graphics is a small triangle, partly because a minimum of three points in space is required to describe a surface. Put enough triangles together and you can finely approximate any image. Creating a realistic scene or image in full motion can require hundreds of millions of triangles. "Ten years ago, the computerized models of a car would have 10,000 triangles and be rendered in real time at 20 to 30 frames per second and at a million pixels per frame," describes Dr. Goh. Each triangle (a mathematical model) is rendered into dots or "pixels" on the computer screen. The pixels fill in the shape of the object. "So, 10,000 triangles go into the system and 1 million pixels come out per frame. By contrast, today, we have data sets of hundreds of millions or even a billion triangles, but commonly 3 to 10 million pixels out. We've gone from a data magnification to a data reduction problem. The latter is therefore the focus of much of our visualization research."

A billion triangles per frame at 30 frames a second (for full-motion graphics) makes for a lot of processing in just 1 second. Such rendering capability is currently the province of super-computer-scale machines. Of those billion triangles, only a fraction of them will be visible from whatever the current viewpoint is that the computer presents to the viewer. The system first calculates which will be the visible ones. Then, it has to angle and rotate each remaining triangle to account for the current viewpoint. Shading, coloring, shadows, texture, and illumination all need to be calculated as wellall calculations are mathematically intensive. The result is a realistic rendering of a certain object at a given view angle under specific lighting and interference conditions (shadows, obstructions, etc.) for one-thirtieth of a second.

Why the dramatic increase in the number of triangles needing to be processed? Users of computer graphics software are now demanding a much more realistic rendering of an image. In years past, high-end computer-aided design (CAD) systems could model only one aspect of a car, for example, at a time, such as the exterior only or the tail section only. Today, automobile engineers want to model and view the entire car, including all the engine components, all the body panels, and the entire interiora 100 percent complete model. Why? Because only then can engineers truly tell that absolutely everything fits together perfectly, has the right spaces around it, and is aesthetically acceptable overall. Given such a complete model, engineers can run extremely compute-intensive (and accurate) thermal simulations, weight-distribution calculations, air-foil simulations, air-flow simulations (around the engine and computer components), and place human models within the interior. The explosion of data needed for more complete modeling in turn begets another explosion of simulation data, and the requisite computing horsepower needed has to grow proportionally to keep pace.

Extremely complex simulations of automobiles, proteins, weather patterns, and earthquakes will in all likelihood get the processing power needed resulting from the new generation of super computers now on the horizon. However, Dr. Goh believes that the technology that goes along with these new computing forms will "trickle down" to smaller and more pervasive computing devices. Soon, PDA device displays will be available with XGA or better quality resolutiona turn of events Dr. Goh believes that will be highly disruptive. Currently, PDA displays are "good enough" for displaying text such as e-mail and very limited Web-style graphics. Greatly enhanced PDA displays that can render a Web page, for example, in finite detail, will make owning a PDA-like device far more compelling to those who have so far resisted the lure of anywhere-anytime connectivity. Furthermore, it will enable scientific and engineering applications to include PDAs as data gathering and presentation devices.

OLEDs

The physical form factor for PDA displays may change as a result of a new type of display technology. Imagine a flat-panel computer display with the flexibility of a sheet of plastic. Within three to four years, displays based on the use of organic, light-emitting diode materials (called OLEDs) will begin to replace the standard CRT and LCD screens in common use today. We say replace because OLEDs promise to be cheaper to make; consume far less power; and project a brighter, clearer image. As an added bonus, they will go almost anywhere.

Because they can use glass, plastic, and metal foil as substrates, OLEDs (and a derivative technology called PLEDs) will appear on wristbands, cylindrical columns in buildings, articles of clothing, and in all the places we now use electronic displays. They can be as big as billboards, as walls, and street signs, or as small as the face of your watch. They could even be embedded in virtual skin. But, their real benefit will be seen immediately by PDA and laptop users. Imagine a computer screen that rolls up and drops into your purse or pocket. Using OLEDs, PDA displays could actually display lots of readable, intelligible information (such as an entire Web page) exactly as you now see it on your laptop, and with better resolution. In fact, OLEDs could allow multifunction PDAs and laptops to converge into a single personal, portable, universal information unit.

The Holy Grail for some visual modeling applications is holographic displays. A holographic image is one that appears in three-dimensional space, as if suspended in air. Right now, the demand for such display technology is low, partly because the image quality is still poor. Systems still cannot process enough triangles in one-thirtieth of a second to give realistic depth perception. Today, a pixel (the picture cell) contains only enough information to display color and brightness. A "hogel" (the holographic cell equivalent of the triangle) needs to have more information (namely "phase" among other details). Looking at a holographic image is much like looking at a real physical scene through a windowpane. To render such detail in real time requires tremendous compute power, which today is only barely available even from super computers. With the help of Moore's law (a doubling of compute power every 18 months), scalability, innovative new hybrid computer architectures, and remote display technologies, the processing power will eventually percolate downward to workstations and PDAs. There are, however, still some challenges.

Another remaining challenge is determining what the actual display technology will be for holographics. Currently, images can be generated using three different colored lasers: red, green, and blue. It turns out that red lasers are inexpensive compared to green and blue. This is partly because of their mass-market use in every CD drive and related technology, and partly because of the fact that red-light wavelength, being longer that green and blue, makes red lasers easier to mass produce. Green and blue lasers are more expensive to produce given the shorter wavelength of the light they radiate, and so they are seldom used in commercial applications (and thus no economies of scale). A commercial application for green and blue lasers is now on the horizon: holographic storage.

Blue-laser data storage disks will be appearing in early 2005. Today, red-laser DVD-type devices can hold up to 4 gigabytes of data. Blue-laser disks have been shown to hold up to 50 gigabytes (equivalent to 4.5 hours of recorded HDTV) partly by virtue of shorter wavelength of blue light. In a few years, true 3D-storage holographic disks using blue lasers will be available, yielding a tera-byte of storage per penny-costing disk of plastic. Some computer storage companies are driving this technology forward in earnest. As discussed in previous chapters, our desire to "store everything indefinitely" coupled with massive data sources, such as movies and videos, will soak up as much capacity as the storage industry can muster upbut only at the right price.

The amazing twist to all this is that the storage industry could essentially be building 3D holographic projection systems (granted, at a different granularity and designed, instead, to hold digital information for computer consumption). Nevertheless, Dr. Goh believes that it will be the massive deployment and success of such technology in the general computer storage sector that will enable holographic scene-projection devices to be built. As Dr. Goh states:

We're essentially just waiting; the prices will come down and the component availability will go upwe'll wait until the prices are low enough and pieces are available, then knit them together and repurpose these storage components for display systems. Realistic holographic displays are no longer skeptical future fantasies. They will indeed be practical and will someday be ubiquitous.

No doubt, holographic displays could forever change the way in which we visualize information. If possible, 3D holographic images will prove to have value far beyond displaying simple scenes. The world will find ways to display "information" in useful 3D forms. Imagine "walking" through your medical recordsturn your head left to see your genome and walk around your gene segments; turn right and wade through a 3D scan of your liver.

Thinking more abstractly and less pictorially, holographic displays will be a boon for the engineers of computer programs and data connectivity devices. Imagine being able to see the relationships of one database tied to another in some sort of artistic rendition. Imagine seeing computer subroutines floating and interacting with data. Computer programmers have seen nearly zero innovations in tools that help the programmer visualize relationships. Most attempts at pictorial programming models have used 2D displays based on "boxes" and lines to show relationshipsa far cry from 3D holographic representations that could even show time as a component.

While building a program, the mind of the computer engineer swells with bizarre artificial models of relationships because no other models exist. No two programmers build the same fantasy modelone reason for the difficulty in maintaining software applications after they are built. In contrast, an engineer building a car or airplane is ultimately building something physical and therefore naturally uses tools to help him visualize the potential end result. Those same tools help others working on the same projectsafety engineers, manufacturing engineers, artists, etc. Holographic display breakthroughs will come from an ability to visualize information that actually has no natural physical rendition, such as parts of a computer program itself, databases, and their relationships to other data sets, where color, texture, depth, age, interaction, and other attributes can be represented in 3D space. Watch for such innovations in "information" display, and then watch massive advances take place for a wide variety of data-centric applications that today have no opportunity for visualization.