Caltech: Combating Future "Bandwidth Bottleneck"
PASADENA, Calif.— The next exciting generation of multimedia on the Internet will be computer images created by three-dimensional geometry. Soon, instead of viewing a simple picture on their computer monitor, a user will be able to view an object from any viewpoint, lighting, or surface. The result will be images that seem to come to life.
It's exciting technology, unless, that is, your personal computer suffers from "bandwidth bottleneck," the frustrating experience that results in the herky-jerky motion of today's so-called streaming video, or audio that breaks up and is hard to hear. Now a solution is in hand for this up coming problem, thanks to a California Institute of Technology professor who's been named a finalist in Discover Magazine's 2001 Innovation Awards.
A team of computer scientists led by Caltech's Peter Schröder, a professor of computer science and applied and computational mathematics, and Wim Sweldens of Lucent Technologies, was acknowledged for developing the most powerful technique to date for computer graphics that will make it practical to send such detailed 3-D data over the Internet. The Innovation Awards are presented annually by Discover, the science publication, and are intended to honor scientists whose groundbreaking work will change the way we live.
Bandwidth bottleneck, a phrase used by Schröder, is caused by too much data being sent through a slow Internet connection to a slow computer. Schröder's expertise is in the mathematical foundations of computer graphics. He and his colleagues developed a compression algorithm—a repetitive computational procedure—that will make it practical to send 3-D images over the Internet, and to manipulate them on personal computers and handhelds.
Without this algorithm, the coming 3-D images would just further bog down the average PC with gobs of data. That's because such 3-D objects describe the actual geometry of an object, such as its depth or height, in detail, with all its measurements. The object could be a human head, or a part for an automobile in an on-line catalog, or a cartoon character. Such digital geometric data is typically acquired by 3-D laser scanning and represents objects using dense meshes with up to millions or even billions of triangles.
The compression challenge is to use the fewest possible bits (the basic unit of information in a digital computing system) to store and transmit these huge and complex sets of data. Efficient geometry compression—delivering the same or higher quality with fewer bits—could unlock the potential of high-end 3-D on consumer systems. The researchers, led by Schröder and Sweldens, report that their technique for geometry compression is up to 12 times better than standard compression methods.
Why this will be exciting for the everyday PC user is the power of 3-D images. "Imagine being able to download a 3-D model of Michelangelo's David to your home computer," says Schröder. "Not only would you see an individual picture, but you could examine in detail the chisel marks on David's cheek, or see what the statue looks like if you stood on a tall ladder."
Today, he says, such riches are reserved for high-end computer users with very high bandwidth Internet connections. "Sometime soon, though," says Schröder, "it can be available to any schoolchild the world over."
