Traditional 3-D technology describes objects in agonizingly exact detail, using dense meshes of millions or even billions of individual triangles. To reduce file sizes and speed transfers, Peter Schršder, a professor of computer science and applied and computational mathematics at Caltech, and Wim Sweldens, director of of algorithms scientific computing research at Bell Labs, set about simplifying 3-D image geometry. They did this by stripping the amount of numbers needed to describe a particular 3-D image point from three to one.

Simplifying point descriptions cut file size by about a factor of three. To achieve additional compression, the researchers turned to wavelet transformations ("wavelets" are a mathematical technique complementary to Fourier transforms?the formulas that lie at the heart of digital signal processing). The researchers already knew that wavelets built with Fourier techniques, while extremely useful for compression, couldn’t handle the geometry of curved surfaces very well. So the team pioneered a technique of generating wavelets without using Fourier transforms.

The arrival of high-quality 3-D images could revolutionize the way people use the Internet. "Imagine being able to download a 3-D model of a home you’re thinking of buying," says Sweldens. "Not only would you be able to move from room to room, you would be able to see all of the features in high detail, right down to the cracks in the wall." Today only users with high-bandwidth connections can take advantage of such visuals, he notes, but "in a few years, they could be available to anyone."