By Rafe Needleman / January 8, 2008 9:55 AM PST
LAS VEGAS– Intel CEO Paul Otellini’s CES keynote was sparkling. In contrast to Bill Gates’ pastel portrait of the future, Otellini presented a concrete vision of a personal, reactive Web, and the challenges to creating it (Silicon, Infrastructure, Context, and Interface). For a full rundown, see Dan Farber’s writeup on ZDNet.
Intel loves where the Web is going. The more interactive and personal it gets, the more processing power is required and the more new chips Intel sells, for both servers and local workstations. The most interesting (and newest) product that Otellini brought to the stage in his keynote was an automatic avatar builder made by BigStage.
BigStage creates a model of anyone’s head by using just three photos–head-on, rotated a little, and rotated a little more. The company processes these pictures on its own servers and ends up with a model that knows which pixels your eyes are (so it can move and blink them), where your mouth is, and so it. In the Intel keynote demo, BigStage found Jonathan Strietzel created an avatar of Smash Mouth singer Steven Harwell. It was eerily good–much better and less creepy than avatars I’ve seen previously.
The technology comes from a CIA-funded project at the University of California. It was originally intended for scanning surveillance cams, since at its core it measures the three-dimensional geometry of key points on a face, for example between eyes, or the shape of a person’s cheekbone. The fact that the algorithm can extract a complete 3D model from only three images, and with what is now reasonably inexpensive computation (this is where Intel comes in) is what makes it commercially viable.