Kinect the dots.

     There's some neat IR video of the Microsoft Kinect peripheral for Xbox floating around on the web.  Basically the device shoots out thousands of points of IR light for the camera to pick up on.  I wonder how the algorithm which deciphers the dot pattern works?  They grid they form seems to be more random than structured from what I can tell.  I wouldn't have thought that the IR camera in the Kinect would have the resolution necessary to pick up on different sized dots for different distances.  (The light beams that Kinect shoots out aren't parallel, things closer to the Kinect have smaller dots.)  Does it pick up elongated dots and interpret them as angled surfaces?  Interesting stuff.
    This video seems to indicate that it just uses the IR grid for depth perception.  That's probably a lot simpler - it doesn't need to measure the individual size or elongation of the dots, only how far apart they are.  You can do this with a much lower resolution (i.e. cheaper) camera.
     Still, it's nice to see MS doing something interesting for once.