Kinect modified to capture true 3D video
University of California Davis researcher Oliver Kreylos has produced one of the most amazing demonstrations of true 3D video I’ve ever seen using an off-the-shelf Microsoft Kinect (a Microsoft Xbox360 accessory). This is not the cheesy stereoscopic tacky “3D” moniker being affixed to recent movie titles but actual video footage that you can rotate in three dimensional space. See the video sample below from Mr. Kreylos. My jaw dropped when he rotated the entire scene.
The way this is working is that the Kinect camera has a depth sensor which allowed Kreylos render a 3D image of himself with the color values mapped onto them. Kreylos used an unauthorized Open Source driver for the Kinect, but Microsoft should really considering hiring Kreylos rather than suing him.
Now if he could just get 3 more Kinects covering his sides and back and another Kinect from up top looking down, we could get something that would almost cover the entire room. Just imagine television coverage doing this for popular events. Bandwidth requirements would likely double and computer processing would need to increase dramatically to render 1080P in true 3D, but this would be a huge boon for networking and processing technology and it should be of great interest to companies like Cisco and Intel.
Update 11/16/2010 – Here is a second video showing the measuring accuracy of the system.