How Kinect Fusion and Kinect Work

Note that microsoft has not yet released Kinect Fusion and may never release its source code. But, the Point Cloud Library (PCL) open source project has implemented Kinect Fusion. It is still under development, but the source code is available and does work somewhat. No color yet though. I plan to contribute to PCL's version of Kinect Fusion. This video contains detailed visualizations and high level explanations of how the Kinect Fusion algorithms create incredible 3D models in real-time. I also forgot to mention that after getting the trajectory of the dots, it also calculates the trajectory into the camera. These two rays give a unique 3D point using triangulation. That's how it determines the 3D position of the dots. And lastly, I have to give credit to everyone that I got video from. I'm sorry, very little, if any of the content is original because I'd rather not spend weeks recreating existing videos. The videos where kinect fusion is actually doing its work came from microsoft videos here: http://www.youtube.com/watch?v=quGhaggn3cQ http://www.i-programmer.info/news/144-graphics-and-games/3122-kinectfusion-instant-3d-models.html One of the infrared dot videos was taken from here. This dude was just a bit too excited :) http://www.youtube.com/watch?v=E-x8Mn1l3rA Another point infrared dot video was taken from here: http://www.youtube.com/watch?v=DxDRMJgUKXI The video depth map of the guy waving was taken from here: http://www.youtube.com/watch?v=sWNrhVGRiXE The ICP (Iterative Closest Point) visualization was taken from here: http://www.youtube.com/watch?v=Xq91aMwYezU The blue images of the kinect dot patterns were taken from: www.futurepicture.org/?p=116