Now that a new patent has shown its face on the Internet, and we know that the Kinect is fully capable of recognizing sign language (which just opens so many doors, it’s almost ridiculous), why don’t we take a look inside the device itself? Just like with any other proper piece of technology out there, the tear-down is inevitable. While this one may not be as in-depth as many of the others out there, it’s good enough for now. And, happily enough, it wields some pretty interesting pieces of intel.
Through their inspection, T3 was able to pull how some of the things work within the motion-based peripheral. For example, the motion sensor is made up of two parts: one IR VGA camera, and the projector. Apparently, that projector will fire out a laser that scans the entire field of play, which the camera will then pick up and interpret, to decipher between you, the second player, and your couch. (Among other things in your living room, obviously.) This is called the “depth field.” Which is pretty appropriately named.
Perhaps more interesting, though, is the way that Kinect interprets the pixels it receives from that depth field. The camera will pick up the image that the laser has painted, and then the software will immediately take that image and run it through a multitude of filters, which will work out what is a person and what isn’t. The camera and the software work together to work with the IR noise that it receives from the pixels of the camera, which is then measured in a varying color spectrum. That’s why you have to stand a certain amount of distance from Kinect, so that it can register you, versus everything else. Interestingly enough, Kinect will paint you in bright shades of red, green, or other colors, while the things in the background are all a gray shade.
In other words, as quickly as it can, Kinect has to work out what a human is, depending on each situation and environment. The system will follow a set amount of system guidelines to determine the human veruss, say, your coffee table. It does this because the designers and engineers have told Kinect that a human has two arms, two legs, and a head. The system will then determine how tall, wide, short, or skinny you are, right on the fly. That’s where that whole “body scan” thing comes in.
Truth be told, Kinect may be targeted at the casual gamer out there, at least right now, but there’s obviously some hardcore technology empowering this thing. We can’t wait to see what it all breaks down into, and what games developers out there can come up with. These are definitely some interesting times, and we can’t wait to see what happens.