Heads up. The “secret” is out. The official branding for Microsoft’s project Natal? Microsoft Kinect for xBox 360. Now, we’ve been talking about Kinect for a while now, so this official branding comes as no surprise, but lets take a closer look at how the technology works.

kinetic-xbox360-microsoft

We know that Kinect uses a multi-sensor strategy to track a ton of 3D data, about what’s going on inside a room. To start, there are a pair of cameras. A simple colour camera for metadata (pictures, face-recognition, etc), and an interesting monochrome CMOS camera sensor working together with an IR emitter. Called the “depth sensor,” the combo acts very much like a radar setup, the IR emitter “painting” the area, and the camera watching and reading the “paint-work”, to build a 3D map of the objects within Kinect’s field of view. These opticals are backed up with some audio tracking, with a multi-array microphone setup to detect location of voices and to cancel out ambient noise. Like sonar…

The sensor array is mounted in a motorised, articulated base. Yes, you heard right. Like a target-tracking radar dish, the Kinect pans and tilts to keep its sensors locked on you as you move around the room. Too cool! The base is quite power-hungry, so unless you have one of the brand new Xbox 360s, which include a special powered USB port, you’ll have to plug Kinect into the mains, as well as your console.

Now for some bad news. Kinect is a slave device. It doesn’t have it’s own dedicated processor for all this tracking alchemy. It’s a cost issue, and the upshot is that the xBox processor runs the whole show. That means a performance hit on both ends and consequently, the Kinect/xBox 360 combination is just not as snappy as the Sony Playstation Move.

xbox-360-slim

Finally a couple of numbers to describe the system:
1. Kinect “sees” 57 degrees of azimuth and 43 degrees of elevation Field of view. (The motorised base contributes 27 degrees.)
2. Depth of field extends from just over a meter to almost 4 meters.
3. Skeletal tracking (accurate, full-body data), tracks 2 active players at a time, while the Kinect tracks up to six people in the background.
4. Resolution is 320 x 240, 16-bit at 30 fps (IR camera), 640×480 32-bit color at 30 fps (colour camera).
5. Audio is 16-bit, at 16 kHz