Slashdot Mirror
Sony Demos Natal-Like Control System
An anonymous reader writes "It's not just Microsoft investigating full body, markerless motion capture. Sony has enlisted the help of Swiss firm Atracsys to develop similar technology. Sony has openly discussed the technology with New Scientist, and has realistic expectations for the new system — it can capture broad body gestures but not individual fingers. That's just one trade-off needed in order to develop a real-time system that anyone can use, according to a markerless motion-capture expert."
It's still in the early stages of development, but the accompanying video shows the use of face-recognition software as well. The demo game has players moving their heads left or right to position their character, and then smiling to "catch" an object.
And Nintendo had motion control with Kibry Tilt and Tumble on the Gameboy Color.
The microsoft version is using Israeli military technology, it's not just a webcam. Or it is a webcam, but it's also something else.
A small time before the camera sends another frame downstream a small led in front of the camera flashes. That flash has a specific color, that's rarely seen in nature. Then it resets one layer of the CMOS camera. Then, a short while later the frame is sent downstream ... with 4 color components.
You have your superstandard red, green and blue channels. And you get a depth channel. The last channel indicates how long a time passed between the flashing of the led and that pixel seeing the color emanated by the led.
In all but the most extreme situations (there are a few extreme angles that don't work as expected, a few materials that are not opaque to the sent-out color, and there's a minimum and maximum distance, which aren't stellar, as well as limitations on how many such camera's can be active at once, as they have to timeshare), that would be the distance between the eye of the camera and whatever object the pixel was on.
Natal does not give you a 2d image, it gives you a 2.5d image (meaning x,y + height, not truly independent x,y,z coordinates), but still. The difference between normal pictures and natal pictures is like the difference between mario bros and duke nukem 3d.
Natal is the combination of a camera + a (relatively short-range) 3d range scanner. That equipment doesn't really exist yet, and while camera's are cheap, I seriously doubt you can find a 2d range scanner running at 60 Hz for less than 50.000$.