Not a member yet? Register for full benefits!

VR Interfaces: The Webcam/EyeToy


Overview of The Webcam/EyeToy
In a paradigm where overcoming physical constraints is key, a webcam may seem a very odd VR interface device. However, what lays true for one section of individuals, dows not always lay true for another. The PS2’s EyeToy has done much in the cause of using physical body form as an input device.

Essentially a web camera with inbuilt microphone, and control software behind it, to separate the moving player from the background, work out which direction their various appendages are heading in, and at what speed, the eye-toy and webcam become pretty much interchangeable.

There are severe problems with this approach, of course. As with all machine vision systems working from a single reference point, the webcam has to interpret a 3D world as 2D, lacking any trace of depth perception. It does not function at all well when two people are in the shot, or there are other moving objects such as cars moving past a window, a wandering pet, or second person in the room.

Even the moving image on a television screen is enough to throw it off.

The camera-VR concept works a little like the immersion in a CAVE – it takes your physical form, or in this case, a 2D image of your physical form, with the background edited out in real-time, and puts it on the screen, in third-person action, looking back at you.

Another problem occurs here, the editing process. Whilst it can cope with perfect collision sensitivity between the representation of you, and whatever is going on, on screen, down to the last pixel, what it cannot do is work out where you end and the room begins, if the background behind you is particularly cluttered, or you are not wearing clothing of a high contrast to the background.

In short, it draws a weird mutant figure with extra limbs, or bits missing, on screen, that shifts and distorts as the background changes. If the user of the webcam is in an electric wheelchair, the system gets so confused, it practically edits their entire lower section out entirely.

This does not even begin to touch on the situation when the user cannot stand to look at themselves in a living mirror, but, if that is the case, alternative input systems can be found, or created.

Untitled Document