Nvidia 3d Vision

[mattwestwick]

It's not hugely complicated but it's a bit of a brain melter getting the calibration done right. I'm trying to make a version that is simple and optimised so it runs fast.
You simply have the coordinates of yer head position control the viewpoint of the character.
Basically you extend from the vanishing point from the source in the gameworld, keeping the trajectory straight as it passes from the virtual focal point, to the screen, then to your eyes.

It works fine for stationary objects in a virtual space, like all the ones we have seen so far. So I will try to have moving objects and see if it works equally well. It should be fine but it's trivial compared to a game engine. Although really it's no different from using the mouse and keys (crouch,jump, etc.) to move the player around.

Where the processing actually comes in is video latency. The little delay from taking the camera feed and processing it into a matrix can make it feel a little sluggish. If you use audio software it's like midi latency, the delay needs to be tiny so it can be used in real-time.

Since there are potentially massive amounts of applications using camera tracking I can see video latency soon being minimised greatly without great cost.

The downside I can forsee is that in all the Virtual reality 3d examples they show off by swooping round the room to emphasise how it works. But who wants to do that for more than 5 minutes. Getting a similar effect while still being able to sit back and relax would be challenging. This is what might make it a non-starter since it's a lot of work to implement what is largely a novel feature.