It has a depth limit, as well. It's truly a 2x2x2 cube. In my current setup, I can lean forward quite a bit, till the monitor is a little too close for comfor

It has a depth limit, as well. It's truly a 2x2x2 cube. In my current setup, I can lean forward quite a bit, till the monitor is a little too close for comfort, but I cannot lean back at all. In playing with it, I've noticed that there's sometimes a bit of latency in the edge detection; I'll lean back, do some eye-ing for a second or two and then all of a sudden it cuts off. It gives me the impression that they're not actually enforcing a cube per se, but that their technique has increasing error rates outside of a sweet spot, and they have an errors-over-time cutoff. I have a lot of sympathy for the difficulties in the implementation. The eye moves only a tiny amount, so you need a high-resolution camera to detect where it's looking with any accuracy (this is why they require USB 3). If the camera is already pushing the limits of inexpensive consumer camera technology, you can't easily expand its field of view without compromising accuracy. Here's a paper that talks briefly about the technology: http://www.tobii.com/eye-tracking-research/global/library/white-papers/tobii-eye-tracking-white-paper/ http://www.tobii.com/eye-tracking-research/global/library/wh...