Crap implementation.
The labyrinth app has used gyroscopes and the accelerometer for a neat 3D effect of its board for ages. I know this is just a proof of concept based on real time head tracking, but it would be sooo much better if it was a proof of concept for combining the two approaches.
Secondly, the face detection is crap.
It is really jittery. Smooth it out a little. Holding the camera perfectly still and myself perfectly still, it still thinks my head is oscillating around like a bobble head.
Also, it thinks my armpit is my face more often than it thinks my face is my face >:(
Finally, why not make something cool like a 3D camera that auto stitches two images like most panorama apps that auto align images?
This way you could show a 3D effect based on gyroscope and head position, and there would be some use to this thing...
tuesdaythefifth about i3D, v1.0