It sounds like you don't have to physically move the camera; the lens physically adjusts to accomodate multiple viewing angles within the same frame of reference, but from a fixed point of perspective.
For those who aren't clear on what that means, think of "bullet time" photography in reverse. Instead of taking multiple shots of a single frame of reference from multiple reference points in sequence, the camera shoots multiple shots from a single reference point (you don't move the camera physically) from one viewing angle, just as normal vision would.
The upside is, you can "turn" an image shot in this fashion on any 2D display. In other words, your single frame of reference has in effect multiple viewing angles. I've seen QT VR files that simulate the same effect.
That data should be enough to produce a 3D image when viewed on a 3D display.