I do not quite understand. I mean, I comprehend that the image becomes 3D via our stereoscopic vision...is that what they mean by 'completely automatically'.
Anyhow, that is really amazing 3D for single photograph.
I guess what I am missing is 'the point'. Is the point that we can see a 3D image based off a single camera lens?
The amazing thing is, it is done completely automatically. i.e. user is not entering the hot-spots etc.
The program takes a flat 2D street photo, and walks you inside the streets with all buildings around you.
Intensive image recognition algorithm in place. Probably building the 3D by making assumptions of shapes, corners, angles, shades, lights etc.