This would involve filling gaps between the individual street view photographs which are spaced several metres apart, during which the environment might have ch

This would involve filling gaps between the individual street view photographs which are spaced several metres apart, during which the environment might have changed significantly. I suspect it would be quite the challenge: much harder than making 3D buildings out of several aerial images. If this were feasible, I would have thought Google would have done it already so that you could fluidly navigate the street rather than being confined to fixed points.