One simple reason (among others) of why 3D has failed to take off as technology, and the movie industry is really struggling to capitalize on its investment, is the following: the viewer does not have voluntary control on the Depth of Field.
In other words, you may have a picture in front of you that gets interpreted in 3D in your brain and you can really understand that there are some objects in the back of the frame and some that are close by, but... you cannot decide to focus either on the near or on the far. You necessarily need to follow what the camera & the director decided to do! If the director wants you to look close, the far end will appear blurry, and vice versa.
This is not the same in real life. Yes, indeed when you focus on an object close to you, the ones on the back will blur out, BUT whenever you decide you switch your focus to the distant objects and now the near ones are blurry.
The difference of course is the voluntary control, and it's indeed quite difficult for the movies to replicate that. You would need to have everything on the frame appear crisp/focused so that the user can focus wherever he wants, but I suspect that would ruin again the effect, as we're not accustomed to see everything crystal-clear. The only solution I can think of is using Lytro-type cameras to capture all depths of field when filming the movie, and then transitioning to a personalized viewing experience by the use of VR headsets like Oculus Rift. These would need to be enhanced with some retina-tracking technology to understand where we're trying to focus on adjust the DOF in real-time (and it would need to be really-really quick).
In terms of technical capability this might not be that far in the future, but it is quite different from the approach the industry has followed, and thus difficult to deploy en mass and to market. And in the end, the differential gain won't be that significant to justify the cost.
