Some basic 3d language from my own observations
Experiences shooting with various 3d rigs for native stereography
The use of very narrow depth of field to focalize our attention, the creation of "depth planes" together with the slightly strobing 24 frames per second "filmic" look, have all contributed in some way to our suspension of belief and identification with the story being told to us. Much of this, indeed, has been what we have perceived as the so-called "theatrical experience", giving us a heightened sense of immersion in the stories unfolding in front of our eyes on the big screen, letting us believe the unbelievable, especially in the cases when the script, storyline, cinematography and direction have been highly effective.
However, with the advent of 3d, there is a whole new added set of tools, and I believe that it can become stimulating to tell stories in a different way, as some film directors have already discovered. Thanks to an added sense of depth now available, it doesn't necessarily make sense to create the same "artificially" anymore; in fact, I have seen from personal experience that it does indeed look a bit odd if we do try and make a 3d film in the way that we have always made traditional 2d "flat" films in the past.
I'm inclined to think that the relatively flat perception has much more to do with the large sensor formats that current 3d films have nearly always been shot with, with their resulting shallow depth of field in the imagery. Using the Cinemascope 2d widescreen format of 1:2.39 can also lead to less stereo perception than 1:1.85, as cutting out the ground and sky (depending on framing) also can mean that we perceive a less three-dimensional image.
So, and without pretending in the least to offer an exhaustive explanation on the subject, here are some of the considerations that I feel I can make, following on from my own experiences in native 3d film production and post, and which may, I hope, prove useful to others about to embark on filming in the format.
Starting from the basics, panning and tilting the cameras, which are essentially 2d movements themselves, don't look all that hot in 3d unless used for minor framing adjustments on Steadicams, Techno Cranes, Jibs and Track moves, as per dynamic zooms.
On the other hand, three-dimensional movements of the cameras do indeed look much more interesting, as we are stimulated by continuously changing depth cues while we watch them, so we actually perceive more depth. Leave a 3d shot with the camera locked off for a while, and it will soon look as if there is little or no 3d, and which, of course, could also be desired - the best stereography should always not be noted at all by the public as its there to help the storyline and not become an interruption to it.
Another example; if we see a wide establishing shot with two people talking, we can see how far they are from each other in 3d space, so when we cut between them in close up's, there is not really as much need to show the shoulders and backs of heads out of focus so that we can see where they are in 3d space, as we have already seen where they are in relation to each other anyway. Also, objects or backs of heads in close up and out of focus look odd coming out of the screen into the theatre - in a logical sense, we should be seeing them in focus in the theatre.
The accurate perception of depth means care must be taken with action shots where stuntmen are fighting, for example. We can quite easily see tricks that might have worked credibly in 2d, like punches that appear to hit home on the adversary, but now, because we can see where things really are in three-dimensional space, they need to be achieved differently.
The timing of scenes and shots changes, in as much as 3d stimulates and motivates a longer look at the surrounding scene. In this sense, I am personally convinced that the best approach is to edit in 3d, and not in 2d first. If there is a 2d version, then this may well need a completely different edit, and I think that it should be done afterwards. Of course, there can be exceptions to this, especially where the underlying essence of a story-line needs to be “discovered” before attempting editing of the 2d and 3d sequences.
I had mused with Dante Ferretti, the world renowned production designer, about the fact that he must have carefully chosen the many beautiful and refined objects for the scenes in Martin Scorsese's recent 3d film Hugo Cabret, and yet, I mused, haven't you ever thought that we almost always see these objects out of focus in the finished film. He laughed. I said that it's as if there were lots of large sheets of perspex smeared with generous layers of vaseline, hovering in front of the objects so that we can't see them in focus.
I'll never forget a test I ran for a director with both a shallow depth of field camera and a small sensor second camera, which was totally in focus. He pointed to it and said "I want that!!" as, with everything in focus, we could scan the entire scenario with our eyes, and that in itself was the story line!