Keeping Things In Perspective

Movies in 2D manipulate perspective, not depth per se.

The latest Star Trek movie 3D version was converted from the 2D master, so I watched it to see what marvelous or abhorrent transformation would be wrought by a high quality synthetic conversion.

Turns out, the retrofitting of 3D onto a 2D conceptualization doesn't do much of anything. J.J. Abrams is a 2D director, that is his cinematic eye. The 3D from the conversion is a presence without significance, in most cases depth without roundness, adding nothing at all and in some scenes, detracting.

This is a well crafted movie, with many examples of creative lens choices. A private tête-à-tête between Kirk and Spock, conducted on the command deck with crew all around, is made confidential by the use of a long lens that foreshortens the distance between the two heads in a tight shot. The camera had to have been far away to make the two actors seem so close together. Their faces are flattened, enhancing the guardedness of their expressions. But in 3D we are looking at the back of one head as a blur on a plane, and the front of the other head as a picture on a billboard.

Depth in the action montages is scarcely noticeable, although realistic. Quick cutting unavoidably slices and dices any perceivable volume into nothingness.

I am a devotee of stereo 3D used to tell a story, but this movie demonstrates that adding stereo 3D as a special effect afterthought is a wasted effort.

What is the heart of the problem? I don't think the issue is conversion technology, which can be quite good with adequate time and budget. After viewing many converted 2D movies, as well as numerous CG features originated as stereo 3D, I've concluded that the use of perspective in 2D movies, and the use of Z depth in stereo 3D movies, are antithetical to each other. They clash when employed together.

In 2D photography the combination of different lenses is a combination of different perspective croppings. The longer lenses employ foreshortening for certain dramatic effects. The wider lenses utilize vanishing point perspective for other dramatic effects. The projection of perspective onto a flat image is the artistic medium for 2D, geometrically.

Perspective does not show depth so much as it creates an impression of distances between objects as measured away from the camera. There are many other depth cues in a 2D image, such as the play of light across surfaces. Painting and cinema have established conventions that allow us to interpret the use of perspective and lighting as indicators of depth, and many movie sets since Casablanca have used these conventions in order to "cheat" a shot so that objects look more distant than they really were on set.

You could not fool the eye with that kind of cheat in stereo 3D. Why not? True depth perception is the product of parallax, not perspective. You can see depth in real life with only one eye open if you bob your head side to side, what some cats do in order to better gauge depth with a wider de facto eye separation. The side to side shift differentiates the depth of objects according to how much they move relative to each other, revealing the parallax differences between adjacent points of view.

Parallax shift is more apparent when seen from wide angle vanishing point perspectives. It is scarcely noticeable at practical lens separations for telephoto lenses. Look through binoculars at the tree trunks in a forest. You know that tree trunks are cylindrically round but through the binoculars they appear flat. The depth from nearest to farthest in view is apparent, but not the roundness of shapes.

Take your binoculars to the nearest railroad line and stand on the tracks, looking along them. This is a good demonstration of the difference between depth and perspective, for with 8x power binoculars, perspective lines scarcely converge to a vanishing point, and yet, you can see the flattened depth planes that are characteristic of binoculars. You could see more roundness if the binocular lenses were very widely separated, in the manner of old style gunnery binoculars.

However, unless one is simulating the view through binoculars, for good stereo 3D, long lenses should be avoided. It would be possible to make an excellent stereo 3D movie with only a matched pair of fixed focal length wide angle lenses, about 53 to 58 degrees horizontal angle of view (about 24 mm to 28 mm on a super-35 sized sensor). What a 2D director of photography does by varying focal length and camera placement in a motif set up for quick cuts, a 3D director of photography should do by varying camera separation (the stereo baseline) and camera rig placement in a motif set up for long takes on a mobile platform. This is the difference between manipulating perspective and manipulating depth.

Good 2D movie makers should just stay away from stereo 3D, and continue to do what they do, without dilution. Those who aspire to be good stereo 3D movie makers should abandon most of what is written in books about 2D photography. They should also study established stereo 3D techniques with detachment, feeling free to discover new variations on the old verities through brave experimentation.

The great stereo 3D productions are yet to come, and they will be created by young people who, because they were never 2D lensmen, have no set of best practices to forget in order to invent a new set of best practices for a fundamentally different medium.

Manipulate perspective or manipulate depth, make your choice. Used together, each detracts from the other.