3D Photography Primer

Hold out your hand in front of you and look at it first with one eye closed, then with the other eye closed. The hand shifts position relative to the background. It is this shift in the relative position of objects at different distances that enables your brain to construct a representation of the three-dimensional layout of the scene in front of you. The brain receives additional clues from the control system that trains and focuses the eyes, from the apparent size of familiar objects, from shading, as well as from other sources. But the crucial information lies in the different spatial relationships among the projections of the same set of objects onto the two retinas. In order to make a photograph appear three-dimensional, we thus have to furnish the eyes with a pair of appropriate images. The way to do this is to take pictures of the same scene from two vantage points and present one picture to each eye. This will create the visual impression one would have if each eye were placed at the vantage point from which "its" picture was taken and were looking in the same direction in which the camera was pointed.

Stereo photography was all the rage in the 19th century. The pictures were taken with special cameras featuring two lenses side by side and were shown in little binocular viewers. Cameras and viewers, as well as scores of the old stereo cards, are still around. There are even a number of modern stereo cameras on the market. These cameras are all designed to produce a "natural" stereo effect by having their lenses separated by approximately the distance between our eyes. Issues of negative size and perspectival shifts aside, this is good as far as it goes. But it doesn't go very far.

Our stereo vision peters out at around 100 yards, when the differences between the two retinal images become too small to be resolved. In order to extend stereo vision further, the vantage points of the two pictures have to be pushed further apart. One can do this by simply moving one camera around. But if the scene changes in the meantime, the effect can be confusing (witness the unnaturally shimmering water in front of the Manhattan skyline or the flickering window cleaning elevator on the World Trade Center behind the Woolworth Building.) I prefer to use two cameras tethered by remote controls. In the picture of the Manhattan skyline, the two camera standpoints were about 40 yards apart. Our eyes are too close together to allow us to see Manhattan in three dimensions if we look at it unaided from across the Hudson river. Looking at the photographs instead gives us the perspective of a giant with eyes 40 yards apart and stereo vision extending out far beyond the city. The camera separation in the other pictures is smaller; in each case it was chosen for the desired effect.

Looking at the photographs cross-eyed is a lousy surrogate of the real thing. My preferred viewing method has one eye look directly at a print and the other at a mirror image of a print in an appropriately placed first surface mirror. This method is vastly superior to the old stereopticon viewer because it takes care of variation in people's eye distance, accommodation habits, as well as convergence preferences. In addition, it avoids any lens distortion, and it handles prints of any size. I have used this method in my various 3D-exhibitions.