3D Or Not 3D, That Is: The Question July 2007

Sfumato is a lovely word. Nausea is not. Both are associated with human perception of imagery in depth.
It's commonly thought that two eyes are required for the visual sensation of distance, but that's not entirely true.

We live in a three-dimensional world--all of us, including those who are blind in one eye. We also commonly mentally calculate depth when viewing two-dimensional pictures--on computer, video, and movie screens as well as in books, magazines, newspapers, photos, and museums.

Our strongest depth cue is probably occlusion. It works in all visual media and for all observers. Consider two dots of apparently equal visual size. Which is farther away? It's the one that gets covered by the other (as in a solar eclipse, for example).

Another depth cue is size. Which horse is supposed to be closer to the observer in a painting? It's the bigger one.

Then there's geometric perspective. A vertical line with arrowheads pointing towards its center can appear to be like the far corner of a room; one with arrowheads pointing away appears like the near corner of a building seen from the outside.
Sfumato, or aerial perspective, may be found in the paintings of Leonardo. Distant mountains are hazier and bluer than closer ones.

The moving-image media offer still more distance cues. As a camera travels past something, that object must have been closer than something still up ahead. Our two eyes offer slightly different views, but those same two views may be seen by a one-eyed observer who simply moves that eye laterally a distance of roughly 2.5 inches.

A number of so-called "3D TV" systems have relied on such movement. One of the more recent simply moved the lens iris instead of the camera.

The NBA All-Star Game that was recently shot stereoscopically was hardly the first effort at imaging video in depth. <I>Business Week<P> reported in 1953 on 3D television broadcasts in Mexico, and the first 3D video was demonstrated decades before that. 3D cartoons have been broadcast in Japan. The Tournament of Roses was carried in 3D TV years ago; so were portions of the Super Bowl and a Rolling Stones tour.

Each 3D TV revival seems to come with the advent of new technology. The 1953 broadcasts required large polarized filters in front of TV screens and prismatic glasses. Color TV allowed cheaper glasses and no covered TV screens, though the results were often poor.

A quarter-century later, a filtered, guillotine lens iris offered simplified shooting, some measure of compatibility for 2D viewers, and a brightness balance achieved by the selection of green and magenta as the view-separating filters. It could be worth examining that last aspect in greater detail.

Colored 3D glasses have often used red and blue filters. In traditional color TV, red contributes 30 percent of the brightness and blue only 11 percent. The remainder is green. The huge imbalance between red and blue was thought to contribute to the discomfort of 3D viewers.

Today's digital era seems ideal for another 3D TV revival. With bit-rate-reduction systems, the additional information required for a second eye's view is a tiny fraction of that of the first. Electronically shuttered glasses can perfectly separate the images and offer full color. Small cameras simplify shooting. And there's no unequal brightness to cause discomfort.

Unfortunately, that's not necessarily the only physiological concern. The 3-D cartoons in Japan made use of something called the chromostereoscopic effect. Try generating some red text over a blue background, and the words may appear to float off the screen. The simple lenses in our eyes need to focus closer on red than on blue, and feedback from the focus muscles suggests to our brains that red objects are, therefore, closer than blue.

Adjustment of an eye's lens focus is called accommodation. Another musculo-visual mechanism is convergence--the aiming of our two eyes at a particular distance. It, too, offers feedback.
There are a few 3D display systems, notably holography, that allow humans to accommodate and converge on distant objects just as they would in the three-dimensional world. In most, however, images appear on a flat plane. Accommodation feedback indicates everything is at the same distance; convergence feedback contradicts it.

Is that difference sufficient to cause nausea for long-duration 3D TV viewers? Perhaps that is the question.Mark Schubin is an engineering consultant with a diverse range of clients, from the Metropolitan Opera to Sesame Workshop.