Re: [Phys-l] A geek's observations on "Avatar"

[Leigh Palmer <palmer@sfu.ca>]

Kyle Forinash wrote:

> Speaking of 3D- I went to the Siggraph meeting (http://www.siggraph.org/s2009/computer_animation_festival/index.php) in New Orleans this past summer and saw a very good 3D video playing on what looked like a normal flat screen TV WITHOUT POLARIZING GLASSES. You had to stand in a 30-40 degree angle from direct center and less than 15 ft away but it was definitely 3D without glasses. Anyone know how this works? It was very impressive.

I don't know for sure, but I long ago figured out how to accomplish something like that. It is very easy to do, but I have never got around to reducing it to hardware.

Have you seen 3D postcards? They can be viewed through a narrow acceptance angle like the TV you describe. If you examine them closely you will find an array (I'm tempted to say "grating") of vertical cylindrical lenses cemented over the picture. The picture itself is made up of two pictures, a stereo pair, printed in a manner similar to the old fashioned TV raster tuned sideways. One picture forms one field and the other picture forms the second field interlaced with the first, the resulting frame being a vertically sliced up pair of pictures in alternating strips. The slices are narrow enough (one pixel) that the missing half of each picture is not missed by the viewer. The vertical cylindrical lenses are two pixels wide and each magnifies by about 3x in the horizontal direction. The left eye sees one picture and the right eye sees the other picture. The perceived result is that the picture has depth. The cylindrical lens grating can also be employed with more than two interlaced pictures (and a different focal length) to display an animated image as the eyes are moved past the postcard.

A less impressive effect can be achieved more simply by replacing the grating of cylindrical lenses with a grating of vertical slits in a flat black sheet, very like the shadow mask in an old Sony Trinitron color CRT. The slits are separated horizontally by about two pixels. One can adjust the widths of the slits and the separation of the grating from the plane of the interlaced pictures to taste, varying the acceptance angle and the brightness of the perceived image. It is assumed here that the illuminant is incident from the viewer's side of the screen. The perceived picture will not be as bright, however, because the black parts of the grating occult part of the picture. It is likely that the contrast ratio will be poor for this scheme as well, because the grating can't be made ideally black. If the picture is illuminated from the rear, however, this can be quite an effective illusion. A high definition TV screen is an ideal object on which to construct this illusion, and I would guess that the TV demo you saw employed this scheme. It would work best (i.e. have a better contrast ratio) in a darkened or subdued light environment.

In the summer of 1980 I spent a month working at the Exploratorium in San Francisco. At that time there was an incorrect explanation on the wall of how the 3D and animated postcards worked, but I corrected it. Frank Oppenheimer taught me more than I taught him, however. It was a wonderful experience.

I have now done further testing on the RealD 3D glasses. The lenses may be popped out of their frames by force. Just wrap a handkerchief around the lower part of the frame and push outward on the bottom of each lens with your thumb. You need not worry about labeling the lenses. They are different; the right lens has a small "ear" on it, and I expect all right lenses have that "ear". It is easy to see by the asymmetry of the lenses (and perhaps their curvature) which side of each lens is the linear polarizer (the side toward the viewer's eye). I did determine that the right lens is a right-hand circular polarizer, and the left lens is a left-hand circular polarizer. (I am working hard to make up a mnemonic to help me remember this fact.)

I also examined an old pair of IMAX 3D glasses, the ones used for the Japanese computer generated films. They are simple linear polarizers canted + & - 45 degrees to the horizontal. I don't know what the projection scheme is for either the old or the newer IMAX 3D processes.

Leigh

(Not proofread - sorry. I apologize in advance, since it's late.)