Re: Color from B&W

[P3D Richard Young <young@xxxxxxx>]

In the original Land (1959) experiments he took
a picture of a colored scene through a green filter onto black-and-white film,
and projected it with no filter, creating a white-and-black display. He
then took a picture with a red filter over the camera, and then projected
through the same red filter with a second projector, superimposing the two
images. Observers see a wide range of hues, including blues, yellows,
greens, reds, white, and black (although a few saturations and brightnesses
for some hues are missing).

The same technique can be done with only one projector (or one CRT, or one
liquid crystal display, or even 1 printed image), by using alternating
lines of the display for the red and white channels - the same way as is
done with 3-D stereo TV for example, but using the odd and even lines for
red and white instead of the left and right eye images (Young, 1990)

The basis of the effect is the way the cells in the brain process
wavelength and intensity information so as to see color. About 90% of this
information is processed by a "white/black" channel and a "red/blue"
opponent channel (Young, 1986), similar to the Y and I vectors used to
transmit color TV in the Y-I-Q system. Land effectively stimulated the
white/black channel and the red half of the red/blue channel, and so got a
wide range of colors.

By also including the blue portion of the red/blue channel, an even wide
range of colors can be seen with only two channels (Young, 1990). This can
be simulated on the CRT by subtracting the red and blue signals to the
monitor, and then stimulating the red gun when the difference is positive,
and the blue gun when the difference is negative. Or, it can be done
directly in a single channel by using a red/blue dichroic polarizer filter
on top of a liquid crystal display with a rear polarizer, alternating odd
and even lines with a white/black polarizer for the second channel (Young,
1990). This "opponent color" method is potentially a cheaper way to make
color video displays for say laptop computers.

Color can also be obtained from a SINGLE white-black channel,  through
temporal and spatial variations of the white light, an effect first
presented in the nineteenth century and known as the "Benham's Top"
illusion, where
black and white stripes in proper locations on a spun half-black/half-white
disk cause unmistakable colors to be seen (Young, 1977). This effect was
used in the 1950's in California to show a Squirt commercial in color on
everyone's black-and-white TV's (at that time). Oddly, many viewers phoned
in to complain that they were seeing color on their TV's and wondering what
was wrong with their sets.

It is even possible to obtain specific colors with ZERO light channels, by
stimulating the eye nerves with weak electrical pulses creating
"phosphenes" (do not try this at home) (Young, 1977).

E. H. Land, (1959) "Experiments in color vision," Scientific American, May.

Young, R. A.(1977)  "Some observations on temporal coding of color vision:
psychophysical results," Vision Research, 17, 957-965.

Young, R. A. (1986)  "Principal-component analysis of macaque lateral
geniculate nucelus chromatic data," J. Opt. Soc. Amer., vol. 3, pp.
1735-1742.

Young, R. A. (1990) "Getting more for less: a new opponent color technique for
two-channel color displays", SPIE, vol. 1250.


Richard A. Young, Ph.D. |:)                       Phone: 810-986-1471
Staff Research Scientist                          GM:8-226-1471
Manufacturing and Design Systems Dept./294 RANB   FAX:   810-986-9356
Mail Code 480-106-285                             e-mail: young@xxxxxxx
GM R&D Center, Bldg. 1-6
30500 Mound Road, Box 9055
Warren, Michigan 48090-9055  USA

Q. What did the Zen Buddhist say to the hot dog vendor?
A. "Make me one with everything."

To be viewed in stereo:

  .   .   .   .   |  .    .  .   .
  .        .      | .          .
     @@@@@        |        @@@@@
  . @ o o @    .  |   .   @ o o @  .
 .  .  =    .     | .  .     =   .
   .   .       .  |  .   .        .
 .         .      | .        .

Hint: fuse both the faces and the "stars".
Looks better with crossed rather than parallel viewing.
(from Jennifer Leong, N.Y.U., with modifications)



------------------------------