P3D Re: MORE two-color nonsense

[aifxtony@xxxxxxxxxxxxx (Tony Alderson)]

A major problem with this thread has been a sloppy use of language all
around--contributors are using vague terms, or improper definitions, and
mainly listening to themselves instead of listening to each other.

The first problem has been the definition of "full-color anaglyph."
Deering seems to mean reproducing the full range of the spectrum in the
image, including saturated reds, greens and blues. Zone and Carter seem to
mean an anaglyph that appears to be a color photo, even if the artist has
cleverly limited the palette to minimize the obvious problems with anaglyph
transmission.

Deering seems to be correct that a red/cyan color system will not reproduce
the full spectrum satisfactorily. It does little good for Dr. Carter to
detail subtraction color theory; the proof is in the pudding and so far all
of the samples shown clearly do not reproduce the full spectrum--the greens
shift towards blue and the blues shift towards green. Of course, not all
scenes have every color, so it's not always easy to tell the difference. It
may be that better results are possible, but an image that demonstrates
transmission of saturated, uncontaminated blues still has not been
presented to P3D.  The best I've seen are in some of the Songer examples in
the April 1974 American Cinematographer, but I doubt this blue is saturated
enough to satisfy Mr. Deering.  But Deering does not seem to be saying that
you can't get ANY kind of color anaglyph through red and cyan filters
(although I initially misinterpreted him in that way), just that you can't
get full spectrum fidelity.

It is in some respects an irrelevant point, however.  The best any anaglyph
can transmit is a two color system; unless we evolve a third eye. Try this
experiment: put on a pair of anaglyph spectacles. Look at the real world
around you. This is the maximum potential of anaglyph for the transmission
of both stereo and color. Now take your glasses off. Did your color
perception change? Is anaglyph capable of transmitting full-spectrum true
color? The very nature of the complementary filters and the retinal rivalry
means that it is not.

Put the glasses back on. It may be amusing to freeview a color stereo pair
for this part. With the image fused, do you get a color impression? Sure,
the color is altered, but there is color in the fused image. Look through
the red filter only. Aren't you seeing a red color separation? Isn't the
view through the cyan filter a cyan sep? ALL ANAGLYPHS ARE COLOR
SEPARATIONS. It matters not if the anaglyph is red-cyan or red-green or
red-green-blue or CMYK or grayscale. It is utterly meaningless to criticize
a red-cyan anaglyph vs. any other anaglyph because the red-cyan records are
color seps, as the same thing is happening in any anaglyph. Look at a color
anaglyph in Photoshop--the red channel is the red separation of the left
image. The green and blue channels are the respective seps from the right
image. Even in a monochromatic anaglyph, the left and right channels are
red and cyan  (or GB) seps of the grayscale images.  The problem with RC
anaglyphs vs. RGB anaglyphs is not that the RC are color seps, it's that
they are color seps that do not transmit the entire spectrum.

But it really doesn't matter when viewing the 3D image--through the glasses
all we can see are red and cyan separations, so no matter how the anaglyph
was made, we won't get any better color transmission than a two color
system. Also, because we view the anaglyph through complementary filters,
we can never satisfactorily see saturated reds and blues. Such colors cause
so much bombardment that they undercut, even destroy the stereo effect. So
it seems rather odd to demand saturated colors to define a "full color"
anaglyph when no anaglyph can transmit them anyway.

Now, I will concede that a major point of color anaglyph is how the image
looks with the glasses OFF, and in this respect a RGB anaglyph may be
somewhat superior to a RC anaglyph, but I suspect most people will notice
the fringes before they notice the inacurrate color. The trick in color
anaglyph is to balance the color perception without the glasses against the
stereo perception with the glasses. To get the thing to work at all, you
have to avoid certain colors, especially saturated pure colors. So no
successful anaglyph will be "full color."    (But "full color" is a vague
term--there were two-strip Technicolor films billed as "full color" to
distinguish them from movies that only had a color sequence or two...Just
as early sound films were billed as "all talking" even though the sound
quality wasn't that good.)  So, for anaglyphs that will work, a RC image
will do about all a RGB will do.  If you show a well-done RC anaglyph to a
lay person (not one of us P3Ders!), they will say, "Yep, that's a color
picture." They typically won't say, "Wait a minute, where are the deep
blues?"  RC seps can capture good flesh tones, and that is the most
important to most people.

Another example of poor use of terminology has been the dismissal of
two-color systems as "duotones." These are not duotones. I quote from the
Photoshop manual (ver 4.0, page 337):

"Duotones, tritones, and quadtones are grayscale images printed with two,
three and four inks, respectively. In these types of images, colored inks
are used to reproduce tinted grays rather than to reproduce different
colors."

This is not the same as red-cyan color separations.  Systems like
Kinemacolor may have limited palettes, but they are definitely color
processes.  A "grayscale" anaglyph is a duotone.

Deering should not be so skeptical of the Land red-green(white)
experiments. I haven't seen the pictures yet, but I suspect one can make a
picture that will fool most people in this way. (But we need to fool Tom to
collect, and that's a lot harder!) Green and blue color seps are almost
identical for most colors. Back when I was running an optical printer, we
commonly used the green sep for both the green and blue pass when printing
the foreground elements for bluescreen composites. Of course, by necessity,
the foreground element in such case has no blue!

So, to sum up, everybody's right and everybody's wrong on this one. Deering
gets to keep his $100 (at least until a better contrary example shows up);
Zone gets to keep earning his living making plausible fallacies.

Anaglyphs are one of those things that shouldn't work, but does; and should
work better, but doesn't.  (Philosophy teachers take note: an analogy for
the dialectic! ;-)  )

Tony Alderson
aifxtony@xxxxxxxxxxxxx




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