P3D Re: Pass the Crow, please!

[Tony Alderson <aifxtony@xxxxxxx>]

Ray Hannisian wrote:
> As for whether anaglyph looses detail, (and thus depth cues), due to
> color manipulation, or if my perception of "cardboarding" is produced
> by perspective distortions or other artifacts, I would appreciate
> further discussion on these matters.

It seems to me we've been partly talking past each other. Two different
effects are happening here, which are both being described as
"cardboarding," but while they may look similar, they come from
different causes.

If we could make grayscale anaglyph and StereoJet prints of Heidi with
the hat at the same size as the printed color anaglyph, and compare them
side by side, we would see the same shape in all three images. All would
"cardboard." This is independent of the mode of stereo, and has to do
with photography and viewing.

But Ray H. and Gabriel are correct, there is something else going on.
When Ray first mentioned "loss of detail," I misinterpreted him to mean
"loss of resolution" and missed his point. What he meant was, in a color
anaglyph, there can be image data seen thru the cyan filter that is not
seen through the red filter (and vice versa). So some image points can't
be matched between the views. Which means some of the stereo image is
undefined; the stereo becomes ambiguous. If the mismatch becomes too
great, the anaglyph cannot be seen in stereo, just as retinal rivalry.

But usually, there are enough common points to define most of the image,
so our brain can try to resolve the ambiguities. If there are other
horizontal points close enough to fuse, we can see some false stereo. We
can try to "patch" the data together by fusing the edges. Commonly, the
undefined area will flatten, defined only by the whatever stereo data is
available. This can look "cardboardy," but it's not the same as the
perspective cardboarding.

What's really happening is there is a "hole" in the stereo image.
Imagine we have a stereo slide, say of Audrey, opposite the Celica ad,
and it looks great in our viewer. Now, by some means, we remove some of
the emulsion from each slide, on homologous areas in her stomach.
Looking at this modified pair, she isn't "cardboarding," there is a hole
in the stereo image. Having carefully saved the emulsion removed
earlier, we restore one of the slides, and view this pair. Well, maybe
we can bridge the gap a little better, but there is still a
stereoscopically ambiguous area. This is analogous to what Ray and
Gabriel are describing, and "backing off" won't cure it; you've got to
either put up with it (in favor of the color), or change your method of
transmission.

I don't think this is much of a problem with the Hat picture, as Heidi
has a pretty good tan, and shading shows up in both images. But this is
a noticeable issue with, for example, the aforementioned pic of Audrey.
This is a real contextual thing, but certainly limits the images that
can "work" as color anayglyphs.

Enough for now

Tony Alderson