Re: Why 3D

[P3D John W Roberts <roberts@xxxxxxxxxxxxxxxxx>]

>Date: Mon, 16 Jun 1997 22:57:30 -0500
>From: "P3D Dr. George A. Themelis" <fj834@xxxxxxxxxxxxxxxxxxxxx>
>Subject: Re: Why 3D

>>The only depth cue we get in stereo photos that's not in flat photos 
>>is parallax ("convergence" in the eye), 

>That's not true.  

Agree.

>As it has been stated many times here, convergence of the
>eyes conveys a very weak depth information.  

That has indeed been stated many times here, but there's never been a
consensus on it, in part because it's wrong! :-) :-) :-)
Counterexample: If I'm in a very dark room, with just a point source of light
in my field of view, I can make a very accurate estimate of the distance
from my eyes to the light source, with convergence as the only depth cue
(well, focus and chromatic aberration may be present as additional cues,
but convergence is the dominant one over a pretty wide range of distances
under these conditions).

There was some excitement a few months ago when the abstracts from a pair of
articles in Nature seemed to make the same statement you made. I read the
articles, and it turned out that they referred mainly to certain inacccuracies
in the coupling between retinal image position and eye position immediately
before and during saccades, which are *normally* suppressed, but which can
be made visible under specific controlled conditions.

>It is fusing the small
>differences in the pair through a brain function known as "stereopsis"
>that gives the sense of depth.  I think (it's been a long time since I
>studied this) flashing a stereo pair in the eyes (to produce after-images)
>is a proof of this.  In the after-images the eyes cannot change convergence
>since these are "etched" in the retina, but the brain can still see stereo.

That's certainly a major cue. Did the study check to see whether the eyes
shift their convergence while "viewing" the afterimage?

>>but when viewing a stereo photo, our eyes are focussed at a single
>>depth. I can easily see depth, in the real world, with one eye, by holding my
>>finger out at arm's length, and alternately looking at my finger and at
>>things in the background.  

>I don't know about you, but I cannot see depth like that.  I can tell that
>one object is perhaps in front of the other but I do not see any depth.
>This focusing is also a weak depth indicator.  

Unfortunately, long-time stereo enthusiasts are probably one of the worst
population groups to use as a test sample. Those of us who enjoy viewing
stereo photographs have long since learned to decouple accommodation and
convergence (or may have had less ability to judge depth by accommodation
than the general population in the first place). Conceivably the ability
to judge distance by accommodation may have actually atrophied to some
extent (while at the same time the ability to judge depth by image differences
and convergence should be improved).

>Plus, if the objects are
>beyond 10 feet, accommodation changes very little but stereopsis is still
>going strong.

Agree. Also, accommodation cues are largely a function of pupil diameter,
and therefore not constant (unless the brain can make a pretty good educated
guess at the current pupil diameter). [This statement applies if you are
trying to guess the distance to an object that is not in focus - if you make
the effort to focus on an object, then feedback from the eye muscles would
be the cue - but still with a smaller pupil it's harder to get a precise
focal distance.]

John R


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