P3D Permissible Depth Range (was: "Too much depth")

[Chris Jones <c.jones@xxxxxxxxx>]

At 14:04 31/10/99 -0700, Abram wrote:
>If I understand Chris Jones right, I'm afraid I disagree to
>some degree.

OK, I'll try and clarify my understanding of the issue.

>There is no boundary at 2 m (in either mode) for any depth
>cue.

There isn't a boundary in the sense that there is a switching from one mode
of stereopsis to another. But it seems that different mechanisms dominate
at different distances.

For distant objects, the eyes are nearly parallel and each eye sees an
almost identical image of the object - stereopsis arises from the
horizontal disparities of the images.

For close objects the mechanism appears more complex, and appears to rely
more on vergence movements, proprioception (here the feedback of eye
position) and some implicit interpretation of the object's structure.

It is the latter case that is so hard to mimic because some of the natural
cues are present in stereo photography and some are missing.

>Children and myopes recognize the depth of tiny wrinkles
>when holding a hand at 10 cm before the eyes, also when
>the eyes are fixed at one point, without eye movements.

For very close objects, you can get a pretty good idea of depth just from
shadows and perspective through a single eye.

>(That stereopsis is independent of eye movements is known
>since the mid 19th century).

It may have been what was suspected but does not appear to be the case. For
a much clearer (and more detailed) explanation than I have been able to
provide, you might wish to try reading an article "Neural mechanisms
underlying stereoscopic vision" by F Gonzalez and R Perez (Progress in
Neurobiology, 55 (3) 191-224, June 1998). Non-biologists are advised to
avoid the latter part of the article! :)

(if you are unable to find this article I can provide other references)

>When surface points of an object are seen sharply by both
>eyes, stereopsis works, at any distance, unless the distance
>is too *large*.

I agree entirely - but I do think that the mechanism by which people
perceive depth is dependent on the viewing conditions, in particular the
distance.

I would therefore expect that producing a "natural" stereo image is not
just about getting good-quality, distortion-free images, but also about
trying to give the brain as many of the depth cues that it is used to as
possible. Which is harder for close images than distant ones.

I hope I've explained myself clearly - if not, or if I'm simply talking
rubbish then please do let me know! I'm quite new to the subject but trying
to learn :)

-- 

Chris Jones 
http://www.c.jones.cwc.net
ICQ #41744518
DALNet nick trickydisko
PGP key available on request