P3D Re: mounting to infinity

[Larry Berlin <lberlin@xxxxxxxxx>]

>Date: Mon, 8 Dec 1997
>From: John W Roberts  writes:
>...............
>
>>Date: Sun, 7 Dec 1997 02:45:56 -0700
>>From: Larry Berlin <lberlin@xxxxxxxxx>
>>Subject: P3D Re: Hypo, Hyper, Ortho
>
>>>From: John W Roberts  writes:
>>>(I would even venture so far as to speculate that to be truly "ortho",
>>>a 3D photo should be mounted to *infinity* (if it contains objects that are
>>>at "infinity"), rather than to *the stereo window*. I'm aware that mounting 
>>>to infinity can cause serious problems for projection.)
>
>>*****  Do you mean that a mounted pair of prints would appear to have the
>>image projecting completely above or in front of the prints?
>
>No, I mean mount it so that objects which are at "infinity" in the photo
>(e.g. the moon in the background of a landscape) also appear to be at infinity
>(viewed with the eyes parallel) when seen in a viewer. 

*****  This implies slides, not prints. In slides it's important that the
infinity point exist somewhere beyond the window. Therefore it's important
to consider both the stereo window AND the infinity points. A hand held
viewer seems to simulate the stereo window itself by placing the viewer's
eyes near the window and peering through. I feel that a properly mounted set
of slides would look good in both the hand viewer and in projection. I have
trouble envisioning a scene where I ignore the window in favor of the
infinity point and get desirable results, unless window and infinity are in
fairly normal relationships anyway. The human eyes can locate and interpret
infinity relative to the contents of the scene, so an absolute adherence to
a fixed infinity separation seems almost pointless, especially if it throws
the window to an awkward place. I would only worry about it when there is
too great a total depth to the scene, which could make the infinity point
difficult to rest the eyes on. In such cases a compromise is in order which
wouldn't necessarily be ideal by any method of interpretation.

>You would then let
>the objects in the foreground fall where they may, instead of the much more
>usual method of trying to mount the photos so the nearest object is at the 
>stereo window (and letting the background fall where it may). The angular
>convergence of the eyes to converge on any particular object in the field
>of view would then be much closer to what it would be in "real life" than it
>would for a stereo-window mounted photo for which the real and virtual
>infinities didn't happen to match up.

****  The latter instance is more workable since image interpretation is
relative to the relationships within the image. If virtual infinity isn't
the same as real infinity, so what? Unless it's far removed, and then other
problems exist too. This is far more critical an issue for hand held
viewers. It might for instance be difficult to converge on items coming to
or through the window, since that by definition in a hand held viewer is
practically on your nose! However, composition for projection, freeviewing,
anaglyphic, or LCS viewing is far better by paying close attention to
relative window placement. For LCS viewing by either page flipping or
interlacing, the key subject matter always looks better if it's at or just
behind the window. When it's behind the window too far, ghosting begins to
be a problem. 

LCS viewing is like placing the viewed objects in space and floating on your
computer screen with you watching them from a couple feet away. The geometry
of hand held viewers would then be comparable to placing your eyes at the
same depth as the computer screen. That would make it impossible to view
anything if the computer screen was supposed to provide the scene! Objects
behind the screen depth have greater parallax, which if certain colors and
high contrasts are present causes ghosting.

Lest that discouage anyone from trying LCS viewing, the results when
properly prepared are very worth while!!! 

>
>As noted before, mounting to infinity can cause significant viewing problems
>if the image is *projected* onto a large screen, causing the disparity of
>the projected image to become greater than the normal human interocular.
>Also, I don't believe that stereo photos should always have as much
>orthoness (be as orthoful?) as possible - but for those who really want
>ortho, I expect that mounting to infinity would be useful.

*****  Mounting to infinity, as I understand it from your discussion here,
then is like taking pictures and ignoring the lens of the camera. The stereo
window is the plane where all elements coexist equally in both the left and
right view. It's a pivot point, where all lines of vergence cross. Move your
head and all relational lines pivot like mechanical hinges at the window,
unmoving only at points that exist on one plane in the visual space. Whether
it's stretch from distance or moving from side to side, all factors relate
fixedly through the window. Mounting to infinity such that a projection of
the image causes too great a disparity is the equivalent to having the
picture be out of focus by being too far or close to the lens. For stereo
images of any type, it's imperitive to understand and use the stereo window
in all mounting and presentation planning. There are few times that ignoring
the window can be considered a safe assumption.

>.......
>P.S. I would prefer to call the procedure I described "mounting to infinity"
>(or perhaps "ortho mounting"?), and the procedure you described "setting the 
>stereo window to infinity".

****  It's true that my original question gave the example of mounting the
window AT infinity. That was to clarify what your intent was, since it
wasn't clear to me at the time. Now I understand your process. I would
propose that it's incorrect to classify mounting to infinity as
automatically ORTHO, since any ORTHO reconstruction could not occur unless
the window is properly positioned relative to all image content, both near
and far objects. If an image has too great a total depth for instance, NO
position of mounting will give 100% Ortho results. Since my eyes can allow
for infinity at a wide range of actual positions and my mind interprets the
scene relative to all these inner relationships, it seems mounting relative
to the window is the more accurate method. If images are mounted with
awareness of the window in mind, like getting a photo focused relative to a
lens, it should work better for either a hand viewer or projection.

Larry Berlin

Email: lberlin@xxxxxxxxx
http://www.sonic.net/~lberlin/
http://3dzine.simplenet.com/


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