P3D Re: stereogram infinity

[Larry Berlin <lberlin@xxxxxxxxx>]

>Date: Tue, 20 Apr 1999 
>From: "ron labbe" <ron@xxxxxxxxxxxx>
>..................
>Larry has an interesting point:
>
>>*****  It's true that a typical stereogram is built in front of a
>background
>formed by the tiled (repeating) pattern, but it's *not* true that infinity
>is at the window!
>
>Again, it's hard to define the window with an single image random dot
>stereogram.

*****  I agree that it's hard to define where the *window* is from strictly
a viewing reference, but in taking stereograms beyond the typical stereogram
image, to the point of creating an infinite stereogram viewable continuously
from parallel to crossed, as far as anyone cares to twist their eyes... it
does become obvious that the stereogram pattern in 2D is *at* what would
normally be considered to be the stereo window. With such an infinite
stereogram designed to be viewed continuously, it is difficult to view
anywhere close to the window due to the power of one's eyes wanting to lock
onto the 2D pattern. As a window shapes what can be seen beyond it, so the
bounds of a printed stereogram limits what can be contained within it's view.


>
>>............. A parallel mode stereogram MUST have a maximum
>pattern repetition distance that is equal to or smaller than the average
>interocular because anything larger and most people wouldn't be able to view
>it.
>
>Exactly- so stereogram "parallel" is not parallel.

****  Well, many of them don't have infinity at precisely the interocular
spacing, which means the viewing lines aren't themselves parallel, but the
infinity point in a given image still appears as an infinity point relative
to the rest of the image content (true of stereo pairs too). Since some
people can go *wall-eyed* or wide eye'd beyond parallel, it becomes possible
to have an infinity in an image that is greater than the interocular
spacing, yet that doesn't affect the definition of infinity.

It remains true that *parallel* viewing of a stereogram puts the apparent
image completely behind the printed pattern by simple viewing geometry.
Therefore the infinity point of that image might be defined as the most
distant point within the image itself. Since most parallel stereograms
define that distant point with the width of the pattern itself, it becomes
virtually impossible for infinity to be *at the window* for parallel
stereograms.

> Larry, you would not
>argue that it is impossible to have a regular parallel STEREO PAIR (where
>the window IS precisely defined) with infinity at the window... would you?
>
>A very interesting argument, indeed.
>

****  Well, it gets more interesting with your comments! It is possible with
an ordinary stereo paired image to set the stereo window at infinity. This
is precisely the result of the raw images from a two camera rig. One could
even postulate setting the window farther than infinity, but this is only
possible because of working with images instead of reality. The results may
not be very viewable.

In stereograms the repeating pattern controls the viewing environment. The
emptiness of distant space is filled with some sort of pattern. Space no
longer is transparent. However it's the viewing geometry that provides the
clue about recognizing that the window is at the 2D pattern and the entire
scenario of stereograms extends outwards from the window. For stereograms it
would seem one cannot move the window at all, though one can move the
infinity point. If you move infinity up to the window, you end up with a 2D
image that does not contain a hidden shape because as your pattern width
approaches zero size, it approaches the window and the 2D pattern. If your
pattern becomes zero width, you have no image and no pattern left either!

I looked at several cross viewable stereograms trying to define where
infinity is relative to the window, and again infinity is moveable. In
crossed mode, the entire image projects in front of the printed pattern. The
printed pattern remains fixed and immovable.The visible shape is in front of
what would normally be considered to be the window. Yet there is no visible
window to measure from other than the limits of what you can view are
defined by the limits of the 2D printed pattern. The window remains the
level at which the view lines cross over and at which one swithces from
crossed to parallel mode viewing.

You will notice that the direction of parallax in an ordinary stereo pair
crosses at the window. This corresponds to the viewing mode change for
stereograms, which is a reversal of parallax direction. 


Larry Berlin

Email: lberlin@xxxxxxxxx
http://3dzine.simplenet.com/


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