Re: Stereogram on the "Stereo Window"

[P3D Larry Berlin <lberlin@xxxxxxxxx>]

>Date: Fri, 20 Jun 1997
>From: P3D  Dr. George A. Themelis comments:
>
>Greg Wageman is using the disparity to explain the stereo window.
>This explanation is easy to understand in projection (or the computer 
>screen) where the two images overlap.  In a viewer things are different.

****  No it's the same thing happening.

>...............
>I agree that the window is located at screen level and it is not moved.
>However it has moved with respect to the scene and changed in size.  

****** Yes it does change it's relationship to the scene as the scene is
moved. It's an effect of scale. Yes, one could choose to describe the
movement of the window relative to the scene, but recognize that in doing
so, the window itself relative to the viewer hasn't moved, regardless of
viewing method.

>
>Greg et al. say that the window is fixed in real space (and I am
>asking where?  7 ft., 15 ft., somewhere else?)

*****  At the plane of coincidence... which is relative to the viewing
method. It still exists in relative proportions for any stereo example. Pick
your image, define your scale, and figure out some specific meaninful number
if you need that. When they say that the window for Realist images is fixed
at about 7 feet, they are referring to the depth where the optics allow a
natural plane of conicicence within the camera/viewer geometry. You can
alter the actual perceived depth factor for that window, by rearranging the
chip, especially if moving a full frame chip within a 5 perf window! If you
try this you will see that the window, defined by the mount you choose,
remains fixed while the scene moves in relation to the window, while the
window redefines itself for each example by the new relationship.

Bottom line is it doesn't matter what the depth is as long as it works for
the elements in a particular image. Discussing a measure in percentage might
be more appropriate than numbers of feet. 

> and the scene has
>moved with respect to the window.  It is hard to accept that the
>rock has moved closer to the camera without increasing in size
>and depth and tha the mountains have moved in half infinity.

*****  They haven't, but you have stretched all the relative factors at once
by rearranging the chips. They remain proportionate to each other whatever
you do with your definition of where the plane of coincidence lies for that
image. You didn't move the camera, just the plane of coincidence relative to
existing image data. Of course that appears to change things.

>
>I accept the fact that by putting the window at 15 feet I reduce
>the separation of infinity points and have my eyes converge when
>looking at infinity.  However this does not make the mountains be
>any closer to me than infinity.  Convergence is overriden by
>other stronger cues.

*****  Convergence isn't a fixed finite thing, it operates relative to
elements within each image. When you stretch parameters, all elements within
the image change proportionately. As long as they remain available to your
eyes, you will perceive depth from the stereodisparities. Infinity is still
interpreted as infinitiy even though that place is defined in the image as a
flat plane existing at a finite distance. If you bring it forward, infinity
has gotten closer but the image still interprets relative to itself. 

Larry Berlin

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


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