[MF3D.FORUM:1012] Deviation, the stereo window, definitions

[Abram Klooswyk <abram.klooswyk@xxxxxx>]

Being somewhat of a definition freak, I'm a little upset :-)
by some of the terminology in discussions between David
Kesner, 
Paul Talbot, David Lee and Richard Twichell, which sometimes
seem a Deviation from the Only Logical Stereo Terms :-).

English phrasing in the following - and previous :-( - is 
probably not correct, but I hope the conceptual meaning is 
clear.

The stereo window is a frame in space through which the 3D 
image can be seen; 3D image and stereo window are cyclopean 
illusions which do not exist without a viewing subject. The 
window illusion is _prepared_ by (among other things) 
positioning of the film chips with regard to the mask 
apertures of the mount. I probably often use "stereo window" 
as a hard ware term, but in fact it remains an illusion.

The strict definition of deviation is something like: 
The Deviation of a pair of homologues on a stereo picture is 
the difference between the infinity separation on the stereo 
picture and the separation of that pair of homologues.

I suppose "homologues" and "separation" are not easily 
misunderstood. 

"Separation" is a perfect term to indicate all X-direction 
distances (left to right distances), lens separation; infinity 
~, far point ~ and near point separation (on stereogram), also 
mask aperture separation and on-screen separation. 
It seems a logic way to avoid confusion with Z-direction 
distances (from camera or viewing person to the scene), 
which can be called just "distance".  
In French exists a similar pair of terms: "écartement" and 
"distance", in German "Abstand" and "Weite", in Dutch 
(since Ferwerda) "separatie" and "afstand". 

The on-film infinity separation on uncut film from regular 
stereocameras is equal to the separation of camera lens 
axes (parallel axes assumed). 

This hardware property of the stereocamera is the basic 
measurement _from_ which other separations _deviate_, 
and it seems the only constant value in different pictures 
taken with the same camera. 
The "built-in window" of stereocamera's actually is the 
deviation of the apertures in the camera from the lens 
separation, but this "window" can be overruled in mounting. 

So "deviation" always applies to a pair of homologous 
points, which _deviate from_ the logical standard separation, 
which is the separation of infinity points, logical because 
that separation is set by the lens separation.

The deviations of points on the chips do not change in 
transposing, when they are seen as absolute values (no 
negative figures).

In strict use, "zero deviation" can only apply to infinity 
homologues. Also, speaking of "_the_ deviation" has no meaning 
unless a specific pair of homologues is meant. Most stereo 
pictures will contain several of such pairs with _different 
deviations_ from the infinity separation.

In looser usage, for the infinity separation on mounted 
slides, which is often not measurable, the far point 
separation is substituted, but this should be avoided
when confusion is possible. 

In stereo base discussions often the deviation of the mask 
aperture is used. Its separation is easily measured on the 
mount, and it is mostly known already. Then only the 
separation of infinity homologues or of far points have to be 
measured to compute the mask deviation on the mounted slide 
(or card). This is often the largest deviation of the slide, 
unless something gets "through the window".

Richard Twichell: 
>Deviation is determined at the moment the shutter clicks 

Yes, except for the mask aperture deviation, which is set in 
mounting.

>In projection, converging the chips on the screen to reduce 
>excessive parallax does not effect the stereo window, but 
>moves the projection window, which is the point in space 
>where the two projected frames coincide.

The chips are in the mount in the projector. Changing the
separation of the projection lenses "does not effect the 
stereo window", indeed it doesn't affect the position of
the virtual or cyclopean window with regard to the percepted
3D scene, but of course it does change the position in 
percepted space of the whole scene, including the 
position of the stereo window.

>excessive parallax
I would prefer excessive parallax _difference_. 
Binocular parallax is the angle between the lines of sight 
to a point in space. A similar description applies to the 
parallactic angles in stereo photography.
The difference in parallactic angles of two points can be 
too large for stereo viewing comfort.
(This definition of parallax is in accordance with the
original astronomic definition.)

>moves the projection window
I would prefer: moves the spatial position of the (cyclopean)
stereo window (to the place where the images of the mask
apertures virtually coincide). 
There is no stereo window without perception, and it is 
the only percepted window.
"Projection window" has no meaning in my set of definitions.

David Kesner:
>If I had two film chips where the deviation between the near 
>points is 2.0mm when the far points are homologous and I then 
>mount (place the window way forward) so that the far points 
>are 2.0mm apart it would spread the near points to 4.0mm. 
>There is still only 2.0mm between the near and far points, 
>but there is 4.0mm total deviation between homologous points.

Sorry, David, I cannot make sense of this in my usage of
the definitions of homologues, separation, deviation.
 
I daily conversation I also frequently use loose definitions 
and less precise formulations, but I believe that in 
>a good, thorough discussion of these issues
(Richard) we must stick to more strict definitions :-).

Abram Klooswyk