T3D Amount of image overlap

["Dr. George A. Themelis" <DrT-3d@xxxxxxxxxxxxxxxxxxxxxxxxxxx>]

John B writes:

> Let's take the example Tom 
> already has in the spreadsheet.  Focal length is 50, near point is 30" 
> and far point is 40".  As we can see, we can shift the camera up to 
> 3.8" without exceeding the comfort limit for most observers. 

I will use this example to demonstrate something that I mentioned
earlier:

> at high magnifications toeing-in becomes necessary.  That's 
> because the translation required to produce a certain amount of 
> parallax (or on-film-deviation) is such that the field of view can 
> be totally different between the L and R pairs. 

John B asked me to clarify this with an example.

When you shift the lens parallel, the stereo window is set at 
infinity.  The film chips must be masked to set the window at
some closer distance.  This is true in all cases (normal photography
and macro photography).  In the case of a normal scene with infinity 
we know that the 1.2 mm OFD must be masked to bring the window at 7 
feet.  In close ups this is very important because the stereo window
is sitting past your most distant object, generating two bands of
non-overlapping image area that must be definitely masked.

The following formula gives the amount of on film masking needed to 
bring the window at distance I from the lens:

Horizontal Masking = F B / I = M B

Where F is the focal length, B the stereo Base, M the magnification.

For example, for F = 50 mm, I = 30" and B = 3.8" we get:

Horizontal masking = 6.4 mm.  So 6.4 mm are wasted.  From the 36 mm
of the 35 mm only 29.6 can be used, the rest must be masked out to
set the stereo window.  Wide RBT mounts cannot be used for this image.
7p mounts must be used. 

As the magnification increases, things become worse.  Consider the
case of M = 1.  The horizontal masking required is M B or in this
case equal to the stereo base.  So, if at magnification equal to one
we want to translate the camera by 12 mm then we will end up with
only 36-12 = 24 mm of useful image so we are forced to use a Realist
mount.  Some combination of near and far objects that predicts a
translation of more than 36 mm (while M=1) is totally useless from
a practical point of view because the two pictures will show a
totally different field of view!

The situation becomes a problem at high magnifications (hence tilt
is required) and is also a bit of a problem in aerial hyperstereos
where large F's and large B's are used.  We know that in aerial
photogrammetry a 60% "overlap ratio" is used.  This means that
40% of the image is wasted in order to obtain a good amount of
depth in the pair.  That's why aerial photogrammetric pairs are
usually tall and narrow.

I would recommend that in your spreadsheets you add a column that
shows either the "percentage of overlap" or the mm that need to
be masked to bring the stereo window at the near point.  And if
the amount of overlap is less than zero then don't even give this
as an acceptable solution.  Anything with less than 2/3 overlap
(66%) should carry a red flag because the user of 35 mm slide film
cannot use even Realist format mounts and is forced to 4p mounts.

George Themelis


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

End of TECH-3D Digest 386
*************************