[MF3D.FORUM:635] Re: Image overlap

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

>I got to wondering whether part of Bill's problem is not
>enough image overlap on some of his wide base shots.  How
>can he calculate that (as opposed to just measuring it
>after the shot)?

The amount of the "unrelated" scene
 is equal to the deviation between near point *and infinity*
or, in my terminology FB/I (F: Focal length, B: Stereo Base, I:
distance of near point).  As you come closer (I increases) then
the unrelated scene increases.  Large focal lengths and stereo
bases work against you too.

A close-up of a relatively flat object (necessitating use of a
wider base) with long lenses (because the subject is far
away) is the worse combination in ordinary stereo photography.

Another way to write this (useful in ultra close-ups) is M*B
where M is the magnification.  If you are working at M=1
and you shift the camera by, say, 10mm, then you just
"lost" 10mm of image overlap.  If you shift 36mm with a 
35mm camera at M=1 then the right and left views will be
totally unrelated (not a very appealing stereo pair).

As you come closer and closer, image overlap (or lack of
it) becomes a real problem and you are forced to use
tilt to record the stereo pair.  The good news is that 
keystone distortion is reduced because
the effective FL becomes M*F (M: magnification) so it is
really large.  SEM stereo pairs with tilt at 1,000 or more
show no keystone distortion from a practical point of view.

In aerial photography I think they aim for 60% overlap
between successive stereo pairs.  If you think of a 
picture right down from an airplane, you can visualize
how F and B are related.  If F is large (narrow angle)
then a smaller traveling distance of the plane (smaller B)
will give you the same 60% overlap.

George Themelis