P3D Re: The "rule" is not a rule at all

[abram klooswyk <abram.klooswyk@xxxxxx>]

In the recurrent Themelis-Deering Debate, always a
pleasure to read (:-)) Tom writes (24 Oct 1999):

>(...) where are those formulas that George recommends?
>One place is http://www.deering.org

When you arrive there and have found your way through
the link 'macros?' to www.deering.org/basis.html, you
read in a section with small print '1/30 rule-of-thumb':
(...)
>The 1/30 estimate of "maximum stereo" is close enough
>for many situations, but it is based on too many
>assumptions to be useful in every situation.
>It assumes a focal length of 35mm or so, and it
>assumes the farthest object is at "infinity."

I'm sorry to interfere in the debate, but "It assumes
a focal length of 35mm or so" seems a conceptual error.
I know (from the Archives) that the statement has been
made before on Photo-3d, when base calculations were
an issue, and that Tom Deering probably only repeats
what other scholars have said.

Still it is wrong. In a medium (or even large) format
*system*, so when camera and viewer focal length are
proportional to *that* format, 1/30 holds equally good
(or bad :-)) as with 35 mm camera's.

This is so because 1/30 essentially is a *viewing*
constraint, which is independent of the system used.
1/30 Rad, or 1/30 as tangent, represents 2 degrees of arch.
When you accept 2 degrees as a practical depth maximum
in viewing, than it also applies to drawings and anaglyphs,
and anything twin-view stereo.
The point is that a so strict limitation (2 degrees) applies
nearly only for projection, and even then not always (see
"double depth").

This was on only a minor detail of Tom Deerings math page.
A more important point is the fact that the so-called
Bercovitz equation assumes that you first have a nearest
point AND a farthest point, and from there you calculate
the base. The closer near and far point are to each other,
the more absurd is the outcome of the equation, illustrated
by the fact all curves become nearly vertical at the upper
edge of Tom's graphs.

This absurd part of the graph is exactly the math background
of George Themelis' objection to Boris Starosta's example of
a portrait.
(I'm sorry not to have math phobia, but to object to the
equation and its graphs on mathematical grounds :-)
"It's heavy math" - really? )

When I get the time, sometime I will show that many names
are attached to the equation, which has backgrounds dating
back to the times that none of us were born, and show that
it can only be used as a guide in a limited number of
cases, not unlike the "1/30 rule", when followed blindly.

Abram Klooswyk

Postscript
>the 1/30 "rule" works only if your camera has a
>certain lens spacing (Tom Deering, P3D a bit of
clarification, 25 Oct 1999)
This is too quick a reaction in a discussion on your
shared pet peeve, Tom. You know that the Bercovitz
equation becomes equal to the 1/30 rule when the farpoints
are very far or at infinity. 1/30 works fine for landscape
hypers with two camera's, having *no* certain lens spacing.

Abram Klooswyk