[MF3D.FORUM:1132] Deviation and Viewers

[Abram Klooswyk <abram.klooswyk@xxxxxx>]

Michael K. Davis wrote:
(...)
>Having studied John Bercovitz' General Solution ... 
(...)
>I have modified it a bit to borrow from David Lee's formula 
>which takes the ratio of viewer lens to taking lens into account.  
(...)
>being a little bit wobbly about the value for d - deviation. 

Beware of discussing General Solution!  Some lists have
gone very nasty doing it. :-)

I could write a book (O.K., a slim one) on base calculation 
and it presumptions. For one thing, the socalled Bercovitz- 
Spicer formula goes back to _very_ similar formulas and 
principles more than a century old, at least five names
should be attached _before_ Bercovitz in order to get a
justifiable name for that formula or equation.

Leaving that for now, I believe you are heading in the
right direction. The only, repeat only, rational for
limiting deviation is a _viewing_ constraint.

It is believed that sticking to a swing or _difference in 
convergence_ of two degrees (between looking at far points and 
looking at near points) generally is a wise limit for
_viewing_. 
As the tangent of two degrees is about 1/30th, the deviation 
limit should be about 1/30 of _viewer_ focal length.

The funny thing is that in several rules and equations the 
viewer focal length isn't even mentioned. Freeviewers however 
know that freeviewing of images with larger deviations is 
easier at a longer distance, which is easy to understand when 
you realize that the swing in convergence from far to near 
point becomes smaller at larger distances.

The next point is that the two degrees limit isn't a law. 
Experts seem to agree that in a series a few slides requiring 
_four_ degrees swing will do no harm, especially when they 
contain subjects at many different distances, gradually 
receding in depth. Earl Krause and Koo Ferwerda were in favor 
of (although limited) use of "double depth" mounting, placing 
the window at half the normal distance. 
I believe that for some selected subjects, seen in viewers, 
even some eight degrees can be tolerable. But I agree to two 
degrees as the standard. But it seems not necessary to 
compute deviation to a hundredth of a mm :-).

Now for some esthetics. From the viewing swing limit you can 
calculate a deviation, from there with camera focal length, 
near- and far point distance you get a base value. 
Remember this is a maximum recommended value. When you use 
near point 8 feet and far point 8.5 feet, the computed value 
will be almost always too large, from an esthetic point of
view. It will cause unpleasant deformations in viewing.

Indeed, the shallower the depth interval in object space is, 
the more absurd the base values which the A.-B.-C.-D.-E.-
Bercovitz-Spicer equations give. That's why several people 
recommend to stick to the 1/30 rule for everyday subjects. 

Abram Klooswyk