P3D 1/30 rule, close-ups, aesthetics

[abram klooswyk <abram.klooswyk@xxxxxx>]

In the reactions on Jeff Toman's posting (Photo-3D Digest 3556,
20 Oct 1999) by Boris Starosta, George Themelis and Rod Sage two
different issues should be distinguished in my opinion:
1. the ratio of the 1/30 ratio.
2. the stretch which can occur in viewing close-up stereo's
   which have been shot with a too large base.

The more these topics are discussed, the more I get the feeling
that there is basically consensus on the technical issues,
but that people differ on the aesthetics. So permanent
discussion is guaranteed, and Photo-3D will live for ever.

Boris (Photo-3D Digest 3557,  21 Oct 1999):
>(...)
>the 1/30 rule limits "on film deviation" and is intended to insure
>that there is not too much parallax in a view, especially when
projected.
>The 1/30 rule is also a simplification, as it assumes that you are
>including infinity in the view.  So the rule is to limit the maximum on

>film deviation of two homologous points at infinity.
This addresses the technique.
>I have seen more shots ruined by the 1/30 rule than
>I care to remember.
This is on aesthetics.

What Boris intended (I believe :-)) is that the 1-in-thirty
rule must not be used mechanically and that it helps to
understand why it was made.

Cameras of the 5P family (StereoRealist group) have a built-in
stereo window at 7 feet. This is about thirty times the lens
separation  of those cameras. When shooting with the closest
object at 7 feet, infinity or distant objects may be included,
and "distant" mounts can be used.
It is more or less accepted that a depth range from 7 feet to
infinity mostly guarantees comfortable viewing. (Still having
a person at 7 feet and nothing beyond except distant mountains
is not recommended, more stuff at different distances
seems better).

Now what about the depth range in close-ups?
I believe it helps a lot when you imagine what the eyes do in
*viewing* stereopictures. Assuming "othostereo" conditions,
in looking at distant objects of the illusionary 3D scene the
eye axes are parallel, making an angle of zero degrees.
Looking at an object at the official window distance of
seven feet, the eyes converge.

Strictly, an angle of one degree is subtended by an object of
1 cm at 57.3 cm (or 1 inch at 57.3 inch), but generally the
ratio 1/60 for one degree is a close enough, and 1/30 for two
degrees. Converging at 7 feet (2.1 meter) with an eye
separation of 6.5 cm gives a ratio of 6.5 / 210, which is
not far from 1/30 (The StereoRealist comes even closer).

So in changing from looking at infinity to looking at 7 feet
the eyes make (together) a *swing* of 2 degrees.
This is a standard for 5P. The 5P mounting system assumes
a maximum swing of 2 degrees in viewing, and from there
come restraints on the depth range in the photography.

Now what to do when the closest object is at ten times the
stereo base? In looking from a point at ten times your eye
separation, to a farther point, swinging the eyes 2 degrees
at most, how far is that far point?
Just compute 1/10 - 1/30 = 3/30 - 1/30, so 2/30 or 1/15.
In accepting the 5P rule, no object should be farther
away then 15 times the stereobase, when the closest is at
10 times the base.
All far point maximums can be computed from the nearpoint
distances in this way. (Seems simple, but some people need
spreadsheets to compute this :-))

Boris:
>(...)
>If you can limit depth in your scene (crude example: shooting
>a head & shoulders portrait in front of a wall), you can get
>away with much more stereobase - or you can get a lot closer
>than the rule indicates.
George Themelis (Photo-3D Digest 3557 21 Oct 1999):
>While I agree with *everything else* that Boris wrote, I tend to
>think that this example is rather unfortunate (>quote)
>The portrait is a very distortion-sensitive subject and needs to
>be treated with a lot of respect.  Better safe than sorry. Less
>is almost always better.  (...)

This obviously is on aesthetics. George doesn't say that the
mentioned portrait is not viewable, but that it can show an
unpleasant distortion.
Rod Sages mentions that the Macro Realist uses about 15 mm
base for 10 cm distance.
But the distortions in viewing Macro Realist slides are often
very obvious, especially with recognizable familiar objects.
Flowers of which you know they are perfectly circular
can become oval, streched in depth.
The cause of the stretch is a significant difference in
convergence when looking at a real object (or person)
compared with the one in the 3D illusion.
Is the distortion always unpleasant? No, but often it is
in portraits.

This is what Ferwerda in his book called "close-up misery".
(The book "the World of 3D", is still available at the
publisher http://www.STEREOscopy.com/3D-books/ and some
resellers).
In the book however, Ferwerda also pointed out how to use
"double depth", a technique which essentially doubles the
swing in viewing to 4 degrees.

Abram Klooswyk