[photo-3d] (Con)fusion limit

[Abram Klooswyk <abram.klooswyk@xxxxxx>]

(Was: again those figures!!!, but it's not on figures alone)

Is this on a practical stereophotography subject or just dull 
theory? It is practical, because some people want you to 
believe that the standard 7 feet window of normal outdoor 
stereophotography is too close, it should be 9 or 10 feet, and 
you should avoid closer objects than that, if distant objects 
are seen also in your stereo picture.

---------
One of my favorite myths in stereoscopy is the allegation 
that close coupling between convergence and accommodation is 
a cause of eyestrain in stereoviewing. 

In the late 19th century L. Cazes in France believed that the 
link between them should never be loosened, and that therefore 
the limit of fusion was set by the _depth of focus_ of the 
eye. This sets a limit of 70 minutes of arch (1.17 degree) to 
the difference in viewing convergence.

>From that resulted the rule that (with a normal 6.5 cm stereo 
base and a scene with distant objects) nothing should be 
closer than 3.2 meter (10.5 feet). For hypers and macro the 
rule was: base 1/50 th of nearpoint distance.

Sometimes is said: the difference between 10 feet and 7 feet 
is only 3 feet, so why bother? But the depth quality or depth 
range in stereopictures is measured in angles. 
A depth range of 2 degrees is more than 1.5 times 70 min, so 
setting a limit at 10 feet in stead of 7 feet is wasting more 
than one third of the possible depth effect. This is so 
because at closer range stereovision gets increasingly better.

What about the link between convergence and accommodation? 
It is easily loosened. This had already been shown in the 19th 
century, it was known to von Helmholtz. 

Accommodation is designed to provide clear vision, and 
vergence to provided binocular single vision. When they work 
together, as in everyday vision, all is well. When they come 
in potential conflict, as in stereoviewing, they primarily 
stick to what they are designed for, as they should, and they 
go wide apart, at least wide enough to permit sufficient depth 
range in stereoscopy.
Apart from science, we know that loosening the link is easy 
because freeviewing would not be possible without it. 

Another convincing case is myopia (short-sightedness). You 
might be myopic yourself, or at least you know several people 
wearing glasses with negative lenses. Let's say someone has a 
moderate myopia of exactly 3 Diopter. With his glasses he 
sees distant objects perfectly sharp with zero accommodation. 
However, when he takes off his glasses, zero accommodation 
will result in distinct vision at 33 cm, one foot. In the 
first zero accommodation case convergence is zero too, but in 
the second case it is more than 10 (TEN) degrees. (Diopter is 
a measure of lens power, Diopter power is 100 cm divided by 
lens focal length in cm).

Would you believe that wearing the glasses gives myopes 
eyestrain, or that taking the glasses off would cause harm?  
Myopes will laugh at you, they do this all the time.

Of course, there _is_ a link between convergence and 
accommodation, but it is so easily loosened that it is of 
little practical significance for stereoviewing. Mostly the 
depth of focus of stereocamera lenses sets a limit to the 
depth range, long before the convergence-accommodation link 
comes into play.

Owen Pearn wrote (Jul 19, 2000):
>(...) disassociation of the accommodation and convergence 
>ocular reflex is a cause of stereoscopic viewing discomfort.  
>there is an abundance of human factors literature which 
>quantifies this. numbers vary, and people vary, but a maximum 
>of "1.5 degrees" is common.  

This 1.5 degrees of convergence range means that pictures made 
with a StereoRealist (70 mm base, 35 mm focal length) should 
have no closer object in the scene than 2.7 meter, almost 
9 feet, if distant objects are included.

The StereoRealist and its 5-P family members were designed 
with a mounting system providing a stereo window at 7 feet,
and this is all laid down in US and international Standards.
Far more stereopictures have been made with 5-P cameras 
than with all other systems together. 

So the 2 degrees stereo range is validated by millions of
stereopictures, but several stereo experts have shown that
at least incidental use of a 4 degree range is also possible,
so called "double depth".

So what to do with the literature which pretends that the 5-P 
window distance is wrong? Forget about it, its is all false, 
in error, based on myths, and not fused but confused :-).

Abram Klooswyk