[photo-3d] Re: Base Calculator

[Abram Klooswyk <abram.klooswyk@xxxxxx>]

I was just a few days away and now I find all this on base 
calculation in the mail ...

George Themelis wrote on 15 Sep 2000 13:03:28:
>The 1.2mm is what is considered the maximum allowable 
>stereoscopic deviation for 35mm film.  It is constant 
>and independent of focal length.

For 35 mm film??  Including 4-p (Nimslo) and 8-P (full 
frame)? and no mentioning of VIEWING?

But fortunately he added (15 Sep 2000 13:19:17,
16 minutes later, coffee?):

>OK, a bit more serious now...
>(...) The maximum shift is at infinity. This is p.
>(...) The focal length has nothing to do with this 
>conventional limit. But it does affect the magnitude of p, 
>all other things being equal.

After some interruptions, and on an additional subject, the 
next day, George Themelis 16 Sep 2000 16:56:06:
>Tom Deering wrote:
>> "Not a problem" in a hand viewer?  Is Mr. Themelis 
>> suggesting that people can diverge their eyes more easily 
>>with a viewer?  Or that it doesn't hurt as much?

>If the viewer has interocular adjustment then you can 
>increase the spacing of the lenses and that will tolerate 
>much more deviation.

But then (16 Sep 2000 18:02:31, one hour later, 4 cups of
coffee):
>After some further thought I do not see any reason why the 
>viewer is more forgiving than projection. Some viewers do not 
>have interocular adjustment.  With my Ekeren viewer (fixed 
>interocular) I cannot fuse stereo slides with infinity too 
>far apart.

But it is a long known fact that the viewer _is_ "more
forgiving". 
OK, not when infinity is too far apart, but definitely when
larger 
deviations are involved with _normal_ infinity separation, so
with 
narrower spacing of the near points on the slide, with means a 
smaller distance of near objects in viewing that slide.
The reasons for this may not be totally clear, but deviations 
three or four times the normal deviation (for that focal
length 
of viewer) are easy to view.

In my opinion, Olivier Cahen made a short but significant
remark: 
"The maximum deviation should be matched to the viewing 
conditions." (17 Sep 2000)
He described his home projection geometry:
>The seat for viewers is located two meters away from the 
screen. (...) The stereo window is located on the screen and 
the background is (...) 64 mm on the screen, so that it is 
viewed with parallel eye axes.<

When indeed the window is "On the Screen" at two meter, and
the 
near points are percepted at that same distance, viewing 
convergence will vary from zero (looking at infinity) to
almost 
two degrees (looking at the window / near objects). 

So this is the (more or less) accepted "easy viewing" 
condition for stereo projection: Looking through a "window" 
at two meter into a scene which extends to infinity. 
This is the Standard Stereo Viewing Space principle.

The tangent of two degrees is about 0.035, close to 1/30.
(Indeed, from there the one-in-thirty rule.) 
Thirty is the "standard" figure for any stereo _system_,
any format, it follows from the Viewing Space standard.
 
Calculating _backwards_ from this primary viewing condition, 
taking the viewer focal length in account, you arrive at the 
maximum near point deviation, or mask aperture deviation, of 
the slide. And viewer focal length is mostly matched to
format.

[Note: deviation is deviation _from_ infinity separation.
 Infinity deviation is zero. Maximal shift is at the near 
 point or window.]

So maximum allowable deviation on slide depends on viewer 
focal length, or on the comparable value: sitting distance 
from the screen (see Olivier Cahen).

Calculating _backwards_ again, from the maximum on slide 
deviation for your set of viewing parameters, you arrive at 
the camera stereo base only by putting in your camera focal 
length.

Now what about the standard "1.2 mm deviation for 35mm film"? 
30 x 1.2 mm = 36 mm 
The 1.2 is a standard only, repeat only, when your _viewer_
lens
focal length is about 35 to 40 mm (often up to 45 mm, no
strict 
value). 

This standard deviation does _not_ depend on film format, but 
on viewing. Stereo is for viewing. 
When your standard camera for 35 mm film would use a 4 
perforation format (half frame), with a 25 mm lens , you 
should try to get a viewer with the same focal length, and 
restrict your deviation to 25 / 30 = 0.85 mm.

But some people have argued that a longer camera focal length 
is better for stereo projection, say 60 mm for 35 mm film and
a 
5 to 8 perforation format. Suppose a viewer again with 60 mm 
focal length, then your standard deviation could be: 
60 / 30 = 2 mm.

I repeat: Maximum deviation depends on viewing. Discussing
maximum deviation without specifying the viewing parameters
is useless.

Abram Klooswyk