Re: Answers to Gary's accumulated questions

[P3D John Ohrt <johrt@xxxxxxx>]

P3D Dr. George A. Themelis wrote:


> >On the other hand, there are the discussions here that lead me to
> >believe that the lenses must match exactly, or 3D isn't possible.

Gary, I think you misread the discussions, they are centered on how to
measure.

 
> But, you did not ask DrT!  All this "nonsense" about matching to
> 0.001% (is that the definition of _exactly_?) can be safely discarded
> as... well, nonsense!

And just how do you know that?  You've never tested it.  And by the way,
there are professional viewers who look at this stuff all day and you
can resassured their tolerances are a whole lot tighter than a casual
viewer.  I'd say you are dead wrong on the assumption that people don't
make camera's that accurate for the heck of it and then sell them to
cost sensitive commercial firms which use them the manipulate aerophoto
data for stereo viewing, amongst other things.  BTW, 0.001% is not that
difficult to achieve but it does require periodic calibration.

>  Sam is a practical man and knows that
> "perfection" is a very time-consuming and expensive affair.  Why be
> "perfect" when "good" is good enough??? :-)

The greatest misconception of technology is the assumption that better
costs more.
 
> Define "_not_ matched".  Usually tolerances of reputable SLR makers
> are tight enough to guarantee an "OK" match.

ROTFL

Test a few.  You will find that a tight tolerance of FL is not a
signficant concern and the "distance to subject" marks are rarely
accurate.  You can include Leicas in the test group :-), my optical
engineers did in one of their Friday afternoon quests for truth!


> I'll let now the professional stereo photographers who populate this
> list explain the faults of my thinking and will go and have another
> beer and shoot a few more pictures! :-)

Not a bad attitude, but I usually perform better before the booze:-)

Up to now, no one has put up a figure to define good enough.  So from a
technical perspective, no one knows what they are talking about :-)



Source MIL-HDBK-141

Military Standardization Handbook fo Optical Design

approved 5 October 1962

Genreally considered one of the best references for a layman.

Quoting para 4.7.5 - Design Considerations

.... The co-ordinated motion of the two eyes must not be disturbed.  A
pupillary adjustment of 50 to 76 millimeters should be provided. 
Magnification differences to the two eyes should not exceed 2%.  Some
people cannot tolerate more that 0.5% while others may tolerate a little
more than 2%.  Occulars must be paired so that increased size
differences will not occur....

The amount of light to the two eyes should be balanced, preferrably
within ten per cent.....

WAIT, don't run out and design anything yet!

At that time the military design was limited to young males within a
specific height/weight range who had been accepted based in part on
vision.

Secondly, you just read a system spec. and that means the design must
budget errors throughout the system.

A sample.


Suppose you select that the system maginfication error must be less than
0.5%, to ensure all can appreciate the wonders of 3d.  Then in a
camera/projector system you might assign 0.25% each to guarantee
compatibility and assume processing doesn't add significant error in
this area.  But if you are design on a camera/viewer function, where
there will be many viewers per camera built, you may wish to assign 0.4%
to the viewer and 0.1% to the camera.

I hope you are getting the idea.

Finally, these are very loose specs for lightly used equipment such as
aiming devices.  I can personnaly assure you that the specs for
binoculars are much higher.  As a lookout and then later officer of the
watch, I have served my time with binoculars with no problem.  Some of
the old fashioned gunsights were no so well fashioned.

All this suggests to me is that a "good enough" match is much better
than 0.5% scale and that some of our viewing techniques and equipment
may not be optimal.

As Sam pointed out, you can scale correct as part of the processing. 
Scale correcting slides is doable, but the slide copiers that could do
it are not cheap by my standards.  If you digitize for computer display,
scale correcting is easily accomplished.

The "light" spec, well that's another area of investigation.

Anybody got another spec?

Finally, this is a hobby :-).

That means you get to do what you want and not answer to anyone :-)

Regards,
John
--
John Ohrt,  Regina, SK, Canada
johrt@xxxxxxx


------------------------------