P3D Richmond, and Base

[Michael Georgoff <Michael_Georgoff@xxxxxxxxxxxxxx>]

Bob Wier said...

>....everyone heading to NSA shortly. If you feel
> that you can't handle the mail volume from the *3D mailing lists
> and I don't see why not since everybody will be in Richmond, and
> nobody posting :-)

Everybody will be in Richmond?  Alas, and woe is me, I gotta "stay
home"...  We who are not going to Richmond will be depressed that we are
not, and when the reports start coming in from those who did go, we will
be depressed all over again. Now, while the Big Boys are in Richmond,
let's not let it get TOO quiet around here!  (Bruce, ya with me out
there?)  

If the Big Boys and Tribal Elders are all in Richmond, hell, I'll have a
go.  And, I'll start with that age-old P3D favorite, Stereo Base. (Okay,
maybe Realist v Kodak gets more attention over the span of time, but
stereo base is always a good one to wrestle with every once in a while.
I was looking forward to watching Heavyweight 3D Pro Battle of the
Stereo Bases duke it out again in response to Andrea's 1:30 question.


Andrea said:

>Can anyone tell me if I should be measuring from the front of the lens 
>or from the film plane (camera) when calculating the stereo base. I am 
>shooting macro with a single SLR and sometimes have 6 inches of lens 
>hanging off the camera but am only 2 inches away from the subject.
>
>i.e. For a stereo base of 1:30, you move the camera 1/30 the distance 
>from the XXX to the subject. (XXX being either the front of the lens or

>the film plane)
>
> Don't get too technical, folks. I can make the images; I'm just trying

> to get accurate information for my presentation this weekend at NSA.
The 
> books I have don't seem to detail this. Of, course I kinda go glassy 
> eyed after too many graphs and charts and scientific formulas, so I
may 
> have missed it!

Andrea's question was about "measure from the front of the lens, or the
film plane", and Dr. T answered "The film plane."  But, I saw the 1:30
and thought I'd comment about the issue of base.

I explored the issue of stereo base a while back, via extensive off-list
email exchanges (and much thanks to all the kind and patient folks
who've contributed to my education), and much poking around for
information.  What I've found is that when it comes to stereo base,
there is no shortage of perspectives on the issue.  As a beginner (not
to suggest I've passed that stage), I figured "Gotta be science behind
all this, gotta be some logic to all this. People been doing stereo for
a hundred years or more.  Science guys, smart guys, educated guys...
studying, writing books full of calculations and theories.  No need to
reinvent this wheel, man has a history of documenting science, gotta be
some standard answers out there.  And, lots of stereo experts out there
today, surely they've come to some reasonably agreed-upon conclusions."
Uh, nope.  I am still baffleghasted by the lack of a simple answer.
Like a new religion, I seek the ultimate Stereo Base Truth God, but
there are many to choose from.

I summarize my findings by suggesting there are two ways to approach it:
1) Don't worry, shoot happy.
2) Use math.  How much math do you need? Well, how much math do you
want?  We got all kinds of math.  What we've got ranges from simple
rules of thumb though the school of thought that starts with
reconstruction (human vision convergence limits, addressing the issue of
on-film-deviation) and working backwards to construction, specifically
stereo base, leaving no room for assumptions.  From easy rule to "Big
Science".  You choose.  As a generalization, small formulas leave out
stuff that may be important.  1:30 does not ask for focal length of
lens, f-stop, an on-film-deviation figure, and so on.  So the math
routes range from simple to complex, with assumption ranges from "many
assumptions made" to "no stone unturned in the quest for a standard
geometry".


1) The "Don't worry, shoot happy" Variants:

1a)  "I don't pay no never mind, I have a [fixed base Realist, Kodak,
etc, or use a multilens Lenticular camera for non-lenticular but paired
results] and I just shoots them pictures."  No real problem here.  In
fact, a fixed base is pretty safe.  Shoot a far away landscape and you
get zilch stereo depth, and you really can't get "too close" as the
camera will generally only focus as close as x, wherein base for a
stereo pair at x focus yields almost too much depth (quantified as
exceeding generally accepted on-film deviation, and certainly dependent
on far-point objects in frame, you gotta watch that in
framing/composition before you shoot).  Only potential for error here
(if on-film-deviation is within range) is that you've got to mask
properly in mounting, but that's no real obstacle.  This is typically 
"self-correcting" in that a beginner (like me) will not take too many
distant-subject "not much stereo here" shots, and not take too many "too
close, my brain hurts" shots either. (Too much OFD = brain hurt, or
wrong mask used in mounting)

1b) "Don't worry, shoot happy" sometimes also applies to those who have
fixed a twin rig at the "least physically possible" base (typically side
by side, as close as you can get 'em).  Applies to SLRs, point'n'shoots,
disposables held together with rubber bands or tape, etc.  Just get 'em
close as ya can, figure out a synch (electrical, mechanical,
finger-and/or-thumb-ical, etc), and shoot.


2) The Math Variants, from rules of thumb to mind-numbing multiple
variable calculations.

2a) The "1 in x" rule is the general rule of thumb for typically single
camera shoot/shift, slide bar, Cha-Cha, etc.  Also applies to twin-rig
with variable spacing...  

2a1) 1 in 30 is the only rule you need, some say. Stereo base = distance
to object/30.   (Which would set the Realist fixed base of approximately
70mm at optimal results for objects/subjects some 6 feet 10 inches, by
the way).

2a2) No, it's the 1-in-focal length rule.  Stereo base = distance to
object/focal length of lens (eg, 1/30 is good for a 30 mm lens, 1/50 for
a 50 mm lens, etc). (Which would put the Realist, with 35.2 mm lenses,
at some 8 feet, 1 inch).

2b) Stereo Base = Near Subject Distance / .80 x Lens Focal Length

2c) Stereo base = (distance-focal length)/(focal length/deviation)

2d) Stereo Base = ((Near Distance X Far Distance X On-Film
Deviation)/(Focal Length X (Far Distance - Near Distance)) -1 
 
2e) If infinity is in the picture: Base = Deviation * ((near point/focal
length) - 1/2)

2f) Or, if infinity, an approximation is: Base = Deviation (Near
Point/Focal Length)

2g) B = d * ( f * ( an + af ) - 2 * an * af) / ( 2 * f * ( an - af ) )

Where B is the baseline, d is the maximum desired deviation in the
negative, f is the focal length, and an and af are the distances to the
near & far objects, respectively.

2h) b0 = d*((an*af)/(af-an))*((1/f)-(1/a))
             
Where:
b0: This is the maximum tolerable separation of the camera's lenses.
d = maximum allowable on-film deviation.  
an: distance from camera lens to nearest object in scene.
af: distance from camera lens to farthest object in scene.
f = the focal length of the camera's lenses.
a = the distance at which the camera is focused.


(Note: If I had been more careful gathering up these formulas, I could
give proper attribution.  I apologies for citing these without giving
credit where credit is due.  I'm sure various of the long standing
members of P3D will recognize some of these formulas.  And, I apologize
if I screwed up any cut'n'paste in the formulas, getting them confused.
So, don't just take these and crank numbers through them, in case I've
made errors.  The intent here was to present evidence of an overwhelming
number of formulas considered by  P3D folks over the past few years.)

2i) Want to calculate base via web browser?  No messy math!  Just enter
your variables:

Bob Mannle's web-based base calculator is at:
http://www.tisco.com/3d-web/papers/calc.htm

Joel Alpers web-based base calculator is at:
http://www.frii.com/~rkymtmem/rkymtmem-3d/stereo.base/base.html

Note that these URLs either use formulas cited above, or explain what
they do use on their respective pages.


I do not profess to understand each of the above formulas in great
detail, so please don't email me with questions and equations, I'm the
wrong guy.  The right guys are probably all in Richmond.

For sheer fun, hit the P3D archives at
http://calcite.rocky.edu/photo-3d/ghindex.html and search for "stereo
base" (483 matches in 259 separate files).  Better yet, try searching
for stereo base and stereo base and formula or equation.   Lots of good
reading there by the masters.

Discussion of stereo base will often make references to published works,
some reference Fritz Wacko, but  the oft-cited and apparently revered
work of Jac Ferwerda surely seems to be the master reference on the
topic.  For an excellent piece of work, a must read is "A Comparison of
Camera Base Calculation Methods Originally published in "Stereoscopy",
the Journal of the ISU. Co-authored by J. Bercovitz and Kiewa Valley
Stereo." Find it at http://www.werple.net.au/~kiewavly/bases.html.
Highly recommended.  Beginners beware, may take multiple reads to get
though it.  Calculator or computer spreadsheet optional, but very, very
helpful.

I am still surprised as a beginner that here are "too many answers" to
what I thought was a simple question, "How do you calculate stereo
base?"  I was, and still am, daunted.  I'll get though the math, the
light bulb will go on, I will sit up in bed and go "Aha!" some night.
In the mean time, I can "don't worry, shoot happy". (But, I may
personally go with Bercovitz/Spicer).

Thanks again to all who've helped me along the way.  As always,
criticism, corrections, advice, rants, raves, clever insults, etc, all
welcome, on-list or off.

Michael Georgoff
San Jose, CA



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

End of PHOTO-3D Digest 2881
***************************