[photo-3d] Re: Re: Re: Quiz #3 - stereo window

[CanterMike@xxxxxxx]

In digest 224 Peter Davis writes, in part:

<< At 11:24 PM 8/31/2000, you wrote:
 >George writes, in part:
 >
 >  << To place a better stereo window you must crop the left side
 >  of the left picture and the right side of the right picture
 >  (in equal amounts).    >>
 >
 >
 >Did I miss something, or isn't that backwards?
 >
 >Since we accept that the Left sees Less on the Left, don't you need 
to trim
 >the extra from the right side of the left image, and the left side 
of the
 >right?  What does it accomplish to take image away from the left 
side of the
 >left image, since there is already less there to start with?
 
 No, this isn't backwards.  Imagine there's a tree to the left of the 
 photographer.  The right lens may miss it completely, but the left 
lens may 
 pick up part of the trunk on the left edge of the shot.  Therefore, 
you 
 would trim the left edge of the left shot.
 
 Put another way, imagine you hold a playing card up to your nose, 
sticking 
 straight out.  You left eye will see whatever side of the card is 
facing 
 left, and the right eye will see the side facing right. >>


A quick review of the LLL topic as discussed in February yields this 
from George (although written referencing slides, it applies equally 
to prints):

  "Actually (and with all due respect) you are confusing things by
   misapplying the rule.  This rule is for setting the stereo
   window when mounting stereo images.  If you look through a REAL
   window, your left eye will see less of the scene along the left
   edge of the window.  Similarly, when you mount stereo images,
   the left chip should show less of the scene in its left side. 
   The rule works!!!

   Now, how can we explain your observation that the left eye
   generally sees more *from the left side of things* than the
   right eye?  When looking through a real window, the left eye is
   forced to look from the RIGHT side of the LEFT edge (and the
   LEFT side of the RIGHT edge).  Hence, it sees LESS in the LEFT
   side and MORE in the RIGHT side.

   [snip]

   Honestly, this is one of the most practical rules I have learned
   in stereo photography.  I now instinctively look to see if I see
   more in the left side of the right chip.  I don't even think
   about it.  It is easy to learn the rule if you remember the
   three L's.  But after you use it for a while, it will become
   second nature, no matter how you apply it."


Peter continues: 
 << 
 >As I write this I hold in my hands an untrimmed set of prints fresh 
from the
 >lab (didn't think I could type & hold prints at the same time, 
didja?) and
 >when I match up the nearest (foreground) set of homologous points 
the right
 >edge of the left print clearly hangs out past the right edge of the 
right
 >image.
 
  Was this pair taken with a twin rig?  Might the lenses have been 
"toed in" 
 slightly?  Alternatively, is it possible the prints were slightly 
cropped 
 in enlarging?  Otherwise, I don't see how *parallel* lenses would 
allow the 
 right edge of the left image to extend past the right edge of the 
right. >>
 

 George took the images that started this thread with a single camera 
on a slide bar.  My images (as referenced above)were made with a 
Burdlo, a dedicated stereo camera made by David Burder by modifing a 
Nimslo.  I doubt very much that the Burdlo has any toe-in, but it 
could very well have no "built-in window" (making the "window at 
infinity", like you would get with one camera on a slide bar).  As
for 
cropping, that is a given for cheap machine prints, as well as an 
unpredictable factor-yet this effect happens with every pair (to 
different degees, depending on distance to nearest object [and 
subsequent window placement]).  Please note that I am aligning these 
prints in the manner suggested by Robert Thorpe at 
http://www.skep.com/3D/holmscrd.htm].  

If I may, I further quote Dave Kesner on LLL (originaly written about 
twin rigs, but it should apply to a camera with no built-in window):

 "Remember "LLL" the left image shows less on the left side.
  The problem is that [snip] there is no 
  "built-in" window or the window is set at infinity. That means 
  that the infinity points will be at the window and the entire 
  scene will be in front of it (coming through it). That is what 
  all that "excess" image is that you see in one film chip and 
  not the other. The only way to get rid of it is:

  1) to have the camera lenses and film gates offset to provide 
                   a "built-in window"
  2) to toe-in the cameras to achive the same thing (sort of)
  3) to mask the excess out ...[snip]  "


Masking out the excess is what we are discussing here.  The question 
is: Which side (left or right) needs trimming from which image (left 
or right)?

I can't explain the mechanics of why this happens.  Perhaps this has 
been covered...any of our math wizards want to draw this out, or does 
this belong to Tech 3D?

Sorry for such a long post. 


Mike Canter