SL3D some more

[P3D <PK6811S@xxxxxxxxxxxxxx>]

Allan Woods:

>This is getting exciting!  Oh boy!  I hope no offence is generated
>by the good natured and passionate approach  ;-)

Much more interesting than over-polite instructions on screw removal ;-)
Reading this list is sometimes like watching a Chip-n-Dale cartoon,
"Thank you.  No, thank YOU".  Which tone is a much-appreciated
change from most lists.

>But, once the image is recorded on a single sheet of ordinary
>photographic film, we cannot reconstruct the original light
>paths just by looking through rose coloured glasses.

Agreed - rose-colored glasses don't work :-)  But the light paths
as well as the two images can be reconstructed:
   Identify a point on the film.  x-y location and color.
   If it has a blue component, trace that back through the blue stop.
   If it has a red component, trace that back through the red stop.

>I don't think the analogy applies because we have only one piece
>of film.

Occupying the same film/space is not a problem, in fact overlaying a
stereo pair on a common surface is ...common:
  - anaglyph, images separated by color
  - interlacing, images separated by synched glasses
  - projection, images separated by polarization

If the usual stereo slide pair is projected using red-blue filters instead
of polarized they will be overlaid on screen and reconstructable by 
viewing with red-blue glasses.  Take a 2D pic of that screen and
you will have what SL3D creates.  Well, at least like what my lenses
create, not so sure about WC's :-)
                                  
Peter posted an interesting extract from the Q-DOS manual.
For the optimal effect, use a medium tele length and open up the
aperture.  Based on tests done with a 135mm 2.8 lens,
this will result in a slab of space that is mostly in focus
and exhibits depth, with a whole lot of unidentifiable blur that seems
to be located further away.  The narrow angle of view helps since
it encourages one to fill the frame with a single object which can
be seen in detail with depth.  Isolating it against a dark background
also helps.

A limitation of the Q-DOS implementation is that the two stops (red & blue)
are constructed to occupy opposing semidisks of the standard diaphragm.
Opening up improves parallax, but sacrifices depth-of-field, you
see more 3D-depth but more of it is out-of-focus :-(  Using separated
stops of smaller diameter would bring back the detail and retain the depth.

Jimmy Songer (thanks WC!):

>...image, i.e., the stereo information is carried in the blur normally 
>associated with out-of-focus region of operation of the lens.

It's only blurry because the stops are so large.  Stopping down both 
sides while maintaining separation will cut the blur so that two
images can be resolved.  That stereo information is carried in the blur
I discussed in a months-old post using a multiple-pinhole model.

>a photograph which resembles the same view of object space as would be 
>seen by a person having two eyes focused on the focal point of the same 
>scene and is therefore free of distinct double images

Double images are _smeared_ into indistinctness because of the large stops.
Two separate images are overlaid, but some portions are not resolvable
because of shallow depth-of-field.

(Since this was one of the goals stated in the original patent, maybe
it would have been possible to patent a lens with small stops, stating
the _opposite_ goal)

Paul Kline
pk6811s@xxxxxxxxxxxxxx


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