Re: Depth reconstruction in freeviewed images

[P3D <PTWW@xxxxxxx>]

Larry Berlin writes:

>******  At least with freeviewed images, the ortho scene typically has an
>infinity that is extremely fixed rather close to the prime subject matter.
>Everything in the image tends to flatten closer and closer to this
>infinity point often placing rather near subjects like a hill or house
>at almost the same distance as infinity itself. However real the closest
>subjects may be, it just isn't natural for infinity to be *just behind
>the apple tree*.  Reality has infinity stretching off to forever and it
>doesn't look that way at all unless you are viewing the image in exactly
>the right way.

Now this really got my curiosity up.  I pulled out a Realist slide from
my recent trip to the Grand Canyon, :) popped it in my $3 viewer, and
saw nothing even remotely akin to what Larry describes.  There was lots
of separation between the near and far, and infinity did seem very much
to stretch on forever.  So I took the slide out and went to freeviewing.
What a difference!  Most of the depth is gone.  The image had very nearly
flattened out to 2D.  Now this raises some very interesting questions!

DrT once posed a question to the effect of why do people who know how to
freeview prefer freeviewing to other viewing methods.  (I didn't know how
to freeview at the time and could not relate.)  Given this loss of depth,
such a proposition becomes most mystifying indeed! (Note that the question
George posted probably dealt with prints rather than slides, however.)
DrT has also been known to voice a dissatisfaction with stereo images
presented on the computer, for a variety of reasons.  But could it
actually be that one of the most important reason for this dissatisfaction
is the inability of freeviewing to recreate the depth in a manner even
remotely resembling what is possible with the lowly $3 viewer?  And does
this now elevate the $3 viewer to status of "viewing the image in exactly
the right way," as Larry says is necessary for good depth reconstruction?
;) ;) ;) ;) ;) ;) ;) ;) ;) ;) ;) ;) ;) ;) <--- all for DrT!

I'm glad Larry included that little "at least with freeviewed images" as
a preface to the paragraph quoted above.  It certainly led to an eye-
opening discovery for me!  If I am correct in deducing from past comments
that Larry works mostly with freeviewed images, while DrT and myself
normally view stereo slides in a viewer or projection, this choice of
viewing method could go a long way toward explaining why some of the
recent opinions about hyperstereo have been so sharply different.

Has anyone explored this depth reconstruction issue with respect to
freeviewing vs viewer-viewing of stereo slides?  Gabriel just mentioned
to me off-list something about freeviewing being like viewing with a
particulary extreme focal length, so I suppose the topic has seen some
discussion in the past.

Paul Talbot, looking for the code to send to the auto-responder to get
             the authoritative lowdown on this topic!


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