Re: Requirements for a new stereo camera

["The Photo-3D List" <photo-3d@xxxxxxxxxxxxxxx>]

Hi 3d-folks,

   Here is the first feedback from Olivier Cahen. As Olivier and I are only
linked by fax, you will not find here those traditional "At xx:xx xx/xx/xxxx
-xxxxx, X wrote:". I apologize if it makes this thread a bit difficult to
follow. In this paper, Olivier refers more particularly to these two posts :

> Date: Mon, 18 Mar 1996 14:20:10 -0600
> From: bercov@xxxxxxxxxxx (John Bercovitz)
> Subject: Re: Requirements for a new stereo camera

> Date: Mon, 18 Mar 1996 19:16:57 -0600
> From: wchang@xxxxxxxxxx (William Chang)
> Subject: Re: Requirement (8-perf)

More feedback coming soon...

Amicalement,
Yannick


+-----------------------/--------------------/------------------------------+
   Yannick Corroenne   /   yc@xxxxxxxxxxx   /   http://www.imaginet.fr/~yc
+---------------------/--------------------/--------------------------------+
         French Stereo-Club member since 1499 (or was it 1994 ?...)


>->->->->->->->->->->->->->->->->->->->->->->->->->->->->->->->->->->->->->->

>From Olivier CAHEN (Editor of the "Bulletin du Stereo-Club Francais")
16 rue des Gres
91190 GIF SUR YVETTE
France

phone and fax : (33 1) 69 07 67 21
Temporary E-mail : ocahen@xxxxxxx


24/03/1996

Dear friends,

   Thanks for your positive comments on my draft on the + ideal stereo
camera ; in your message dated 18 march, 14:20.

   I would much like to agree on John's directive approach. The point is
that the all the camera manufacturers, like all industrial managers and
everybody in the world, think that they can imagine something original and
that they also know something on stereoscopy as well as on any other topics.
So I first thought that it might be better to let them some choice. We are
not giving the specs for an order in which you pay the development expenses
like the DOD could do, but we try to help them deciding to start a new
development. If you actually feel that you could entice someone to start it,
the directive approach would be better. In any other case we must tell them
what is acceptable and what is not.

   About the tolerance on focal length, you may be right since it is not
easy to measure focal length at a low cost with a high precision. But that
we require them is only to sort pairs of lenses with the same focal length.
Since the standard manufacturing RMS lies around 1%, I think that nearly
every lens can found a twin brother without a too high extra cost. The
important point is that it is not each contribution to the vertical
deviation, but the overall result, that must be kept within 0.1 mm: one mil
for focus deviation, one mil for film guiding, one mil for film width
tolerance, and so on: there are many causes for this vertical deviation, so
that we must not allow half of the total for a single type of error.

   A device measuring the distances of the foreground and the background
would be very pleasant, but as far as I know it can only measure the
distance of the foreground, at least at a reasonable cost. The objective of
minimizing the cost should be maintained.

   Of course, the priority must be to the aperture as the primary
autoexposure mode.

   The current rule, since the time of Sir David Brewster, is a limit of two
meters for the window (20 toe-in), and even some authors recommend a window
at three meters. My own RBT camera has a built-in window at three meters. If
you accept it at five ft only, it is due to your Realist camera with very
short focus. My personal preference is between 2 and 2.5 meters, but this
point is open for discussion within the ISU. But do not forget that some
people, like... yourself, John, in your message sent one hour later,
recommend a longer focus, i.e. 50 mm.

   O.K. if you find how to measure the background distance, but I do not
like a two-step procedure for image shooting. Do not forget that a simple
operation is also an important objective to be met.

   You proposed a film gate interval of 69.25 mm. This is not an integer
number of perforations, so that stepping the film at each shot by this
amount can be a problem of accuracy at the lowest possible cost. I prefer
your later proposal for 15 perf as the film gate interval.

   I started from the idea that a step by an integer number of perforations
would be cheaper and more accurate. This point must not be specified, but
discussed with the camera manufacturer.

   I found your argument, William, to be sensitive: eight perf has the
advantage of standard process. I also found your proposal for six perf clips
in a 8-perf frame, to be sensitive. If this option is finally kept, I can
give you another suggestion: a seven perf clip in a eight perf frame. The
stereo base would be about 69.5 mm with a 1.75 shift for the window, the
film gate width about 33 mm, the film gate interval exactly 15 perf i.e.
71.25 mm. You would lose about 4 mm per slide, much less than in the RBT S1
or with your 6 perf clip.

   This solution would also avoid the tricky problem of designing and
manufacturing the film loop with a good accuracy and at a reasonable cost.
It is compatible, like yours, with standard processing, and with automatic
mounting in pin-registerred slide mounts, since the film gate interval is an
integer of perfs.

   You would much appreciate a WYSIWYG camera (I suppose that you mean with
the window in the viewfinder); my last proposal is not compatible with this
feature with reflex viewfinders, but it would be with a pair of cheaper
Galileo-type viewfinders, since their interval can be different from the
lens interaxial. On the contrary, your six-perf format would be compatible
with WYSIWYG reflex.

   If the autofocus device gives the operator a warning in the case of a
foreground closer than the window, the controversy of mounting to the
infinity or to the window will fall down: the window will always be correct
with an automatic mounting at the infinity in pin-registered slide mounts,
so that the infinity points will no more jig back and forth at every new
slide pair.

With my best regards,
Olivier

<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<-<


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