P3D Re: IMAX 3D projection

[boris@xxxxxxxxxxxx (Boris Starosta)]

I hope no one minds this voluminous "public" discourse between two enthusiasts.

Paul:

I think I am probably even less versed in the technical side than you are.
My understanding of things has always been more intuitive, and this fools
me into thinking I know it all, and I get sloppy with the language.  As to
the question of orthoscopy vs. orthostereoscopy:  what I mean is the
perception of "being there."  If that is achieved through orthostereoscopy,
then that is the term I shall endeavor to use henceforth.

I do realize that a "true" orthostereoscopic viewing experience is only
possible from one seat in the theater - however, I think that something
close to orthostereoscopic viewing is possible from a broader range of
positions (when I viewed "Into the Deep," I was sitting rather close to the
screen and a bit off center, but I still had what I thought was a fairly
orthostereoscopic view).  When I say that I think Imax shot the film
orthostereoscopically, I mean that they used a lens spacing close to that
of the human eyes (now confirmed by an earlier posting), and a lens focal
length whose field of view will somewhat match the perceived field of view
in the theater for most visitors in the middle seats.  Is there another
term that describes such a near-orthostereoscopic technique?

>From: Paul Talbot <ptww@xxxxxxxxx>
...
>So you suppose these objects were even a lot closer than 4-5 ft
>from the camera?  Is that "where they went wrong" in that film,
>in your estimation?  How would you want them to keep from presenting
>that close to your eyes?  (Which, I agree, is too close.)

I don't think that they made a mistake shooting such objects, or shooting
that close to the camera.  It is the presentation that was perhaps too
challenging.  To mitigate the difficulties, I would have at least edited to
minimize "jumps" between extreme close-ups and more regular close-ups (but
the Imax people were surely trying to minimize such jumps already).

However to accomodate a greater number of people, I think a more pro-active
solution would be better.  That would be to move the left and right images
closer together on the screen - so people don't have to cross their eyes as
much.  But the resulting image will then be quite different
(non-orthostereoscopic): instead of a tiny crab "in your lap," the observer
perceives a giant crab at about the distance of the screen (or whereever
you want to set your close distance limit for such shots).

This solution can be effected at the time of taking the picture (toeing in
the cameras, or shifting lenses in, or shifting the film gate and film
out).  My guess is that this would seriously complicate the design of the
camera, as well as the camera operator's tasks.  (Having learned from a
recent response in this thread that a set of mirrors is involved in getting
the image on film, perhaps a tilt or shift of the mirrors would be a
possible solution for extreme closeups).

Or the shifting can be done in post-processing, resulting in the loss of
some of the image (just like mounting your realist slides in close-up
masks).  The problem there is doing this step for only a few shots - do you
do it over and over again for each print that you manufacture, or do you do
it once on an internegative - a "duping negative" - resulting in the loss
of a generation?  (Sorry about all of the cinematography "lingo" - I fear
my usage).

Or, lastly, you could shift close-ups during projection.  That would be
most difficult - requiring information on-film about projector aiming
requirements. Plus the projectors would have to shift rapidly, and/or cuts
prior to projector shift would have to involve short fades of the image, to
accomodate the time needed for the shift.

It goes without saying that for such shots, and such shifting of the left
and right image, depth within the image must be tightly controlled at the
time of shooting.  Because the end result of such shifting is to place an
extremely close image further from the observer, this also means that more
distant objects (esp. infinity) get pushed even further back.  Depending on
the amount of shifting, items only a few feet into the scene at the camera
might project on screen at beyond infinity (i.e. forcing the observer to
diverge his eyes to focus on homologous points).  This is to be avoided,
obviously.

Another way to look at this (please anyone correct me here, if I'm wrong.
I'm doing a lot of guessing in this thread.), is that when the left and
right images are shifted together for projection, the perceived scale in x,
y, and z is enlarged.  So care must be taken that z is not enlarged beyond
infinity.

Now, for another subject.

Paul Talbot wrote, in reply to my assertion that orthostereoscopic
projection onto a large distant screen such as is the case at Imax, without
window violations, would result in a rather flat stereo image:

>I think you might mean to say "ortho-stereoscopic."  According to
...
>So, unfortunately, I still do not understand the assumptions you
>use to reach your conclusion.

Given my additional comments above, and my understanding / definition of
orthostereoscopy, unless you have stereo images in front of the screen,
assumed to be 50 feet away from the observer, you will have a very flat
image.  Many more people would get no depth perception from such an image
than would be the case if images are kept much closer to the observer.
(This assertion rests on my experience, related via posts last year and my
website, that stereo perception appears to run through a spectrum.  That
is, some people have only moderate ability to perceive stereo.  Like with
all things, there is probably a "bell curve" distribution of stereo imaging
capacity in people).

So by shooting close to the camera, Imax is just giving a higher percentage
of people a truly stereoscopic thrill (at the expense of tiring the rest of
us at the top of the bell curve).

>> looking as if through a window six stories tall, 80 ft. wide
>
>Is this the actual screen dimensions, or numbers derived from
>some input variables?

These are actual dimensions according to an announcement at the theater.

...
>If five inches disparity is acceptable with infinity at the screen,
>why impose the 2.5 inch limit when placing infinity somewhere behind
>the screen?

I am saying perceived infinity is just as likely to fall behind the screen,
even behind actual infinity, as it is to fall onto or even in front of the
screen.  So for Imax to shoot for infinity at the screen seems very
reasonable given the technical limits / tolerances of the system.  Cripes,
they couldn't even keep vertical alignment to within a foot!  And that's
the easier variable in a horizontally running film, I would have thought.

>My concern is not with the window, but with the problem of watching
>"ghost-like" images that dance in thin air and pass through other
>patrons in the theatre.

I acknowledge this is a problem.  I can't think of any way to fix this,
without removing the offending audience members who might be sitting in
front of you, or without keeping the image very far away.  It's a situation
where tradeoffs are going to be inevitable.

>Thanks for the patience.
>

I'm enjoying it.

Sincerely,




Boris Starosta

usa 804 979 3930

boris@xxxxxxxxxxxx
http://www.starosta.com
http://www.starosta.com/3dshowcase



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