This is such a hairy topic, distortion on curved screens, for a number of
reasons. First, every screen/projector set-up produces *some* distortion, even
on a flat screen. Deeply-curved screens are another creature altogether.
Distortion (aside from the uniquely-3-strip forms like join lines and bent
horizons) was less obvious in their original use for 3-panel Cinerama, since
each projector was shooting onto a screen which was only about a 50-degree
curve, which is quite shallow, and which reduced distortions such as keystoning
due to off-center projection angles. Problems multiply when you try to shoot a
wide, rectangular image from a single position onto that same deeply curved
screen. To the projector's "eye" the screen is bow-tie shaped, not
rectangular, since the center of the screen is so much further away than the
edges. (Forgive me; I know this is old hat to most of you!) The center looks
much shorter in height. This brings up the age-old debate of: Do you crop the
image to the shape of a bowtie in the aperture, losing picture content but
keeping screen height consistent, or do you project the entire image and mask
the screen on the sides, reducing the "wow"-ness of a curved-screen effect.
Neither is perfect. Regarding Mad World, I found it interesting-- but not a
bad thing-- that at the Cinerama Dome they ran the film with no aperture plate,
but with significant masking on each side. (See
http://hometown.aol.com/mlutthans/myhomepage/photo.html for a crude pictoral
representation.) Note that Mad World in unrectified UP presentation in October
used the ENTIRE width as used for 3-panel Cinerama, which in the dome is
roughly 120 degrees. Some sources have said 126. Regardless, it is well short
of 146. If you continued to enlarge the IAMMMMW image to fill the height, it
would also expand outward. I'm not sure that it would have filled an entire
146 degrees if enlarged and cropped in the aperture, but it sure would have
come close. Some very, VERY rough math: The image size would have to be
enlarged by roughly 25% to entirely fill the height of the screen at the
edges-- which was standard practice at many Cinerama Theatres in the 1960s.
(Butterfly apertures were common.) If they did that in the dome for IAMMMW,
the width would have also increased by the same percentage (or even more-- this
is anamorphic U-P) as the height, but since projected image width and actual
screen width do not correspond on a deeply curved screen, we would probably be
talking about roughly a 15-20% increase in width. Let's say 15% to be
conservative. A 15% wider portion of the arc would come out to 138 degrees
filled, and 20 % would be 144 degrees filled, top to bottom. Of course, about
20-25% of the image would be missing from the middle of the picture, but for
some reason, most people don't seem to mind that. (It drives me up the wall,
but I'll admit that I'm in the minority.)
Distortions on curved screens also have so much to do with where YOU SIT in the
building, and even the rectification process only decreased the perceived
distortion in some seats. It added distortion in others! Additionally,
IAMMMMW was the first single-film "Cinerama" movie, and was probably projected
with considerable keystoning in many facilities, since it was pretty common for
Cinerama 3-strip facilities to have slightly high (under about 4 or 5 degrees
off-center) projection booths. (Again, it didn't much matter for 3-strip.)
I've got blueprints for the Cinerama in Honolulu, for example, and it had a +2
degree angle of projection. The Martin in Seattle had a zero-degree or "dead
center" projection, as did the D-150 just down the street. Oddly enough, some
architects didn't much worry about this problem even when a building was
conceived for single-strip 70mm, as the final 70mm Cinerama theatre built, the
Southcenter in Tukwila, WA, had about a 4-degree downward projection -- and
significant keystoning, especially in 70mm and scope. What I'm trying to point
out is that many people equate curved screens with "smiley face" distortion.
This is often evident, but it can be avoided with proper theatre design. The
folks at D-150 were very aware of this, for instance, and sought to always have
dead-on projection.
A few final ramblings:
1. 2001 at the Cinerama in Seattle used the butterfly aperture. The opening
copyright ran almost off the screen at the bottom. I mean it was right on the
edge above the carpet. The image filled about 120 degrees of the arc, with
black masking dropped along either side, as you would expect. 2. One scene
about IAMMMMW has always made me question its original presentation. If indeed
butterfly apertures were used, how did that shot in the 2nd half when all the
"heads" of the main characters are shown looking down into the big hole under
the Big W look? Were the heads on the bottom (and/or top) of the shot just get
cut off? That seems to be a shot which really cries out "full aperture" to me.
3. This is purely a matter of semantics, but it seems to me that many of the
problems of projecting onto a curved screen are framing issues as opposed to
distortion issues.
4. It would be great if there was a lens that would increase the height of the
projected image gradually as you approaced the edge of the image to correspond
with the consistent height of the curved screen. Maybe someday....but of
course, it would actually be adding distortion to the image! It might look
good though.
I've given myself laryngitis of the digits. Sorry about the babble.
--Matt Lutthans, Seattle