[Tig] Fw: CRT scanners & anamorphic/unsqueezed 2.35:1 (TIG question)
Mon Feb 25 17:05:27 GMT 2002
A note from the folks at Cintel that I believe might be of interest to many
on the TIG:
eCinema Systems, Inc.
ecinema at pacbell.net
----- Original Message -----
From: "Peter Swinson" <peter_swinson at compuserve.com>
To: "Martin Euredjian" <ecinema at pacbell.net>
Cc: "Adam Welsh" <awelsh at cintel.co.uk>; "Gary Welch" <gwelch at cintel.co.uk>
Sent: Monday, February 25, 2002 3:37 AM
Subject: CRT scanners & anamorphic/unsqueezed 2.35:1 (TIG question)
2.35:1 Scope image data scanning. What H & V pixel count
A good question.
Depends on how you wish to scan/sample, a choice with CRT based systems.
I assume the requirement is to scan the whole frame, therefore I have not
addressed TV type pan scan.
Basically two modes are available.
1: Preserve the anamorphic squeeze.
2: Generate unsqueezed data files.
The difference really depends on whether your religion is square pixels
and/or constant film surface resolution. In addition how dedicated are you
to precise 12um (2K) or 6um (4K) film area sampling.
Lets look at the raw squeezed version. This is nominally 1.175:1 with a
higher frame height than normal 35mm acadamy. Lets set the horizontal
sampling to the (2K) data standard of 2048 pixels. To maintain square
pixels relative to the film image AND the film's intrinsic resolution, a
vertical pixel count of 1743 would be required. When the data is
unsqueezed to 2.35:1 the pixels will no longer be square but will aquire a
2:1 aspect ratio. Bear in mind that this scenario provides a data
resolution that equates to a resolution of about 11um FILM RESOLUTION as
the 2048width is over an acadamy width rather than a S35mm width.
Now whether this is good enough depends on ones point of view.
Cintel would suggest that 35mm film offers better than 12um resolution with
all but the worst camera lenses, most film material is certainly capable of
In such circumstances precise 6um pixel counts, relative to camera
apertures would be 3683 horizontal pixels by 3112 vertical pixels.
What we do provide is a 4096 x 3112 scan sampling which represents S35mm
camera apertures. Now it is fairly obvious that setting the CRT raster to
fill the anamorphic height, which is near the S35mm height will entirely
encompass the anamorphic image. At scan time we then crop the horizontal
pixel count to 3683, thus removing the soundtrack area before storing image
files of 3683 x 3112 square pixels.
This scenario offers sufficient resolution to make 2:1 scope unsqueezes
without loosing resolution. The unsqueeze will give 2:1 pixel aspects
however a simple 2:1 interpolation of the vertical 3112 to 1556 would re
assert square pixels giving an unsqueezed file size of 3683 x 1556. This
represents 2.367:1. As you will realize different horizontal pixel count
crop will provide different ratios.
Of course we can apply the same rules to 2K scans giving a squeezed file
size of 1840 x 1556 square pixels.
Being CRT scanner manufacturers we have the luxury of designing scan
formats for whatever area you want and then sampling the area however you
want. Sounds easy, however we tend to offer particular industry standards
rather than infinate solutions, so as to avoid a "Tower of Babel" of non
Hope the above helps
ps Feel free to post this to the TIG if you think it appropriate.
More information about the Tig