[Tig] Neophyte with calibration questions

Martin Euredjian ecinema
Mon Apr 28 03:36:01 BST 2003


From: Chris Swinbanks

> ... 3D mapping of the colour ...
> ... not addressed by simply adjusting the tonal scale of the monitor with
a 2D LUT.

This is an interesting emerging issue, particularly with various new
display/projection technologies.  I've been working with 3D LUT technology
for about a year and a half, maybe two years.  Someone correct me if I'm
wrong when I say that there are no workstations or monitoring systems in the
market today that allows real 3D mapping (I'm working no one but it's not
out yet).

The problem is that one possible implementation (10 bit, 4:4:4 or 22:22:22
YCbCr or RGB input) would require at least 3 to 4 Gigabytes (depending on
output width) of LUT memory.  That's a chunk.  And that's a chunk that is
hard to adjust in real-time, as in, turn a knob and see the result in the
next frame.  One box I'm building does allow for real-time tweeking of this
huge LUT space, but it has been a real challenge to say the least
(~100GByte/s processing).

Addressing the same issue for 12 bit RGB systems is even more difficult,
with a mind-bogglind 200 to 300GB LUT.  Of course, there are ways to work
around this, but the fact remains that every input value must be evaluated,
which leads to a minimum of 2^30 (for 10 bit components) to 2^36 (12 bit)
input values.

Oh, yes, and, let's not talk about doing this in real time in 2K or, worst,
4K.  Lots of data/second.  Probably doable, but very, very expensive.

Also, there is no way to contort any YCbCr space into the greater RGB space
and recover colors that couldn't be represented in YCbCr.  If they didn't
exist in the first space they are not going to be found in the second. That
might not be an issue for the workstation guys working in RGB, but it is for
those doing digital cinema in HDTV (22:22:22 or 22:11:11, it doesn't
matter).

Finally, the last link in the chain is the monitor itself and its interface.
Some are not capable of more than 8 bits per channel, regardless of how rich
the source information might be.  It is my opinion (and I am biased on this)
that CRT's are on their last legs.  They lack performance in areas that
other technologies seem to be addressing better and better with time.  Maybe
CRT's will be delegated to television use, where, by definition, they are
the standard.  Higher contrast, dynamic range and resolution (both spatial
and color) devices will probably surface in the future and gain a strong
foothold in the digital cinema domain.

A question I've been asking myself for a while is: How does the viewing
environment affect the dynamic range and color resolution that the human
vision system is capable of perceiving?  In other words, is it a waste of
time to try to get to 10 or 12 bits of color depth unless the viewing
environment is laboratory-controlled?  I don't know if anyone's done
research in this area.


~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Martin Euredjian
eCinema Systems, Inc.
voice: 661-305-9320
fax: 661-775-4876
martin at ecinemasys.com
ecinema at ieee.org
www.ecinemasys.com












More information about the Tig mailing list