[Tig] Color Grading in NLEs

Craig Leffel craig
Thu May 8 00:49:44 BST 2003

 Kevin wrote:

> How do you choose your color combinations? I asked a number of
> colorists I know and they have not heard of using color theory to assist
> in color choice. I assume that once you get to be a pro colorist, this
> is more intuitive-I can understand that. But am I on to something with
> this idea?

I would say you need to re-think your approach. Coming up with a color
balance that "works" will always be dependent on the material you are
starting with. You need to break your thinking into; Primary, Secondary, and
Terciary reasoning. I don't mean secondaries in the classic color correction
model either. I mean RBY are primaries, and their combinations are
secondaries.  Combinations of 2 secondaries, or a secondary and a primary
produce a terciary. In video, as well as photographic printing, all colors
can be derived from 3 basic primaries. The only difference is Green becomes
a source for luminance as well as a primary. It's STILL a blend of Blue and
Yellow ( original real world primaries ).Relationships of the original
primaries in a finished grade are based purely on the source material and
what the desired outcome is. I have been involved in a number of dicussions
about this subject for workstation color enhancement programs. There is NO
"preset" my friend. It takes work, experience and an education either from
an institution or the school of hard knocks. Do not expect that you will be
able to come up with "preset" pallettes of color combinations and have it
make any sense to anyone operating the box. We are in the middle of doing a
number of webisodes / commercials / DRTV work for the United States Army. We
are "color correcting" much of it in an Avid DS. The editors would like to
know how they can save what we do. It's really a problem for Avid and the
color correction impaired programmers they have working for them ( or FCP
for that matter ), but the idea is that there should be a way that you can
save color corrections that you build in a seperate list, with a picture
attached. Just in case anyone is listening, 2 pictures would be better. One
of the image before color correction, the other AFTER. Then the editor ( or
better - Colorist ) might have a slight clue about how the color correction
is going to affect the image they are working on. Simply naming them and
saving a file is not going to cut it. Try and come up with a good name for a
correction that adds 50% contrast, removes all blue, adds 25% saturation,
leans green about 10% more than Red, has a matte channel that acts like a
power window for a vignette, has a luminance curve manipulation, and has a
gamma push. EVEN IF you could name it something that fits, the only way
anyone could use a name only system is by defining a large list and playing
the double click game down the entire list until one works. That does not
sound efficient to me at all. I have a better idea, just like I tell our
editors, when you need to color correct something, call the colorists and
one of us will come down and sit in your chair for a few minutes...

BTW, you can tell the folks at FCP to throw away the curve base portion of
their color corrector. Why all the yokels out there building these things
think a curve based part of the corrector is vital is beyond me. I realize
many graphic artists use this kind of system everyday. Up until now they
haven't had any better tools. "Color correcting" by moving points on a curve
and waiting to see what happens is like drinking beer out of a baby's
bottle -- it's the wrong tool for the job. Curve based color correction is
for specific tasks -- not for making images. While your at it, throw out
Histograms too. They're just as worthless.

Hope it all helps Kevin --

Yours in Knobtwisting,

Craig Leffel
Senior Colorist

More information about the Tig mailing list