Re: RGB vs CMYK: some experiments

From: Katharine Thayer ^lt;>
Date: 12/04/04-02:14:43 AM Z
Message-id: <>

Jason DeFontes wrote:
> Katharine,
> It's clear to me now that you are right about the CMYK settings affecting
> the results in my RGB channel example. I realized it while I was sitting on
> the metro today. I only did what I did to prove to myself that you could get
> anything at all from RGB separations, and in that regard I am both surprised
> and satisfied. I am in total agreement with you that RGB and CMYK are both
> models that are different ways of describing the same thing, and either is
> capable of producing equivalent results. Had I not done it for myself (in my
> admittedly limited way) I would not have believed it.

I think what you did was great; I wish more people would go to the
trouble of working things through for themselves, in whatever way makes
sense to them. It doesn't matter that you were a little off in what you
thought the visual meant; the main thing is that you proved something to
yourself and showed us the results. That kind of thinking and sharing is
always useful and productive in my opinion.

I was just so thrilled when I realized what I was looking at: an actual
visual of what the default CMYK does to the color values in a whole
image. I'd been looking at the question one pixel at a time, one color
field at a time, going around and around through the numbers til I was
sure I understood what was happening at each step, and I was just very
interested to see the effect in a whole picture instead of in a color at
a time.

My gum prints from default CMYK separations (that I spoiled the first
set of by doing them on Arches paper which made them all speckly, so I
have to start over next time I get a chance-- those two of the grey
patch were the only ones that survived that effort) don't look nearly
that good in their color balance, but that doesn't mean anything at
all. I'm only doing this comparison because someone, in the last go
round of this discussion, challenged me angrily to print some RGB and
CMYK separations side by side and show the results. So I'm committed to
doing it, very slowly and bit by bit when I have the physical strength
to go out to the studio and work at it, even though I don't think
there's anything much to be gained by it.

The reason I say there isn't much to be gained by the comparison is
that, say for example, in my comparisons, the RGB prints look fairly
close to an accurate color balance and the CMYK prints look like
something otherwordly, but that's because by experience and practice and
habit my printing method is calibrated to printing RGB separations. I
can't help it, it just is. So when I run a comparison holding everything
constant, using the same printing method for both prints in order to
give a true comparison of the separations, then the RGB print looks
right and the CMYK print looks like something totally else. I don't know
that either of them would qualify for Joe's label of "accurate" but at
least the RGB print is fairly realistic in color; in most cases the CMYK
print is not even close.

I don't have energy or interest to develop a good printing method for
CMYK separations; I am willing to assume that given time and practice I
could learn to print them at least within the ballpark. But someone who
is used to printing from CMYK separations and who gets good results from
them, their printing method will be calibrated to the CMYK separations,
and for them a comparison between the two should come out the other way
around: a nicely balanced print from the CMYK separations and a
weird-looking print from the RGB separations, which is what Joe showed
us a while back.

To me the question of which kind of separation gives *better* balance
(in some kind of larger, RULE OF GUM PRINTING, sense) is not an
interesting question, because it can't be answered in any definitive
way. We can only answer it each for ourselves and for no one else. My
answer is that I prefer to use the straight CMY separations because I
don't see any reason to use a profile that was developed for SWOP inks,
and I like the results I get from the RGB separations. If I didn't like
the results, I would be more inclined to manipulate the curves directly
rather than do a custom CMYK setup, because I don't think in CMYK terms
about gum printing. But as I keep saying, there's no one right answer to
this question. Someone else could say well, the SWOP profile works great
for me. And it sounds like you will want to develop your own CMYK
profile; both of these and any other possible answers are as good as my
own choice. Whatever works, is the right answer.


> By the way, thanks for the level of effort you've made to describe and
> illustrate your process.

Thank you so much for that; you're very welcome.

> > -----Original Message-----
> > From: Katharine Thayer []
> >
> > Hi Jason,
> > Thanks for the great visual, which shows how the default CMYK
> > alters the
> > CMY values from the true values.
> >
> > If you had inverted the RGB channels, turning them into negatives for
> > CMY, and had an expert gum printer print them on gum, I would expect a
> > closer approximation to the original than you got by importing the
> > channels to CMYK, which of course altered the color values to
> > match the
> > SWOP profile, which was my whole point, and thanks for the
> > illustration.
> >
> > It looks like the settings dialogue box has changed since my
> > version of
> > Photoshop (I kept up from versions 1 through 5 and then stopped
> > upgrading) although I did check to make sure that the CMYK default
> > settings haven't changed in the newer versions (they
> > haven't). It's too
> > bad that the new color settings dialogue just names the CMYK profile
> > rather than describing the settings. I go through all the settings on
> > the page that I would guess you didn't take the time to read,
> > since your
> > page seems to show no knowledge of the information contained therein,
> > such as how an RGB file turns CMY when you invert it, and how the
> > default CMYK alters the true CMY values to match a profile that's not
> > terribly relevant to gum printing. I'll give it here again
> > just in case,
> >
> >
> >
> > Katharine Thayer
Received on Sat Dec 4 10:10:41 2004

This archive was generated by hypermail 2.1.8 : 01/03/05-09:29:43 AM Z CST