Re: multichannel profiles
Re: multichannel profiles
- Subject: Re: multichannel profiles
- From: "joe borne" <email@hidden>
- Date: Wed, 15 Nov 2000 17:14:56 -0500
>
> We are not adverse to spending reasonable money to achieve results
>
> inhouse. Can anyone out there recommend a 6-ink/multichannel
>
> profiling package? (That perhaps works with the ColorSavvy CM2s
>
> spectro)
>
> --
>
I think you are looking at Monaco or ColorBlind
>
Regards,
>
Steve Upton
You can make 6 color profiles with ColorBlind or Monaco and I have done
both. Doing so can be a long tedious process if you want it done right.
However, there is a basic flaw in all of the systems that are currently
creating 6 color profiles.
The mathematics in use are essentially nothing greater than the old UCR/GCR
color replacement methods. For example: in a typical Hexachrome profile,
wherever there is a mixture of magenta/yellow that can be replaced by
orange, it does so. Then at the end of that range of hue, it can reach out a
bit farther to get more vivid oranges than the tint builds can provide. This
is nothing more than old world color-replacement math with a little stretch
on the end (so to speak). Not that this is a bad method, its been the best
darn thing so far.
But, what would really be the correct way of doing it is to generate a
profiling target that shows all of the tint values you have available when
you have 6 inks to mix.
Using simple math: We know that there are 100 possible tint values of each
channel in CMYK. Therefore simple factoring of 100 to the 4th power is the
total number of possible combinations (= 100,000,000, a billion).
Fortunately things like ink limits and other factors reduce this number
significantly. If you use the law of permutations and limit the calculations
to groups of 3 with no redundant sets (321 = 123), it also begins to drop.
And since most combinations interact in a predictable manner we can usually
use less than 1,500 patches to predict a CMYK device that behaves in a
reasonable linear manner.
Now, every time you add a channel and try to do a REAL mapping, you increase
the necessary number of patches by a factor of 100. So your initial
necessary number of patches jumps to 1,000,000,000,000,000. (Someone please
jump in a tell me what the next iteration of the "illion" is above trillion,
because that is the proper term here.) If you use grouping and the law of
permutations to reduce this you get a much lower number. But you still have
an astronomical figure.
Some would argue that this method is not needed. They believe that we would
be simply be creating colors that already had matches in the existing CMYK
gamut for most of the mapping. I do not agree, because combinations of 7
color systems using specific hues of red and yellow can reach out and hit
colors in the orange hue range. I have personally demonstrated this
phenomena. With replacement math systems, this is not possible. Extra gamut
colors are not mixed with CMYK primaries to create new tints. Its a shame
that current mathematical models in use by profiling packages are
insufficient to this task. They will provide better color for sure, but are
only scratching the surface of the potential.
So what is the solution? Well, I can tell you that I and some others are
delving into the math. There are skunk works in progress to achieve this
goal. I foresee a "true" multi-channel profiling package being available in
the second quarter of next year.
--
Joe Borne
Color & Graphic Technology Consultant
(859) 282-0393
-- "The worst thing to do is nothing" --