Re: Spot Color tables in Best ColorProof
Re: Spot Color tables in Best ColorProof
- Subject: Re: Spot Color tables in Best ColorProof
- From: Roberto Michelena <email@hidden>
- Date: Mon, 02 Jun 2003 01:01:53 -0500
Marco,
>
> That means, for the in-gamut Pantones, filling their Lab
>
> values into Best's table should be enough.
>
>
That may be true some of the time, but each printing device lives within its
>
own color space, and the same coordinates that describe an LAB color in the
>
abstract do not necessarily produce the closest result to the ideal on the
>
device at hand without further intervention. One may still have to move the
>
values around somewhat to find a closer match, specially if the Delta E
>
starts to exceed 3.
Even if the dE is zero, that doesn't guarantee a visual match; the best
possible visual match can only be achieved by visually tweaking.
Sounds terrible, but the fact is that too many things intervene: the
instrument used to measure the library is not the same brand and model than
yours; the light source under which you view is not exactly D50; the
"standard observer" functions do not match your rods and cones; the ink and
paper contain some fluorescence; etc etc etc...
Nevertheless, for most purposes a good numeric Lab match is a good enough
match in most jobs.
>
> Meanwhile you could try a Photoshop trick in which you paint Lab gradients
>
> around the value you want, then transform to the device profile and back to
>
> Lab, and subtract that resulting image from a flat tint of the Lab value
>
> you wanted to achieve.
>
>
Correct me if I'm wrong, but that device profile in your test would need to
>
be something other than the one created via a CMYK test chart for CMYK
>
output (say, the TC3.5 CMYK): it would need to be one that charts the
>
LARGEST printable color range on the device
<snip>
>
In other words, if the device's CMYK profile only describes A PORTION of all
>
possible colors achievable on a specific inkjet printer, how do you map line
>
colors that are outside that CMYK range but still INSIDE the device's
>
printable range?
Marco, you got me wrong. The device profile I am referring to is the proofer
profile. It is obviously not a portion of itself, and it describes all
colors achievable by the proofer.
I will take more time now to describe the process in detail; I've not tried
it in print, just made it up now and tried in Photoshop, but it seems to be
valid and I don't see why it wouldn't in real output.
Imagine you want to achieve a Pantone whose Lab values are 80,-40,50 and it
lies outside your proofer's gamut.
1) make a new blank Photoshop document in Lab; a small square, for example
2x2 inches. Select background color of 80,-30,50 and foreground of
80,-50,50. Fill with vertical gradient.
Now select background color of 80,-40,40 and foreground of 80,-40,60. In
channels palette deselect "a" channel. Fill with horizontal gradient.
Right now, you have a square that has your desired Lab value at the center
(80,-40,50) and a range of neighbor values around, at a fixed Lightness but
varying in a* and b*. That is called a portion of "slice" of Lab space cut
at the L* axis.
2) duplicate this document 8 times (image->duplicate), for a total of 9
identical documents.
In each of the copies, fill the L* channel with a solid value, stepping from
76 to 84 and skipping 80 (because that's the original).
Right now you have 9 documents, which only differ in their L* value (always
solid), which steps from 76 to 84.
3) Assemble those 9 documents into a single one (size would be 6x6 inches if
each one was 2x2 inches). For that, you should expand the canvas size of the
original one, keeping the image at the center, then move each copy into that
canvas and position it, finally flatten. Everything is still Lab. Now you
have a single document, let's call it Doc-1.
4) Make a copy of that document (image->duplicate). Convert the copy (let's
call it Doc-2) to the proofer colorspace, by choosing "convert to profile",
with absolute colorimetric rendering.
Then convert it back to Lab, again absolute colorimetric.
5) What you have now in front of you, called Doc-2, is the Lab values that
(according to the profile) will get actually printed when you request the
original Lab values of Doc-1.
From here, you could do channel operations ("difference") between the L,a,b
channels of Doc-1 and Doc-2, and get delta-L, delta-a and delta-b grayscale
documents. But that hits a dead end since there is no way to convert those
grayscale documents into something resembling a usable delta-E.
Instead, use Photoshop's built-in assessment of what constitutes a close
color, in the following way:
6) with Doc-2 in front, set your foreground color to the desired color
(80,-40,50 in our example).
7) use "select->color range" and you'll see where the closest colors to your
desired "ideal Lab color" are!!
8) Even more, to really pin them down, do set the fuzziness to something
useful and press ok, and then save the selection in a new channel, deselect,
and adjust those channel's levels (via "treshold") to eliminate everything
but the really closest ones, and make those 100% selected ("Load selection")
9) Take note of the coordinates where those "closest" colors lie. By looking
up those coordinates in Doc-1, you'll know which Lab values to request in
order to get the closest match.
There will probably be more than one point with the same "similarity"
(Photoshop's equivalent to dE); and in fact, there is more than one way to
achieve a delta-E, and it corresponds to an intersection of what might be
called a "delta-E sphere" around the target color, and the gamut surface;
you might want to make a patch of each, print them and visually select one.
Long but sure beats filling up the wastebasket with aimless trial and error
printouts. Helps the environment, too!
It depends heavily on the profile's precision and probably even on the CMM,
so I don't know if it will actually work... Besides the disclaimer that the
lowest delta-E doesn't necessarily correspond to the best visual match.
I think I'll try it myself soon.
Best regards,
-- Roberto Michelena
EOS S.A.
Lima, Peru
_______________________________________________
colorsync-users mailing list | email@hidden
Help/Unsubscribe/Archives:
http://www.lists.apple.com/mailman/listinfo/colorsync-users
Do not post admin requests to the list. They will be ignored.