> Objective:
> I want to get a fair estimation of the Lab colours from the ICC profile,
> when using relative colorimetric conversion.

Hi Auke,

do you mean, you want to estimate those L*a*b* values, which you can
_measure_
on the printed paper, if you perform a relative colorimetric transformation
from e.g. Adobe RGB to printer RGB?

This would corespond to a proofing transformation

  Adobe RGB --relative-colorimetric--> printer RGB --ICC-absolute--> *Lab

Example:

Q:
Which CIELAB color can I expect to measure on the printed paper, if I
convert
AdobeRGB=[150,160,170] with relative colorimetric intent to printer RGB?

A:
$ icctrans -c0 -t1 -i AdobeRGB1998.icc -o *Lab -p printer.icc -m3
little cms ColorSpace conversion calculator - v1.8

Enter values, 'q' to quit
R (0..255)? 150
G (0..255)? 160
B (0..255)? 170 

L*=59.5236 a*=-3.4766 b*=-12.1094

Is this what you want?

> c. I don't *wan*t to use absolute colorimetry as it changes the
> RGB values too much. For instance white (Lab 100,0,0) is trnaformed
> to 229,246,255.
> If I print this, I don't get paper white anymore, yet my paper is
> becoming even more blue :(

That's pretty strange, IMO rather the opposite should happen.
If PCS white (D50) is transformed with ICC-absolute intent to the printer
color space, then I would rather expect that the printer puts down a bit
of yellow ink to make the blue-ish paper more neutral.

Or do you possibly talk about an ICC-absolute conversion from AdobeRGB
(or sRGB) to the printer color space? This will indeed produce a _strong_
blue-cast, because AdobeRGB and sRGB have a D65 whitepoint.

Please notice, Marti and I are just talking about about an ICC-absolute
transform from the _profile connection space_ (PCS) to the printer
color space, and _NOT_ about an ICC-asolute transform from a source
RGB space (e.g. AdobeRGB) to printer RGB space!

Btw, I'm still not sure:
Do you use the generic profile supplied by the manufacturer,
or did you build your own profile? I think you said, that
you're using ProfileMaker?

> Some thoughts:
> The ICC profile *knows* about the actual colour <> RGB
> relation (it is the major function of the profile).
> Therefore, I would think, it should be able to specify,
> approximately, using its 3D LUT, what the Lab values will
> be when I print using a particular RGB combination.

Right, that's exactly what the ICC-absolute intent is supposed to do.
If you transform printer RGB values with ICC-absolute intent to CIELAB,
then you are supposed to obtain the CIELAB values _as you can measure_
them on the printed paper.

Example:

$ icctrans -c0 -t3 -i printer.icc -o *Lab
little cms ColorSpace conversion calculator - v1.8

Enter values, 'q' to quit
R (0..255)? 255
G (0..255)? 255
B (0..255)? 255

L*=91.7417 a*=0.3047 b*=-6.5195 

where the result [91.7, 0.3, -6.5] is approximately the
measured paper color.

Btw, all what I said, applies under the ancillary conditions that
- you built the profile for D50 illuminant
- you did not create the profile with FWA compensation
- and that the measured L*a*b* values are D50-based L*a*b* values


As Marti said, first try to compare the results of "icctrans -c0 -t3
-i printer.icc -o *Lab" for a couple of printer RGB values (e.g. 100
patches) with actual measurements. If you achieve an average error of
<= 3dE for say 100 patches, using a custom profile, then IMO you can be
pretty contented, for a consumer printer.

Don't expect accurate results from a generic, manufacturer supplied
profile (they often have a large error, nevertheless the perceptual
intent can still give reasonably pleasing results).

Also don't overestimate the capabilities of consumer inkjet printers.
In his paper "To Develop a Universal Gamut Mapping Algorithm", Jan Morivic
has also briefly evaluated the repeatability, spatial uniformity and
temporal stability of a DeskJet 850C. My conclusion of his evaulation is,
that
depending on the used consumables, an average error of up to 3dE, and
max. error of up to 10dE can be easily explained by the flaws of the
printer.
Another conclusion is, always grant 24-48 hours (!) of drying time after
printing a test chart, before making measurements!

Regards,
Gerhard

-- 
NEU +++ DSL Komplett von GMX +++ http://www.gmx.net/de/go/dsl
GMX DSL-Netzanschluss + Tarif zum superg�nstigen Komplett-Preis!


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now. 
http://productguide.itmanagersjournal.com/
_______________________________________________
Lcms-user mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/lcms-user

Reply via email to