[Openicc] linearisation targets, Argyll, G7

[Graeme Gill]

Jan-Peter Homann wrote:
> Hello Graeme, hello list
> I don´t think, that a linearization for a shared visual appearance is 
> only useful for set up CMYK printing presses.
> Especially in in the Inkjet area, we have a lot different printers and 
> media. If the the linearization targets for all of this printers / ink / 
> paper combination lead to to similar gradation and gray balance, 
> workflows with standard profiles will be much more stable and predictable-
> 
> Most users will not need individual profiles if the usage of a standard 
> profile for a paper class on a linearized device will lead to the 
> expected gradation and gray balance.

This is not in agreement with my experience. While I'm sure such an
approach can be applied, and may well improve compatibility when
printing CMYK without profiles, the reality is that inkjet and laser
printers are not very like printing presses. Most presses that are
setup to meet one of the standards will be using ink-sets that meet
the standards chromaticities. In contrast digital printers often
use a wide variety of inks with highly non-standard chromaticities.

A calibration process that attempts to meet a CMY grey target
is going to be far more complex than one that just makes the
individual channels match a per channel response target, and I think
this added complexity will be for dubious benefit. The best approach
in my opinion is not to attempt CMY grey calibration with popular
digital printers, but to use a CMYK input profile to provide compatibility
with standard CMYK files, in combination with the device specific profile,
a separation to it's inkset and per channel calibration

The other thing that is often done with printing presses is to re-target
the CMYK using a 'through' black CMYK->CMYK device link. My experience
with this type of thing when applied to digital printers is that it
is also of dubious benefit. The GCR has too many side effects to leave it
to the input files separation algorithm. Instead it needs to be setup
carefully for the particular printing device, to strike the appropriate
balance between ink load, "black grain", illuminant and observer variation
robustness.

> As many printers have very strange curves in the pure linear state, the 
> creation of 1D LUTs for a shared visual appearance is an iterative 
> process. Black is calibrated by firstly printing a pure black scale. 
> Based on the maximum darkness, which is reachable on the device, the 
> appropriate target curve for the linearization is choosed.
> 
> For the CMY-scale a scale with different near neutral CMY patches is 
> printed and measured or visually compared to reference print samples.
> The linearization goal could e.g. be, that C50 M40 Y40 should lead to 
> L*57 a*0 and b*-1.
> If e.g. the CMY patch C30 M25 Y30 leads to L*57 a*0 and b*-1, the 
> correction-curve for linearization should correct C50->30 M40->25 and 
> Y40->30.

This sounds like a difficult process to automate for unknown digital
printers. Unlike a press, the devices behaviour cannot be assumed.
It can't even be assumed that a particular test set of near neutral
CMY values will in fact be near neutral, nor can the ink mixing model
make any assumptions about how the test values should be interpolate
to arrive at the correct output grey.

> As the eye is very sensible for the neutrals, a re-linearization / 
> calibration to a good gray balance and gradation will do the job, most 
> users expect for a good printer-setup.

Right, but a good profile will provide this, without the need for
CMY neutral calibration.

> Both the basic linearization and also the re-linearization/calibration 
> could be done either by measurements or by visual comparisons with 
> reference print samples. As I described in my last e-mail.

One of the features of a useful calibration system should be that
re-calibration should be fast an simple. Printing a set of test wedges,
reading them in and then instantly being calibrated is pretty close
to this. I'm sure someone setting up a press for many thousands of
prints for a job worth thousands of dollars is prepared to take a bit
more time and trouble than someone taking care of their personal inkjet on
a day to day basis.

> If anyone is interested to make more real world tests concerning a G7 
> based ink linearization with GutenPrint, you could contact me off-list.

It certainly sounds like G7 has an important part to play in press
standardisation, but I'm far from convinced that it is the best approach
for personal digital printers (or even proofing).

cheers,

Graeme Gill.