[Openicc] Assigning icc profiles to driver settings

[Graeme Gill]

Leonard Rosenthol wrote:
> On Tue, Jan 18, 2011 at 11:26 AM, Jan-Peter Homann<
> homann at colormanagement.de>  wrote:
>
>> 1) PPD colour keywords:
>> The PPD for a given printer must be able, to describe all color relevant
>> driver settings with all parameters (incl. low level driver settings if
>> necessary)
>> an ICC profile will be assigned to a proper parametrized PPD and the PPD
>> will completely set up the color options in the printer driver.
>>
>>
> I (and other folks at Adobe) support thsi model.  In fact, we've been trying
> to convince various printer vendors to do this for their PPDs for YEARS.

This is only a good approach for PS based front ends to the printer, where
a PPD parser is readily available. It is less desirable for other, simpler
front ends (ie. a direct raster print pipe).

>   However, all of the major vendors are unwilling to do so for various
> business and/or technical reasons.  So any provided PPDs would need to be
> MODIFIED versions of the originals...
>
>> 2) Driver settings implemented as metadata-information into the ICC-profile
>> Like with the PPD-workflow, all color relebant printer driver settings are
>> stored as metadata into the ICC-profile.
>
> Modification of ICC profiles is a BAD IDEA!
>
> As noted elsewhere in the thread, it breaks the ID and prevent the use of
> the profiles in other workflows where comparison of profiles is important.

Another disadvantage of this approach is that it makes it hard to
re-use profiles for similar but slightly different printer set-ups.

One of the practical issues is that there is a combinatorial
explosion of printer settings, so it's impossible to profile
all possible combinations. One of the advantages of separating
calibration from profiling is that it's possible to
mix and match for good (but not perfect) color. This involves
using profiles for set-ups other than the one they were created for.

For instance, changing the screen pattern or DPI often changes
the response transfer curve, but it's feasible to calibrate
for many of these combinations, and then mix it with
a profile that has been made for that paper & ink combination,
yielding quite reasonable results.

Graeme Gill.