[Openicc] Assigning icc profiles to driver settings

[edmund ronald]

Please understand that in this debate I am not arguing an abstract
position, I am explaining the intended direction of Gutenprint, which
has been in transition for a number of years as features realized in
code fro each printer have been progressively migrated into more
accessible parameters. The intention is now to expose as much as
possible of the print system to power users and domain experts (print
experts) in order to turn them from simple users into print-recipe
contributors.

The thing about driver settings is that they're not monolithic, they
are trees. For Gutenprint, we would like that a domain expert who
creates a media/ink/printer combo pick and choose from what went
before, then adjust what needs to be adjusted. This is reassembling a
tree from a bunch of pruned branches.

Let me give an example:

To print on some known A1 rag sheet on an Epson 7880, with sheet feed
(I am making this up):

You could take eg. the model inking and curves from one set (Robert's
Trivial Matte 7880), the raster dot pattern (Robert's720x360
Epson7800)   the art paper platen settings (Kai's art paper platen
setting 9600) , a paper transport adjustment (Kai's art transport
adjustment 9600) a paper size (Graeme's mini-poster  A1), and a choice
of paper feed (7600 sheet). etc etc.

In other words you are ripping settings from a bunch of *other model*
printers which users have experimented with to create this one.

At the moment, only the programmers who go into the code and build
system can really do the things above. But these programmers are
running out of steam faced with the combinatorial explosion of media
types and printers. So the idea is to expose more settings and let the
users do more of the work themselves.

I think that working with these trees is easier in XML than in the ICC
profile format.

Edmund

On Wed, Jan 19, 2011 at 7:27 AM, Kai-Uwe Behrmann <ku.b at gmx.de> wrote:
> Am 19.01.11, 02:16 +0100 schrieb edmund ronald:
>>
>> On Wed, Jan 19, 2011 at 2:14 AM, edmund ronald <edmundronald at gmail.com>
>> wrote:
>>>
>>> Graeme wrote:
>>>>
>>>> Because it's compact, and quite easily read and written by computer ?
>>>> Because the code to read and write ICC profile tags is already present,
>>>> and there's no need to introduce another separate library to
>>>> do a similar thing ? (You certainly don't want users to have to use a
>>>> text editor to configure the printer!)
>>>>
>>>
>> BTW, I disagree that profiles are easy to read and write by computer.
>> I wrote a camera profiling tool some years ago, and the color code
>> worked fine, but I never managed to write usable profiles with
>> anything except Matlab.
>
> For open source projects we offer support here on this list. The same is
> true for the argyllcms, lcms and oyranos lists.
>
> And IccXML had changed the rule for textual editing of ICC data.
>
> kind regards
> Kai-Uwe Behrmann
> --
> developing for colour management www.behrmann.name + www.oyranos.org
>
>