BW Printing and ICC profiles... why use a profile...

[Ferp]

Many thanks for your answer, I really appreciate it.

Likewise.  I now have a pretty clear idea of what you're doing, except how you create the DeviceLink profile.  Is there any more detail info on this? 

I agree that you can effectively make ABW linear within Photoshop, but I still that this is beyond the scope of the typical ABW user, such as the OP.

With ABW from Photoshop, there is still the issue of how PS prints untagged or GG2.2 images when you select "Printer Manages Colors", which is made more complex by the fact that when PS does a silent profile conversion to such images is different on Win and OS X.  It's amazing how little understood this is.  I don't know how your DeviceLink profiles are treated, but it may be enough to "solve" the problem.

[digitaldog]

Likewise.  I now have a pretty clear idea of what you're doing, except how you create the DeviceLink profile.  Is there any more detail info on this?

You need software to do this, usually kind of expensive (for example. I1Profiler with Device Link module: http://www.rpimaging.com/store/manufacturers/x-rite/x-rite-i1-publish-pro-2.html?gclid=CjwKEAiAmNW2BRDL4KqS3vmqgUESJABiiwDTvwsWzfZkLoX6t7XOD-v1Nnaqss2rwFrxm3R35utMWRoCcBbw_wcB)

[NeroMetalliko]

Andrea, thanks for that explanation. What was required to build your tool? Is it C++ or python or???

My own experience with the Canon ABW equivalent and the papers I use is that it all works fine out of the box, but I'm academically interested :-)

Hello,
if you give a look to my old post here
http://forum.luminous-landscape.com/index.php?topic=78142.0
and read down a little bit there is some short descriptions of the various tools used (even if some of the indicated things are now changed).

In short:
- I have coded a XML filter to import the CxF exported files (from X-Rite spectro) in LibreOffice Calc and I save then them as CSV. This was needed because Colorpicker for Windows does not export the data directly in CSV.
- The math work is done in Octave (free open source Matlab equivalent). I have coded some .m scripts to do all the task from command line and saving then the correction curves in TXT (and CSV) files.
- The real DeviceLink profile is done using a free open source tool called ICCXML: there is a ICCfromXML.exe command that allows me (using some batch files I have specifically done for the purpose) to build a DeviceLink profile starting from a right pre-formed XML text and the curves stored in my TXT data files previously generated.

The DeviceLink profile is applied to the picture in Photoshop (Edit -> Convert to profile (with Advanced setting engaged).

So, all what I have developed are some scripts/code for free open source stuff... it's not "academically" nice?
But it is not a straightforward one click process for sure...

Ciao,
Andrea

[NeroMetalliko]

Likewise.  I now have a pretty clear idea of what you're doing, except how you create the DeviceLink profile.  Is there any more detail info on this? 

Hello,

as answered above for the physical ICC build I use a command line based toolset called IccXML
you can find it on Source Forge
https://sourceforge.net/projects/iccxml/

More specifically I use the ICCfromXML command with some batch files that I have coded. I use an XML template of a DeviceLink profile and I insert the A tables with the numbers taken from my calculated data saved as TXT. Once you have all ready (which is not trivial) it works like a treat in one click!

With ABW from Photoshop, there is still the issue of how PS prints untagged or GG2.2 images when you select "Printer Manages Colors", which is made more complex by the fact that when PS does a silent profile conversion to such images is different on Win and OS X.  It's amazing how little understood this is.  I don't know how your DeviceLink profiles are treated, but it may be enough to "solve" the problem.

There were long debates on the net in the past regarding which profile is expected from ABW. I thought initially a gamma 2.2 (or AdobeRGB) was the proper one, but after some experiences/test I have collected more doubts and it could be that ABW expects something with a sRGB gamma curve.
In any case the good news for me is that it does not matter too much, because my linearization will fully compensate for everything in between. I still prefer to start from the best point possible and I usually use a gamma 2.2 profile form my work, choosing the best setting for the paper (often Neutral Dark in ABW).
The nice thing of this "universal" approach is that the printed results are what will drive the curve, so whatever happens in Photoshop/ABW, assuming is repeatable, it will be perfectly compensated.

Another nice thing is that the linearization works even for ICC + "Photoshop manage color" workflow obviously, with the possibility to neutralize the gray axis or to cast some pre-programmed split tones.

Recently, and unexpected, I have found that the linearization works very well even for color prints too: last week I have linearized a new Canson Baryta paper box starting from the Canson canned profile and found out that, in addition the L* linearization obtained, the overall color fidelity was improved (at least from the results of a personal test based on the 24 GretagMacbeth patches). But better to stop going OT here...

I really wish to have more time for all of this.

Thanks again,
Ciao.

Andrea