[poppler] lcms usage in poppler

Olaf Drümmer olaflist at callassoftware.com
Sun Aug 4 04:19:18 PDT 2013

Going from the source colour space in the PDF to the destination space (in this case that would be the destination profile in the OutputIntent), the rendering intent must always be the rendering intent specified in the PDF page description.  The rendering intent can be set independently for each graphics object in a page description. If not set explicitly the rendering intent defaults to relative colorimetric. There are a number of ways to set the rendering intent: content stream operator, entry in ExtGState, entry in image XObject, entry in inline image object.

In the PDF graphics model, the rendering intent indicated inside an ICC profile **never** plays any role.

Also note that - in a second step - for converting from the destination colour space to some actual output device, for example a computer monitor to be able to display/simulate the colour appearance as it would occur on the output device described by the OutputIntent, another colour transform will be necessary, and one would usually use relative colorimetric for that, unless the colour appearance of the originally intended substrate (paper colour of the colour space associated with the destination profile/OutputIntent) is to be simulated, in which case one would use absolute colorimetric.

Please make sure to always honour the rendering intent encoded inside the page content of PDF files, and never pay any attention to the rendering intent encoded inside ICC profiles.


Am 4 Aug 2013 um 12:34 schrieb Thomas Freitag <thomas.freitag.bbr at gmail.com>:

> Am 02.08.2013 19:50, schrieb Albert Astals Cid:
>> El Divendres, 2 d'agost de 2013, a les 17:05:38, Thomas Freitag va escriure:
>>> Hi Folks!
>>> Can anybody tell me how I can really test the lcms integration in
>>> poppler? I thought it should work, but now I'm no more sure. Reason for
>>> the change of my mind is that I tried to implement the usage of
>>> OutputIntents.
>> Is your implementation similar to the one i did 2 years ago?
>> https://bugs.freedesktop.org/show_bug.cgi?id=34053  
> Thanks for the hint, Albert. In the meantime I already encountered, that creating the transformation with INTENT_PERCEPTUAL instead of INTENT_RELATIVE_COLORIMETRIC works fine with the GWG tests (This time I attach only one result, but the two other are also okay). Now I'd read the bug comments, but I did't find any intent entries in the PDF. BUT: as far as I understand the PDF spec, the settings in output profile should be used if there are no intent entries, and indeed, the profile pointed by the OutputIntents dictionary says:
> icc:RenderingIntent               : Perceptual
> Should I try fix it in that way and and attach a patch to bug 34053?
> Cheers,
> Thomas
>> I could not find either why it was not working.
>> Maybe you can mix your idea and my idea and get something that works?
>> Cheers,
>>  Albert
>>> Motivation is the GWG, patch series 7 (07_OutputIntent),
>>> but the result of my first tests is not really satisfying, s.
>>> attachments. Of course it's a little bit better than without my
>>> implementation, see i.e. the shading cross. But it's not like it should
>>> be. So I'm wondering now if the reason for it
>>> 1. is my wrong implementation (but in the debugger I saw that the lcms
>>> transformation is called) or
>>> 2. is the integration of lcms in poppler, at least in the CMYK area (I
>>> need to test it there, because the profiles in the PDF are CMYK profiles) or
>>> 3. is lcms with this kind of profiles (the profiles are REALLY small for
>>> CMYK profiles, normally I see CMYK profiles with a size of at least 512
>>> kByte).
>>> Cheers,
>>> Thomas
>> _______________________________________________
>> poppler mailing list
>> poppler at lists.freedesktop.org
>> http://lists.freedesktop.org/mailman/listinfo/poppler
> <iso71.jpg>_______________________________________________
> poppler mailing list
> poppler at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/poppler

More information about the poppler mailing list