Jeremias Maerki a écrit :
There's no "treatment" or the image itself, just evaluation of
information inside the image being loaded. FOP reacts on the and tries
to put the right information inside the PDF.
This was not clear in our mind since I had a vague feeling I had seen
things like gamma correction while browsing the code very quickly, but
this may be somewhere else in a totally different context ?
This hint helps us a lot.
The key change is in org.apache.fop.render.pdf.ImageRenderedAdapter (in
the case of PNG images) where you'd need to override the ColorSpace
sent to the PDF library with the device-specific variant.
This is second very, very helpful hint. I've been once more through the
code of this class; not sure at the moment I grasp everything about the
way it works or how it fits in FOP general picture, but at least now I
know what to look for and where we should put the effort...
Please note: I consider this a hack or a desperate measure if all else
fails. Ignoring the color management is a way to fall back to
device-specific colors which is known to produce the color fidelity you
expect on screen. I'd rather find out if there's a bug in the way FOP
handles color profiles. But if I see so many other programs fail at
producing the colors you expect I have low hopes. Maybe your PDF gives
me a clue.
Well, I guess the customer is desperate to a point where a hack will do
the job until in the long term someone (you / us / someone else) comes
up with a clean solution.
So that is not really a problem.
