Hi Graeme
Post by Graeme GillThe profile should be the one that corresponds to the same relationship between
device values and colors as the measurements you are going to make.
You you are measuring the printer in its calibrated state, so the profile
used to inform target should be the calibrated device values,
so the ICC profile without the calibration curves applied to it.
Ok, that makes sense. It's around the other way to what one might
naively think, but I see the logic.
Post by Graeme GillI initially tried targen -c xx.icc (the one without the calibration applied) but got the
"Failed to re-seed the veronoi after 100 tries" until I reduced adaptation to about 0.8.
Then I tried targen -c xx.icm (the one with the calibration applied) and it appeared to
work at adaptation 1.0
The former should is the correct one. It's a bit of a puzzle as to why it should be less
well behaved than the raw (uncalibrated device space) profile.
You could try reducing the resolution of the pre-conditioning profile (ie. -qm or -ql)
or increasing the -r parameter, but reducing the adaptation level is also fine.
targen completed ok (and very quickly) using -r, but I went ahead with
-A 0.83. Then "printtarg -K" and chartread. However every row gives a
warning "There is at least one patch with an very unexpected response!
(DeltaE 49.046876)" so I've got something wrong.
Could I ask you to cast an eye over the complete workflow below? I have
to say that using argyllcms have given me far better results than any
other product, so thankyou! And the profile from the "Initial Profile"
stage below is pretty good. There was a tremendous improvement after I
changed to doing the calibration separately and then combining them.
It's just the refining that I can't get right.
This is a very straightforward situation, with a modest inkjet printer
that doesn't itself support calibration. Despite hunting around I
couldn't find any complete end-to-end workflow for calibrate/profile
then subsequent refines, so I put this together. I'm using 1.6.2,
ColorMunki Photo, and this particular printer is a Canon iP4700. (I
reverted from 1.6.3 because of the CM re-calibration problem when
profiling displays).
Initial Calibration
----------------------
P=iP4700; D=300
printtarg -v -iCM -h -b -T$D -pA4 ${P}_c
**** print using Lightroom (printer-managed; print driver no correction)
chartread -v -H ${P}_c
printcal -v -i ${P}_c
Initial Profile (4 x A4)
----------------
P=iP4700; D=300
targen -v -d2 -G -f840 ${P}_p
printtarg -v -iCM -h -b -T$D -pA4 -K ${P}_c.cal ${P}_p
**** print using Lightroom (printer-managed; print driver no correction)
chartread -v -H ${P}_p
colprof -v -S ../AdobeRGB1998.icc -cmt -dpp -D${P} ${P}_p
applycal ${P}_c.cal ${P}_p.icc ${P}.icm
**** Use ${P}.icm in Lightroom for printing (print driver no correction;
good results)
Refine Profile (1 x A4)
-----------------
P=iP4700; D=300
targen -v -d2 -G -f210 -A 0.83 -c ${P}_p.icc ${P}_p
printtarg -v -iCM -h -b -T$D -pA4 -K ${P}_c.cal ${P}_p
**** print using Lightroom (printer-managed; print driver no correction)
chartread -v -H ${P}_p **** DeltaE > 49 on every row ****
colprof -v -S ../AdobeRGB1998.icc -cmt -dpp -D${P} ${P}_p
applycal ${P}_c.cal ${P}_p.icc ${P}.icm
**** Use ${P}.icm in Lightroom for printing (print driver no correction)
Any comments on the overall workflow would be very welcome, as well as
clues as to why the chartread in the refine is going so wrong.
I do feel that including some complete workflows like this (for the
simplest cases) into the documentation would be of great help. There is
lots of detail, but it's sometimes hard to put it together right. I
appreciate that there are myriads of possibilities, but profiling a
simple inkjet (with the benefits of the separate calibration phase) must
be one of the commonest....?
Many thanks for your help, and for all your work on argyllcms.
Regards
Chris