display calibration and profiling

[Frédéric Bron]

I have just been able to get a colorimeter (i1 display). I have used
it with argyllcms to calibrate (can only change the brightness) and
profile my laptop display.

I first tried the one step option with dispcal:

$ dispcal -v 3 -m -yl -qh -g2.2 -Y p -o profile1.icc profile

The profile I obtained changed dramatically my display colors which
were much too blue! It has improved a lot my photographs. I am already
very happy. Thanks a lot to Graeme for this wonderful tool.

I then tried to go further using
$ targen -v 3 -d3 -f2000 -cprofile1.icc profile
$ dispread -v -yl -k profile.cal -Y p profile

I then get a profile.ti3 files.

I can then create a matrix profile with:
$ colprof -v -q u -qm -as -O profile2.icc profile

My questions:

1. for profile1.icc, I run dispcal 3 times and chose the one that has
the smallest worst case at the end of the log "Failed to meet target
0.400000 delta E, got worst case 0.612002". Is that useful? Sometimes,
the worst case varies from .6 to 2, I choose .6

2. is the matrix profile created with colprof (profile2.icc) supposed
to be better than the one created by dispcal (profile1.icc)? With
colprof, I cannot find any comparable output as for dispcal like "got
wors case ...". Instead I get "peak err = 10.749372, avg err =
1.225101, RMS = 1.723312". Is that delta E error?

3. I would like to try the other option that creates a LUT based
profile to see if I can see any difference with the matrix profiles
but I am lost in all the options. In particular, an input profile must
be provided: "For a LUT based profile, where gamut mapping is desired,
then a source profile will need to be provided to define the source
gamut." What does that mean? Should I provide the Adobe RGB profile as
I want to view my photographs that have been coded in Adobe RGB?

Thanks in advance,

Frédéric

[Nikolay Pokhilchenko]

Post by Frédéric Bron
I then tried to go further using
$ targen -v 3 -d3 -f2000 -cprofile1.icc profile
$ dispread -v -yl -k profile.cal -Y p profile
I then get a profile.ti3 files.
$ colprof -v -q u -qm -as -O profile2.icc profile
1. for profile1.icc, I run dispcal 3 times and chose the one that has
the smallest worst case at the end of the log "Failed to meet target
0.400000 delta E, got worst case 0.612002". Is that useful? Sometimes,
the worst case varies from .6 to 2, I choose .6

This is not bad. It means that while calibrating certain gray level the printcal was unable to look-up the RGB values which are close enough to the desired gray. Sometimes this is because of display and instrument instability and noise. Sometimes because of quantization errors on display interface and inside the display. Your result is OK and highly acceptable.

Post by Frédéric Bron
2. is the matrix profile created with colprof (profile2.icc) supposed
to be better than the one created by dispcal (profile1.icc)? It seems that both should be equal.
With
colprof, I cannot find any comparable output as for dispcal like "got
wors case ...". Instead I get "peak err = 10.749372, avg err =
1.225101, RMS = 1.723312". Is that delta E error? The dispcal profile have the errors of the same order. This is because of nature of color correction inside the display, The displays with digital color correction inside are not truly additional devices. So they can't be  profiled by matrix profile flawlessly. Have You tried "native" mode of color correction in display settings? In some cases (rather often) the native mode (without a correction) resulting much better peak error with the matrix profile.
3. I would like to try the other option that creates a LUT based
profile to see if I can see any difference with the matrix profiles
but I am lost in all the options. Just try colprof profile. That's all.
In particular, an input profile must
be provided: "For a LUT based profile, where gamut mapping is desired,
then a source profile will need to be provided to define the source
gamut." What does that mean? Should I provide the Adobe RGB profile as
I want to view my photographs that have been coded in Adobe RGB? If Your images are mostly inside of the display gamut I don't recommend You to involve gamut mapping into display profile. The display profile with relative LUT only will be optimal. If You wouldn't provide source color space or image gamut You'll get good "classic" LUT display profile without mapping.

The mapping from profile gamut (not from image gamut) have desaturation of colors in directions where display gamut was smaller than source profile gamut. The desaturation in such case will persist even if the image is in gamut of the display.

[Graeme Gill]

Frédéric Bron wrote:

Hi,
sorry this reply is somewhat delayed.

Post by Frédéric Bron
1. for profile1.icc, I run dispcal 3 times and chose the one that has
the smallest worst case at the end of the log "Failed to meet target
0.400000 delta E, got worst case 0.612002". Is that useful? Sometimes,
the worst case varies from .6 to 2, I choose .6

Probably not a lot. Typically this sort of variation means that
dispcal is fighting quantization. At the end of the day the
profile takes the calibrated displays response into account.

Post by Frédéric Bron
2. is the matrix profile created with colprof (profile2.icc) supposed
to be better than the one created by dispcal (profile1.icc)? With
colprof, I cannot find any comparable output as for dispcal like "got
wors case ...". Instead I get "peak err = 10.749372, avg err =
1.225101, RMS = 1.723312". Is that delta E error?

They are the comparable output. Since they do different tasks in a different
way, they report different things. Yesm the colprof output is
profile fit delta E.

Post by Frédéric Bron
3. I would like to try the other option that creates a LUT based
profile to see if I can see any difference with the matrix profiles
but I am lost in all the options. In particular, an input profile must
be provided: "For a LUT based profile, where gamut mapping is desired,
then a source profile will need to be provided to define the source
gamut." What does that mean? Should I provide the Adobe RGB profile as
I want to view my photographs that have been coded in Adobe RGB?

It's not to hard to try this. Simply add the "-la", "-lx" or "-lX" options,
plus the source profile/gamut ie. "-S sRGB.icm".

Graeme Gill.