HW-calibrated display, color profiles and darktable workflow

Ah, great. Thanks. That’s good to know.

Hi,

If you have a Xrite compatible colorimeter you probably can use ‘displaycal’ on Linux to create a color profile, but I don’t think you really need it :wink:

What is the build-in correction sensor? Is it an built-in calibration device (colorimeter) and supports self-calibration? And does your monitor support hardware calibration, which means it stores the color mapping table in the monitor hardware?

If both can be answered with an ‘yes’ then you don’t have to boot to windows and create full blown ICC profiles.
I used to think the same, as I have a dell monitor with an built-in colorimeter. But I was wrong, at least partially (see below [1]). And ‘yes’ you can do this as you describe, but because your monitor probably stores the calibration data internally you only need an ICC profile that contains the gamut information and not the whole color lookup table.
Thus you can use just a standard AdobeRGB ICC profile, either shipped in the ‘icc-profiles-free’ package or the one linked by @Leniwiec. Just set it up in your Gnome/KDE/Xfce4 color management settings, use ‘system-default’ in darktable and set your monitor to AdobeRGB color-space. As long as all your sRGB jpgs do have embedded color profile information all images will look fine on your monitor.
In the meantime I checked this workflow on multiple machines and all my Images exported from darktable with embedded sRGB profile do always look fine (no color deviations).

If you use very specific, in monitor manually defined, color space the related profile can be generated using ‘displaycal-synthprofile’ from ‘displaycal’ package. Again the only information that is stored in the profile is the gamut: white point, red, green, blue, luminance and gamma. In my dell monitor there are two configuration slots for self defined color-spaces, thus I had to use this process to create matching ICC profiles, but in the end I don’t use those slots as AdobeRGB more than enough for me and provides better compatibility among different systems and software packages. You can find more information about it here: https://www.dell.com/community/Monitors/UP2720Q-ICC-profiles-for-Linux-or-Mac/m-p/8179908#M141056

On the other hand, if you use an ICC profile with color lookup table on a self calibrating monitor you end up with double correction of the color data, by the system first and then by the monitor later. This certainly isn’t what you want.

[1] The calibration verification data for self-calibrating monitors is only necessary to validate the deviation between color-space definition and the actual colors displayed by your monitor. If the deviation is negligible the gamut information in the ICC profile is all you need. OTOH if the deviation starts to grow, you’ll need to calibrate your monitor, with for example ‘displaycal’, and create a proper ICC profile with a lookup-table.

-Milosz

Edit: sorry the AdobeRGB profile was linked by @ggbutcher

1 Like

Perhaps creating the “full blown” ICC profile can be really convenient when you - like I do - don’t choose to limit your screen in any slightest way and pick “Gamut = native” during calibration :wink:

Hi Milosz,

Thanks for your help. The (used) monitor supports hardware calibration with setup using Eizo’s Color Navigator software (doesn’t support DisplayCal, I believe). The built in sensor cannot be used for a full calibration, I think, only for maintaining existing calibration settings. I guess Eizo “reserve” the former for their more expensive monitors.

“This sensor records your adjustment results, and maintains the white point and brightness values by automatically readjusting the settings at specific intervals that you determine. Even if the monitor is switched off or not connected to a computer, it will stick to its preset schedule and self correct.”

I will read up more on this topic as I don’t yet understand a lot of the fundamentals, but in the meantime I’d like, if possible, to correctly calibrate what I have now. I think I will use the Color Navigator software to try to calibrate the screen as that’s supported by the hardware.

I think in my initial question to Michal, I misunderstood that the colour mapping is stored in the monitor rather than on the PC.

I think the main issue for me then is that Colour Navigator won’t easily run on Ubuntu (in theory it will but I can’t see reference to anyone who has made it actually work). Therefore I would need to run Color Navigator in Windows (I dual boot Ubuntu and Windows 10) to calibrate the monitor, before switching to Ubuntu, where I do my Darktable editing.

I have one niggling concern (and here I’m probably going to make clear my black hole-sized level of ignorance) that in the switch from Windows to Ubuntu, the operating systems will themselves alter the color mapping. (I calibrated my old monitor using displaycal and generated a profile that I currently use in the Color settings in Ubuntu. I guess I can ensure that both Ubuntu and Windows are both using the same standard profile. (I’m not sure if any of this makes sense… but, hey ho).

Cheers!

Hi @Leniwiec,

indeed this is a tricky part and I don’t fully understand it.
My monitor has an option to use the ‘native’ color-space as well. All the gamut information for this color space is available through the EDID monitor information. Using ‘displaycal’ (file->create profile from EDID) the gamut info can be used to generate an intermediate ICC which later can be used to create an ICC profile for the ‘native’ color-space based on the gamut and correct gamma and white-point information (the gamma and white-point from EDID is off a factor in my case).
Doing it like this you’ll theoretically end up with an ICC without color lookup table (not ‘full-blown’). So I did it like this and all images seem to look fine as well.

But: the problem is now that my monitor doesn’t support neither calibration nor verification of the ‘native’ color space. So I have no clue what would really happen if you create a full-blown ICC profile for the ‘native’ color space.
May be it would be wise to use an external colorimeter HW to verify this color space only, just to get an impression how the deviation from all the EDID values looks like. But I have the feeling it wouldn’t be a good idea to create an ICC profile with color lookup table as long as the deviation is very small.

Hi @TonyBarrett,

Maybe there is a way to use it on Ubuntu as well. But you’ll probably try to contact Eizo directly to get the software for linux. It has been mentioned here: Red Hat applications on Ubuntu? - Ubuntu - It's FOSS Community but the link inside doesn’t seem to work.

-Milosz

Yes, I’ve seen that but I’ve also seen that others with far more expertise than myself have been unable to get it to work. I’m guessing it’s possible but unless there’s a step-by-step dumb guide somewhere, I think it’s probably beyond me for now.