I have trouble understanding whether I have some bad configuration or misunderstanding with monitor profiles under Linux.
Here’s my setup: I am calibrating and profiling my laptop monitor with DisplayGUI and a Spyder2 for some years now under Arch Linux. The profiles are installed in the OS and loaded into the video card gamma table. It works out of the box, but when selecting the monitor profile in RawTherapee or digiKam additionally, the colours look over saturated like if the colour correction is applied twice. Without any monitor profile or a generic profile like RTv4_sRGB in RT the colours look natural.
In contrast, if I reset the vcgt and use the monitor profile then, the colours also look unnatural. What do I need to do to see the “real” colours in RT or digikam with my monitor profile selected. Is my feeling correct, that the colour correction is applied “twice”? Is this a misunderstanding of the colour management as such or something I may reconfigure?
I hope I understood your question correctly: When selecting the ICC profile which was created by DisplayCAL during profiling previously in the pull down menu in the lower part of RT or by selecting the system profile (which is the same in this case).
@betazoid: What makes you think that it’s broken? The created profile looks good when loaded in the vcgt. Much better than without (my laptop monitor has a quite low colour fidelity).
well Spyder2’s are quite old already and the filters in the device age.
Are you sure that you understand color management correctly? Especially the difference between profiling and calibrating? Calibration does not affect the saturation of the colors whereas profiling does. In the case of laptop displays, you usually need both calibration and profiling since there is no way to hardware calibrate laptop screens. But you might try profiling only as well.
How large is the color space of your laptop screen - at least accoring to Displaycal/Spyder2? Is the color space of the created profile siginficantly smaller than sRGB? Than color management indeed does add saturation to the colors. Well, there might be several reasons for this behavior.
One more thing: color managment only works with apps that can do color management. You do not see the effect of the color profile outside of color managed apps. You only see the effect of calibration.
Calibration/Profiling is not only about “fixing” screens that have a color cast.
Thanks for the quick answer. As written in my first post, I may have a misunderstanding here, but I think I got the difference between calibration and profiling. The calibration is done when the the calibration curves are loaded into the video card gamma table, right? On top of that there’s the ICC profile which is read by the colour managed program like RawTherapee. What I probably did not understand properly is the combination of calibration and profiling. What do I see when I open an image in RT on a calibrated monitor without a loaded profile and what with the profile loaded?
I don’t know the colour space of my laptop screen. But I assume I can measure it with DisplayCAL.
In the flatpak, it is not possible to compare a screen profile with sRGB (screeshot - comparison profile is p3 which is useless), but on Arch it should be possible. If your screen profile is significantly larger than sRGB, color management will decrease saturation, if it is significantly smaller, it will increase saturation.
Yes it is. No wonder the colors look oversaturated. Don’t use that screen for editing.
Go Get a 100-euro/dollar external screen. There is also the possibility to transform the screen profile so the clors look better but that screen is basically… well not suitable for editing photos. What kind of laptop do you use? If it was a budget laptop (under appr. 1000 eur) the screen is probably very bad.
I would have to look up how to do that perceptual transformation, maybe try to google the thread in the official Displaycal support forum. It was me - username probably betazoid - who asked a very similar question.
Maybe asking a stupid question: isn’t calibrating what you do with your display without any measuring device (Spyder2), and profiling what you do with your measuring device to refine what you have already got with calibration?
According to this explanation (and many others in other pages) most of the time calibration means adjusting brightness (not true for high end displays), and profiling means «doing our best to get the most accurate colors we can from our display».
Anyway, I think the profile shown in the screenshot is rather old (from 2014-12-16), and also think you’re right politely saying that display is «not good» for editing photos…
Well calibration curves can alsofix or induce color casts. This is visibile in apps that don’t do color management, too. In my experience, especially laptops usually need a “software calibration”, too, because the only thing one can tweaked on the dispay hardware is the brightness/backlight.
The laptop was not cheap, but it is vintage in computer years (> 10 years old). It’s a ThinkPad T410 which was around 1500 € at that time. I do have an external screen, but I don’t use it so often because of laziness. Looks like I need to change my mind. Here’s an old curve of the external screen:
@XavAL: The profile of my laptop is dated 2019-10-06, not 2014. But I have curves from 2014. In the 5 years in between, there was a slight deterioration.
I’ve never done that (to my knowledge), so I missed that information. Thanks.
Let me disagree: I’m also a dispcalGUI user, and on my profiles the field «Created» (7th from the top) shows the date when the ICC profile was created by the program. So in both your ICC profiles screenshots, that should be 2014.
By the way, either Argyll and dispcalGUI have much current versions than the ones used to make those profiles. Maybe you can get a bit better results now (I wouldn’t expect much better results, though).
And just to add some more info, the last 2 lines of that profile information tells you the gamut your display can render, with respect to sRGB and Adobe-RGB.
I’m pretty sure that I made the profile in 2019. Furthermore, the profile is called “ThinkPad 2019-10-06 2.2 S 3xCurve+MT” and the ICC files have a file date from 2019. Could it be that the date is still 2014 because I just updated the profile in DisplayGUI?
Whatever, I see I must create a profile from scratch again with the latest software versions.
If you can afford it, it’s definitely a good idea. An i1Display Pro would be a good choice.
But prior to that, I would recheck with a new profile which could be compared with the older one. I see two differences that could make a big difference (in my experience):
the Luminance in the new profile is lower than in the older profile (151.84 vs 172.08): depending on the display hardware and how aging has affected it, a higher display brightness could lead to a wider gamut
you need to create the same type of profile: XYZLUT+MTX vs 3xCurve+MTX (in my display the difference is not that big, but there’s a difference)
Uncheck the LUT option and it should render a look closer to the older profile (not the same gamut, though). And just for fun, change «Device > A2B > PCS» to «PCS < B2A < Device»…
The rendering intent should not make any difference here, in my opinion. Even though, while looking at the profile information, change «Absolute colorimetric» to «Relative colorimetric», so you can have a better impression of your display gamut in real life.
Thanks for your recommendations, I’ll try them later today.
I read that these device seems to have problems with the long term stability of the rubber because the softeners deteriorate. I hate this. Any experience with this?
Furthermore, there’s also the i1Studio, which is a spectrometer an thus more expensive. Does it make sense to invest this money?
@betazoid: I did check the difference with the external monitor. The difference between loaded vcgt plus ICC profile and loaded vcgt only is much less than with my ThinkPad monitor. However, there’s still a slight difference. It’s a bit lighter in dark areas with the ICC profile loaded. Is this because of the missing colour space on the upper side of the curve?
I dont think so. Probably you just created a profile where black point compensation is active. You find the option under profiling - profile type. If you select e.g. curves+matrix, bpc is automatically activeted. Also, with xyz profiles, bpc is baked in, afaik, so there is no need to activate this option. But it will still do something with the blacks/darks, probably brighten them.
The missing part of the color space just means that you cannot see some saturated yellows and greens on the external screen. It has nothing to do with the blacks.
Mine is around 8 years old and looks and feels like new. Can’t say this will always be true.
About the i1Studio, it is just the new Colormunki version. I don’t know if they upgraded anything in the hardware side (I read they didn’t when it was released, but newer models may perform better).
If you need printer profiling, then it is a must (or some model similar to this one). If you’ll only calibrate/profile your display, there are comparisons saying that the 1iDisplay Pro gives better icc profiles, but there is not much difference between them.
It’s a ThinkPad T410 which was around 1500 € at that time. I do have an external screen, but I don’t use it so often because of laziness. Looks like I need to change my mind. Here’s an old curve of the external screen:
