- This topic has 16 replies, 2 voices, and was last updated 7 years, 3 months ago by
.
-
Posts
-
Greetings!
I’m relatively new to color management and still learning. I got myself a Spyder 5 Express to calibrate and profile my Acer P221W screen on my Arch Linux system.
I let DispCAL download the suggested correction for my screen, then proceeded to calibration and profiling. The calibration phase was a success. I used the hidden factory menu of my screen to adjust gains until the bars matched up pretty well.
However, after profiling was done and I loaded the resulting profile my colors turned all yellow/amber. I’m not talking about a slight discoloration, but completely off, like I was looking through a glass of tea. I tried profiling several more times to no avail. First I loaded it as a system profile in KDE, then I tried to instead use it only in Krita (color managed painting program), but the result was the same strong amber tint.
In the end I removed the profiles and hoped the calibration was good enough – it apparently was, as my artworks looked on the print shop’s Eizo screen and on the prints themselves exactly as I remembered them from my home screen.
Still, it bothers me why the profiles gave me such bad result, and I want to fix it. Also, if correct profiling has any chance to improve my experience on top of calibration, then I would like to exploit that.
So any ideas?
Note: The issue may be in KDE 5 itself, it’s still in development and has some issues. I can’t tell as I don’t have access to another system to test the screen and profile with.
Hi,
I let DispCAL download the suggested correction for my screen
It is important to understand that what you downloaded is not a suggestion – a query via the online database just looks for matches between your instrument and display model name, and presents all matching corrections. All the corrections in the online database are provided by various users, and whether they improve the result with your particular instrument and display is always up to you to evaluate.
A correction created for one specific instance of instrument and display combination may not work well in another instance of the same combination due to manufacturing fluctuations, for example.
Your best bet in such a case is to use one of the vendor-provided measurement modes (=generic corrections) instead (probably “White LED” in your case).However, after profiling was done and I loaded the resulting profile my colors turned all yellow/amber.
What settings did you use for calibration? Note that e.g. a whitepoint with a correlated color temperature of 5000K is considerably warmer than what most displays look like when they come from the factory.
In the end I removed the profiles and hoped the calibration was good enough
The profile contains the calibration so removing the profile gets rid of the calibration as well.
It is important to understand that what you downloaded is not a suggestion
Oh… I was under the impression that I can’t get proper reading without selecting one. The automatic download downloaded for me this:
Spectral: LCD CCFL PVA (NEC MultiSync P221W Native (GretagMacbeth i1 Pro))I have no idea what all that means, apart from “P221W” which is my screen’s model, and that I was under the impression that i1 devices and Spyder 4/5 are compatible.
I tried importing from the Spyder 5 driver, but no new entry has been added to the Corrections dropdown.
I’ll try calibrating again using “auto” when I have the time… it takes more than an hour.
Your best bet in such a case is to use one of the vendor-provided measurement modes (=generic corrections) instead (probably “White LED” in your case).
You mean the “Mode” dropdown? There is no “White LED” option, there is “Generic LCD” and it was on that already. But how does this dropdown relate to the Corrections one? I’m a bit confused.
What settings did you use for calibration?
Which exact settings you are asking? I left the vast majority on default since I don’t know much about them. I used the quick start guide to do the basic guided calibration and profiling.
Note that e.g. a whitepoint with a correlated color temperature of 5000K is considerably warmer than what most displays look like when they come from the factory.
I tried to set my whitepoint around 6500, since my work is primarily seen in digital, I rarely print. The initial measurement reports it as about 6600.
The profile contains the calibration so removing the profile gets rid of the calibration as well.
No. I hardware-calibrated using the adjustments in my screen’s service menu.
Spectral: LCD CCFL PVA (NEC MultiSync P221W Native (GretagMacbeth i1 Pro))
This correction is not suitable at all (look at the manufacturer).
You mean the “Mode” dropdown? There is no “White LED” option
Menu “Tools” -> “Correction” -> “Import colorimeter corrections from other display profiling software”. This runs automatically when the instrument is first detected. Did you cancel it?
I left the vast majority on default since I don’t know much about them.
That’s fine.
No. I hardware-calibrated using the adjustments in my screen’s service menu.
Calibration consists of two parts, adjusting the monitor’s brightness and whitepoint is only one of them. The other part are the calibration curves that are embedded in the profile, and that get loaded into the graphics card.
Sorry for the slow reaction and thank you for your reply!
This correction is not suitable at all (look at the manufacturer).
Alright I’ll repeat it with no correction one I have the time.
Menu “Tools” -> “Correction” -> “Import colorimeter corrections from other display profiling software”. This runs automatically when the instrument is first detected. Did you cancel it?
No I didn’t. I imported from the Spyder of software downloaded from Datacolor. This created “spyd4cal.bin” in ~/.local/shareArgyllCMS but that’s all it did. There was no correction added to the Correctin dropdown. Now I deleted the GretagMachbet one that you said is not suitable, now the only option left in the dropdown are “none” and “Auto (none)”
I’m going to run a new profiling with Correction set to “none” and see if the resulting profile works. Not sure when I can do that though since it takes hours and I can’t use the computer in the meanwhile. Thanks for your help so far!
I did several calibrations today. Tried various settings, 5000K and 6500K, 2.2 gamma and 2.58 which is the value reported by “report on uncalibrated device”. I also dual-boot Windows now so I tried creating a profile with the proprietary Spyder 5 software under Win 10, then I copied it back under Linux. The results are better than previously using the wrong correction, but there are still issues:
– After setting white point to target 6500 greys feel blusih and cool – however, all profiles, including the one made with Datacolor software feel orange-brown and too warm. Profiles with 6500K target should be cool but they feel definitely warm, and much warmer than the uncalibrated screen after setting white point via the screen’s service menu.
– The profiles created with DispCal will make the screen much darker than the Datacolor profile, regardless of gamma setting. Often this feels too dark.I don’t know if my eyes are cheating me because I was looking at an uncalibrated screen for way too long, or my Spyder is actually faulty? Is there a way to find out? I don’t know anyone who has a colorimeter so I can’t borrow another one to compare.
After setting white point to target 6500 greys feel blusih and cool – however, all profiles, including the one made with Datacolor software feel orange-brown and too warm.
That sentence doesn’t make sense. The result can be either cool or warm, but both at the same time isn’t possible.
The profiles created with DispCal will make the screen much darker than the Datacolor profile, regardless of gamma setting. Often this feels too dark.
Are you using ambient light level adjustment (you shouldn’t)?
After setting white point to target 6500 greys feel blusih and cool – however, all profiles, including the one made with Datacolor software feel orange-brown and too warm.
That sentence doesn’t make sense. The result can be either cool or warm, but both at the same time isn’t possible.
I didn’t say in the same time…
It feels too cool (grays are blue) after adjusting my screen to match the white point target at the beginning at “interactive display adjustment”. This is _before_ any profile is applied. _After_ profiles are applied it turns from too cool to too warm (grays are orange).
Are you using ambient light level adjustment (you shouldn’t)?
No, it’s unchecked.
The profiles created with DispCal will make the screen much darker than the Datacolor profile, regardless of gamma setting. Often this feels too dark.
In your above screenshot though, you’re using a gamma of 2.58. That is very contrasty and possible even too much for a pitch dark room. Use the default 2.2 instead.
It feels too cool (grays are blue) after adjusting my screen to match the white point target at the beginning at “interactive display adjustment”. This is _before_ any profile is applied. _After_ profiles are applied it turns from too cool to too warm (grays are orange).
That doesn’t make sense either. After adjusting the display to meet the target whitepoint, there’s nothing left for the calibration to adjust, and so the profile would have exactly the same whitepoint as established during calibration. Attach the profile + logs please (use the “Create compressed archive” button next to “Settings” with the profile selected in DisplayCAL).
In your above screenshot though, you’re using a gamma of 2.58. That is very contrasty and possible even too much for a pitch dark room. Use the default 2.2 instead.
2.58 was given to me by “Report on uncalibrated device”. I tried both that and 2.2, with no noticable difference.
After adjusting the display to meet the target whitepoint, there’s nothing left for the calibration to adjust, and so the profile would have exactly the same whitepoint as established during calibration.
This is exactly why I suspect my Spyder might be faulty, because even the profile created with Datacolor’s own software makes colors much warmer when the profile is loaded.
Attach the profile + logs please (use the “Create compressed archive” button next to “Settings” with the profile selected in DisplayCAL).
Done. The second one is the profile created with Datacolor’s software under dual-booted Win 10.
The profiles seem fine and there’s no indication that anything might be faulty. I would still re-calibrate with gamma 2.2 and disable “Auto” black point correction so you get the native contrast ratio.
I don’t know if my eyes are cheating me because I was looking at an uncalibrated screen for way too long
I guess that’s it. In uncorrected state, your monitor has a strong blue cast in the upper end of the grayscale below white, which the calibration of course corrects for, so the result looks warmer what you’re used to.
I would still re-calibrate with gamma 2.2 and disable “Auto” black point correction so you get the native contrast ratio.
I’ve tried many combinations of settings. I’ll try again one more time, but I’m pretty sure I did that combination already.
In uncorrected state, your monitor has a strong blue cast in the upper end of the grayscale below white, which the calibration of course corrects for, so the result looks warmer what you’re used to.
But you said the opposite previously:
That doesn’t make sense either. After adjusting the display to meet the target whitepoint, there’s nothing left for the calibration to adjust, and so the profile would have exactly the same whitepoint as established during calibration.
At first you say the profile shouldn’t change the temperature, then you say it does. Now what? :s
I still don’t think calibrating to 6500K should result in everything being super warm like there is an orange tint. I made a crappy video made with my phone. Note: The phone’s white is too cold, so what appears in the video to be a change from very cold to slightly warm is in reality a change from slightly cold to very warm. But it should still give an idea of how wild the change is.
https://www.dropbox.com/s/61c6zexvxxlfblg/VID_20170101_004729.mp4?dl=0
But you said the opposite previously:
Not really. All I said is the calibration will make all of the grayscale the same hue as the measured whitepoint if it’s already on target, and that is exactly what it did in your case.
I still don’t think calibrating to 6500K should result in everything being super warm like there is an orange tint. I made a crappy video made with my phone.
Videos or photos are not suitable means to convey this information because obviously they cannot show what you see (white balance and image processing of the camera as well as the display one is viewing the video on influence what one sees), but from what little I can extract from the video I can assume with some confidence that the calibration seems to be working as intended, i.e. it gets rid of the very strong blue cast of the grays and makes the whole grayscale neutral with respect to the whitepoint.
But it should still give an idea of how wild the change is.
Well, your display is old and probably has acquired, through aging, a very strong blue cast in the upper grayscale that is also apparent from looking at the calibration curves. The calibration of course corrects for this.
Sorry for the slow reply, been busy.
the calibration will make all of the grayscale the same hue as the measured whitepoint if it’s already on target
OK, hmmm… let me ask you a few things because I feel my understanding here is incorrect or at least incomplete.
- Am I right to conclude that the sole point of setting the white point is to give calibration something to work with, and it’s OK if the resulting screen settings will make colors look off, as the calibration will fix that? So setting the white point according to the measurement is not supposed to make colors look better on its own?
- How important is adjusting black point via screen settings? I did that based on the info shown in “check settings” part of the initial display adjustment, changing base settings in the screen’s hidden service menu until black point measure was close to target. But should I really bother with this? I experience soma banding in the very low grays and I have a feeling it might be because of this. I made a note of the original values in case I need to restore them.
the calibration seems to be working as intended, i.e. it gets rid of the very strong blue cast
[…]
your display is old and probably has acquired, through aging, a very strong blue cast in the upper grayscaleI seem to see that orange tint primarily in the dark grays instead of the light grays. Or that’s just my eyes misleading me?
of the grays and makes the whole grayscale neutral with respect to the whitepoint.
So if everything feels to warm I should try calibrating towards an even cooler target? tried to just use the calibration for a while to give my eyes time to adjust but it feels so off to me that I couldn’t stand it for too long. I changed my desktop color scheme a week ago to use only shades of gray with no any color or tint, hopefully that helps my eyes.
My desk is located in a room with white walls almost next to a window, plus I had a desk light (LED “bulb”, no idea of color temp) because my eyes start to hurt if I don’t. This is farm from ideal (About to get some black carton and craft a hood for my screen to alleviate somewhat) and I can’t afford to set up consistent lighting.
I’m going to try a few more things when I have the time, such as calibrating for an even cooler target, and re-adjusting screen gains.
Am I right to conclude that the sole point of setting the white point is to give calibration something to work with, and it’s OK if the resulting screen settings will make colors look off, as the calibration will fix that?
Adjusting the white point to a given target is not a necessity, but if you (e,.g.) want to match several displays to one another, there’s usually no way around it.
So setting the white point according to the measurement is not supposed to make colors look better on its own?
Only the white itself will be affected by white point adjustment. Grayscale will only be correct after calibration. And colors will only be correct in color managed applications after profiling.
How important is adjusting black point via screen settings? I did that based on the info shown in “check settings” part of the initial display adjustment, changing base settings in the screen’s hidden service menu until black point measure was close to target. But should I really bother with this?
Probably not. The usual approach is to leave the black point at its native setting.
I seem to see that orange tint primarily in the dark grays instead of the light grays. Or that’s just my eyes misleading me?
Unlikely, as the human visual system is quite sensitive to hue changes in neutrals, although this is less so the darker the stimulus gets (and we perceive “neutral” in relation to the white point our eyes are adapted to).
So if everything feels to warm I should try calibrating towards an even cooler target?
Yes. Many “consumer” instruments are not particularly accurate when it comes to assess the white point, that’s why the default in DisplayCAL is to leave the whitepoint “As measured” (not sure if you have tried that already).
My desk is located in a room with white walls almost next to a window, plus I had a desk light (LED “bulb”, no idea of color temp) because my eyes start to hurt if I don’t.
Maybe the monitor is either too dim or too bright for the environment? That is often the primary cause for eye strain.
and I can’t afford to set up consistent lighting.
You probably don’t have to spend that much. Closing the blinds and a single high quality bulb for your desktop lamp (CRI >= 90) with a CCT rated anywhere from 5000-6500K (“daylight”) should already help.
