- This topic has 61 replies, 6 voices, and was last updated 1 year, 1 month ago by
.
-
Posts
-
I have an Acer XV282k, which already looks pretty good out of the box, but I would like to verify that it’s actually correct, and also make sure my monitors at work and home match as closely as possible..
Running a calibration with Default settings makes everything too bright and washed out in Color Managed applications. Using “Monitor RGB” in Photoshop will make things look normal again, which I guess means the issue is in the ICC profile.
Setting the calibration Tone Curve to sRGB instead of Gamma 2.2 makes the end result look somewhat closer to “Monitor RGB”/the monitor default, just a bit less saturated and still a little washed out. Is this the correct setting?
Both settings will make the
http://www.lagom.nl/lcd-test/black.php#blacktest.png test image (imported into Photoshop) look suspiciously washed out. Without calibration, I can barely make out the 1 and 2 squares, but after calibration they are much brighter than it seems like they should be.I’ve spent a dozen hours reading on this forum and elsewhere, but I can’t find a straightforward answer regarding which settings to use. How can I make sure this calibration is actually correct?
My settings:
Gamma: 2.2 / sRGB
White point: 110 cd/m2
Whitepoint: 6500K
Everything else default
Acer XV282k
Same issue with both iDisplay Pro and Spyder X
Same issue on Windows 11 and 10
SpyderX Pro on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.I have an Acer XV282k, which already looks pretty good out of the box, but I would like to verify that it’s actually correct, and also make sure my monitors at work and home match as closely as possible..
Running a calibration with Default settings makes everything too bright and washed out in Color Managed applications. Using “Monitor RGB” in Photoshop will make things look normal again, which I guess means the issue is in the ICC profile.
No. It means that you are not familiar about how things should look.
Monitor RGB = see content con color managed (wrong)
Setting the calibration Tone Curve to sRGB instead of Gamma 2.2 makes the end result look somewhat closer to “Monitor RGB”/the monitor default, just a bit less saturated and still a little washed out. Is this the correct setting?
No. Choice of calibration target gamma/TRC has no effect in colorspace extension (saturation). It only corrects grey.
Both settings will make the
http://www.lagom.nl/lcd-test/black.php#blacktest.png test image (imported into Photoshop) look suspiciously washed out. Without calibration, I can barely make out the 1 and 2 squares, but after calibration they are much brighter than it seems like they should be.No. sRGB as a brighter TRC near black than 2.2. Imported to photoshop and assigned to sRGB profile that PNG should render very dark greys brighter… because sRGB TRC behaves that way.
I’ve spent a dozen hours reading on this forum and elsewhere, but I can’t find a straightforward answer regarding which settings to use. How can I make sure this calibration is actually correct?
It does not matter in color managed apps (if profile type describes dispay accurately, not idealized), but aim for 2.2. for non color managed. If display has a colorspace BIGGER tahn sRGB you’ll need to correct it to with other tools.
Thanks for your response – I appreciate any help on this bewildering topic.
I am aware that Monitor RGB shows non-color managed. Like you, I would also expect that the color-managed view would be correct, but it looks much worse and significantly more washed out than every other screen.
I have compared a bunch of different illustrations that I am familiar with, and they look worse to the point that I doubt that the washed out look is the artists intention. The result is similar to applying a 1.1-1.2 Gamma correction in Photoshop.
Furthermore, the images generally look pretty much the same on every monitor – my monitors at work, my home monitors, my ipad and iphone screen, other people’s monitors. They are more or less the same, and images look good.
After calibration, my screen looks worse and different than every other screen I’ve tested. Is this not reason to think that something is wrong, that there is a bug or incorrect settings somewhere? The problem is, how can I verify that this is the case?
This is an sRGB monitor, by the way.
Since I am creating sRGB images that will be viewed in non-managed contexts (video games and web), is it better to just not bother with calibration then?
No. Choice of calibration target gamma/TRC has no effect in colorspace extension (saturation). It only corrects grey.
I’m not quite following – are you saying the standard setting of Gamma 2.2 is the correct one? The results from 2.2 and sRGB looks quite different, and 2.2 looks worse.
No. sRGB as a brighter TRC near black than 2.2. Imported to photoshop and assigned to sRGB profile that PNG should render very dark greys brighter… because sRGB TRC behaves that way.
Are you saying the test image is only valid when viewed outside a color managed app, such as the browser?
I tried the Verification option with these settings – I guess it claims it’s correct? I couldn’t find a good source on which settings to use or how this feature works generally.
Thanks for your response – I appreciate any help on this bewildering topic.
I am aware that Monitor RGB shows non-color managed. Like you, I would also expect that the color-managed view would be correct, but it looks much worse and significantly more washed out than every other screen.
Every other screen are not a “reference”.
I have compared a bunch of different illustrations that I am familiar with, and they look worse to the point that I doubt that the washed out look is the artists intention. The result is similar to applying a 1.1-1.2 Gamma correction in Photoshop.
Furthermore, the images generally look pretty much the same on every monitor – my monitors at work, my home monitors, my ipad and iphone screen, other people’s monitors. They are more or less the same, and images look good.
same as above.
After calibration, my screen looks worse and different than every other screen I’ve tested. Is this not reason to think that something is wrong, that there is a bug or incorrect settings somewhere? The problem is, how can I verify that this is the case?
This is an sRGB monitor, by the way.
Since I am creating sRGB images that will be viewed in non-managed contexts (video games and web), is it better to just not bother with calibration then?
No. Calibrate & profile. Then simulate whatever scenario you want, including simulating a true sRGB screen in non color managed envirements , for example with DWMLUT.
No. Choice of calibration target gamma/TRC has no effect in colorspace extension (saturation). It only corrects grey.
I’m not quite following – are you saying the standard setting of Gamma 2.2 is the correct one? The results from 2.2 and sRGB looks quite different, and 2.2 looks worse.
No I’m saying that calibration only corrects grey ramp.
Then after calibration a profile is made that captures actual display behavior after grey calibration is applied.
That profile can be accurate (detailed) or idealized (assume perfect additive behavior and infinite contrast)Color managed apps read image’s colospace and reencodes (RGB numbers) them on the fly to display colorspace and sends them to screen.
If display colorspace (profile) does not match display (because user choice of profile or innacurate device) it does not wok as intended.
Hence running a profile verification must be the 1st test.Example 500:1 contrast display profile calibrated to 2.2 (the best it can… becase a 500:1 CANNOT BE 2.2 near black) where profile type BY USER’S CHOICE is “gamma” (powerlaw instead of actual TRC) + black point compensation (display reports as infinite contrast) will lift very dark greys beyond its expected value… and all is working as intended.
A requierement for this to work is that measuremente device is accurate and it is used as it should be (colorimeter corrections)
No. sRGB as a brighter TRC near black than 2.2. Imported to photoshop and assigned to sRGB profile that PNG should render very dark greys brighter… because sRGB TRC behaves that way.
Are you saying the test image is only valid when viewed outside a color managed app, such as the browser?
MS paint. Now there are lots of browsers color managed.
I tried the Verification option with these settings – I guess it claims it’s correct? I couldn’t find a good source on which settings to use or how this feature works generally.
I was answering while you posted XD
Aim for 2.2 and matrix profile BPC since its 1000:1. Also it is pointless to aim for a detailed profile and keep calibration “fast”, your display is not good enough and you ended with color tint in some greys, use “medium” speed.
Of course dark greys 2.2 non color managed will look darker than the same image (sRGB profile embeded) color managed… and it should behave that way by definition, see TRC plots for example on wikipedia.
If you “believe” that your content will be played non color managed on a true 2.2 power law gamma display with exactly sRGB gamut (that is a VERY BIG “IF”), then encode content that way (image profile is sRGB colorspace but with 2,2g TRC).
EDIT: Posted before I saw your previous reply.
Every other screen are not a “reference”.
I would argue that they kind of are in practice? Why would I want to create images with “correct” calibration, if that calibration looks worse to everybody else?
Furthermore, I accept that most monitors are not exactly calibrated, even if they come “calibrated” from the factory. However, if they all look very similar to each other, but hugely different from my calibrated monitor, then I believe it is rational to suspect something is wrong with how images are displayed on my screen. At least, it is worse in the sense that it lowers my ability to predict how the end user will perceive my work.
Simple recipe for other screens:
-if the actual TRC of other display do not match the TRC in OS default display ICC, that display will not render images properly color managed.
Measure them.SImple recipe for your screen:
If you calibrate to 2.2g a 1000:1 display (with its limitations, it cannot be 2.2 near black) and display profile captures accurately display behavior then:
-sRGB content non color managed will show darker near black greys than color managed. It is expected and it is working as intened.
-if you believe all other screens will behave like some colorspace X (if actually tested they will match some colorspace with some kind of accuracy) and your main target is those screens non color managed, then encode content in colospace X
-if you want to simulate how sRGB content will show “non color managed” in some other display described by colorspace X then DO NOT use “monitor RGB” in photoshop (that is YOUR display non color managed), use softproof on that colorspace X (its profile) with “preserve RGB numbers”. “X” and your display colospace may match, but not always.Example WLED 99% sRGB display (it is a little bigger than sRGB, intersection gives 99):
Calibrate to 2.2, simple matrix profille 1 curve, BPC active. Let’s assume that resulting profile describes calibrated display with reasonable accuracy
-sRGB content with embebed sRGB profile will show brither dark greys in Photoshop than in MS paint, as intended
-sRGB content with embebed sRGB profile using “Monitor RGB” will show a little oversaturated primaries because typically they are beyond sRGB (although close, see a gamut plot of your display profile) and darker greys than in sRGB.
-sRGB content with embebed sRGB profile using softproof + prevserve RGB numbers on a synthetic Rec709/sRGB profile with 2.2g will show how things will render non color managed on an ideal display with exactly sRGB gamut and 2.2 gamma.Maybe this last recipe is what you want.
This display shows 90% of DCI-P3, while you use wrong correction for near-sRGB panels. Here I attach .ti3 measurement file for BenQ PD2705Q that has similar gamut, select it in correction building window, also set near 160cdm lightness and build a personal correction for your display/colorimeter. Your test shows that all is OK, but significantly wrong gamut may result in wrong color lightness and saturation. Also check Photoshop defaults for image profiles, check for sRGB profile in test image (left bottom corner in Ps, switch from size to profile). I don’t think that there are signal limitation problems (16-235 for TVs instead of 0-255), but also check it in driver settings. Use DP connection with nVidia cards.
Thank you, I feel like I am slowly inching my way towards understanding this.
-if the actual TRC of other display do not match the TRC in OS default display ICC, that display will not render images properly color managed.
Measure them.Measure how? You mean do a Verification in DisplayCal without a profile?
matrix profille 1 curve
Out of curiosity, why select this over the default “XYZ LUT + matrix” in this case?
encode content in colospace X
With this, do you mean embedding a color profile? Would this be the same as just saving a .jpg straight from PS, since images are assumed sRGB when they don’t have an embedded profile?
softproof + prevserve RGB numbers on a synthetic Rec709/sRGB profile with 2.2g will show how things will render non color managed on an ideal display with exactly sRGB gamut and 2.2 gamma.
This sounds like what I want. And the result is like you say: with softproof on, the darks are slightly darker.
However, the difference is much smaller compared to Monitor RGB, which will crunch the blacks a lot more and add saturation, also like you said. Monitor RGB looks closer to what I would expect and what I think looks good and what I recognize from other screens.
Working in the darks of an image is actually more difficult after color correcting, since the bigger difference between values creates a bunch of banding. I can’t get completely smooth transitions in the shadows anymore.
So I guess I am still somewhat at a loss. My own definition of “correct” or rather “desirable” colors would be basically be the average of good quality consumer monitor out there, in other words that which lets me predict how the greatest number of people will see my work. And so far the closest I’ve gotten to that is to not color correct at all? I think?
That is by the way how every other artists at the game studio I work at operates. Part of my hope with finding a more robust way was actually to convince other people to work in a consistent manner, to avoid different screens showing slightly different results.
This display shows 90% of DCI-P3, while you use wrong correction for near-sRGB panels. Here I attach .ti3 measurement file for BenQ PD2705Q that has similar gamut, select it in correction building window, also set near 160cdm lightness and build a personal correction for your display/colorimeter. Your test shows that all is OK, but significantly wrong gamut may result in wrong color lightness and saturation. Also check Photoshop defaults for image profiles, check for sRGB profile in test image (left bottom corner in Ps, switch from size to profile). I don’t think that there are signal limitation problems (16-235 for TVs instead of 0-255), but also check it in driver settings. Use DP connection with nVidia cards.
Here is the panel for my monitor: https://www.panelook.com/M280DCA-E7B_Innolux_28.0_LCM_parameter_45661.html
It just says “WLED” so I chose the “LCD White LED” Correction in DisplayCal. Do you know if there is a more appropriate one?
My test images don’t have color profiles embedded, which in my understanding should mean Photoshop treats it as sRGB by default, right?
Thank you, I feel like I am slowly inching my way towards understanding this.
-if the actual TRC of other display do not match the TRC in OS default display ICC, that display will not render images properly color managed.
Measure them.Measure how? You mean do a Verification in DisplayCal without a profile?
Remote (non color managed browser)
Plugged to your computer (using whatever profile you want, EDID or driver icc for that monitor seem a good choice, it “should” behave that way)
with HCFR
.. etcmatrix profille 1 curve
Out of curiosity, why select this over the default “XYZ LUT + matrix” in this case?
simplifies work to Photoshop, since it has no dithered ouput. Less banding de to rounding errors in color management.
Only valid if display can be described by such idealized profile (it should work in your display)encode content in colospace X
With this, do you mean embedding a color profile? Would this be the same as just saving a .jpg straight from PS, since images are assumed sRGB when they don’t have an embedded profile?
no. Embeding may be assigning or converting. beware. choose what you really want to do.
No profile but you belive that content is X = assign
reencode to show same color in colorspace X = convertsoftproof + prevserve RGB numbers on a synthetic Rec709/sRGB profile with 2.2g will show how things will render non color managed on an ideal display with exactly sRGB gamut and 2.2 gamma.
This sounds like what I want. And the result is like you say: with softproof on, the darks are slightly darker.
However, the difference is much smaller compared to Monitor RGB, which will crunch the blacks a lot more and add saturation, also like you said. Monitor RGB looks closer to what I would expect and what I think looks good and what I recognize from other screens.
but not accurate (it is non color managed).
Working in the darks of an image is actually more difficult after color correcting, since the bigger difference between values creates a bunch of banding. I can’t get completely smooth transitions in the shadows anymore.
So I guess I am still somewhat at a loss. My own definition of “correct” or rather “desirable” colors would be basically be the average of good quality consumer monitor out there, in other words that which lets me predict how the greatest number of people will see my work. And so far the closest I’ve gotten to that is to not color correct at all? I think?
That is by the way how every other artists at the game studio I work at operates. Part of my hope with finding a more robust way was actually to convince other people to work in a consistent manner, to avoid different screens showing slightly different results.
sRGB content showed in Photoshop with no softproof and without disabling color management it is the accurate way. Same grey ramp (same RGB numbers) have brighter dark greys than in “sRGB gamut + 2.2.g” colorspace… and it should be that way. PERIOD.
If they look brighter than it should, its creator did it in a wrong way and it is his fault.Some assume that sRGB content should be shown on 2.2g display with exactly sRGB colorspace, but that assumption breaks color management (you won’t be able to show content on any other colospace accurately). If you want to create content with that belief, use softproof method (recipe 3)
Using Monitor RGB is wrong.
