- This topic has 34 replies, 4 voices, and was last updated 1 year, 4 months ago by
.
-
Posts
-
Hello everyone,
Recently, while re-profiling my monitor, I decided to dig into a problem that had caught my attention before.
I took this issue seriously because it concerns the monitor’s black levels, which significantly impact my photographs.
For the past week, I’ve been experimenting, testing, and reading online, but I now need your observations as a reference.This is long post, so for those short on time or confident in their monitor’s calibration/profile, could you please run the tests below and share your results with me?
1: http://www.lagom.nl/lcd-test/black.php#blacktest.png
- What is the difference between the background and level 1? Do you need to look carefully, or is it clearly visible?
Which level is the first one you can comfortably see?
2: https://www.drycreekphoto.com/Learn/Calibration/monitor_black.htm
- The same questions as for Test 1.
I tried to take two photos as a reference, please see the attached file “Custom ICC & sRGB ICC”.
If you could provide your answers, I’d greatly appreciate it.
Thanks in advance!—————————————————–
For those interested in more detail, feel free to keep reading. =)
System Information:
Operating System: macOS (also on windows)
Monitor: Dell U2723QE with a factory-specified 2000:1 contrast ratio, %100 sRGB, %95 DPI-3 color. Good monitor..
Calibration Tool: SpyderX Pro (using its native software). It reports the monitor’s contrast ratio as -/+1750:1.
The monitor comes with a factory calibration report for sRGB mode, guaranteeing Delta E <2 accuracy.My verification process for factory calibration:
Given my available resources, I validate the factory calibration as follows:A. I set the monitor to sRGB mode and profile it. Almost no changes occur in the colors, and the results are excellent on both.
( Comparison: sRGB IEC61966-2.1 vs. my custom profile. )B. I set the monitor to Native mode (Custom RGB) and profile it. The color changes are noticeable.
( Comparison: Dell EDID vs. my custom profile. )When I compare the profiles from A and B profiled, I find that the colors are identical.
( Comparison: Native profiled vs sRGB profiled vs sRGB factory sRGB IEC61966-2.1 = same collors )My conclusion:
The sRGB factory calibration appears to be solid.
Do you agree?————————————
The Problem / Dilemma:
Regardless of the monitor’s preset mode, the only thing that significantly changes after profiling is the representation of very dark tones.My test Results:
1: Lagom Black Level Test: http://www.lagom.nl/lcd-test/black.php#blacktest.pngA: Monitor: sRGB Preset | Profile: sRGB IEC61966-2.1:
Level 1 is barely visible; in a dark room at 160 cd/m², it’s only noticeable with careful observation.
Level 2 is more distinct.
Level 3 is even more visible.B: Monitor: sRGB Preset | Profile: My Custom Profile (After Profiling):
Level 1 is easily visible.
Level 2 is clearer.
Level 3 is much more distinct.2: Dry Creek Photo Black Level Test: https://www.drycreekphoto.com/Learn/Calibration/monitor_black.htm
I will perform this test soon, but it should be technically similar to Test 1.Please see the attached pictures. Also added extra information about monitor.
————————————–
The Question:
Which black level is more accurate?If your results align with (A) sRGB IEC61966-2.1, I will stop calibrating with the SpyderX and trust the factory calibration while using the sRGB IEC61966-2.1 ICC profile. (maybe sell the SpyderX and buy x-rite when needed after a while).
If your results match my custom profile (B) (where Level 1 blacks are easily visible), I will continue using the monitor in Native mode with my own profiles.
This is critical because the contrast differences in my edited photos are unacceptable with this .
If I stick with sRGB, I’m also considering returning to Windows with peace of mind. =)
————————————–
Workflow:
Edit photos in Lightroom, export to jpeg with sRGB color space.
When needed, export from Lightroom to Photoshop and save as jpeg with sRGB color space.Thank you very much for your thoughts and responses!
SpyderX Pro on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.I’ve been reading more online and wanted to add a few additional points.
I always use gamma 2.2 when profiling.The first attached photo: sRGB IEC61966-2.1 (A)
The second attached photo: Custom SpyderX ICC with gamma 2.2 (B)The monitor’s mode, whether sRGB or Native gamut, does not make a difference in the blacks after profilling. There is always big difference in end blacks before/after profilling.
Sorry for spamming, can’t edit my post before.
I just realise that I can take screenshot of it.
I also testen Eizo gamma test -> https://www.eizo.be/monitor-test/
sRGB IEC -> Between gamma 2.2 & sRGB
Profiled -> sRGB.I also measured with “Color Sync Utility” calculator.
1,1,1 is translated to 6,5,6.
2,2,2 is translated to 10,9,9The Question:
Which black level is more accurate?The measured one, although SpyderX is not as reliable at it should be.
You are on macOS, color managed desktop although very limited profile type support.
Also ArgyllCMS priorizes correcting grey color (no black unless instructed) over target TRC because as long as you record actual TRC in ICC profile color mangement engine can manage TRC conversion. In a limited contrast display measured TRC will “bend” (bending on 2D plot gamma Y vs input X) target TRC near black, while in a “fake”/ideal EDID generated ICC profile won’t bc iit’s artificially idealized to deal with deficent color management engines like Apple’s.
I also measured with “Color Sync Utility” calculator.
1,1,1 is translated to 6,5,6.
2,2,2 is translated to 10,9,9Useless comparisons due to display profile limitations in macOS (fake infinite contrast vs crushing blacks in apps using apple CMM – Adobe suite uses its own engine).
This also means that you cannot perform ANY kind of visual assesment of calibration ALONE in Safari, Edge or any other color managed browser or macOS app.
At best you can try to get that visual assesment downloading those Lagom’s PNGs and watching them with old MS paint in Windows.Hello Vincent,
First of all, thank you for your response. Unfortunately, I feel a bit ignorant compared to what you’ve said. I read your message 30 times =)
I performed the following tests/comparisons with a Windows 10 machine, as you suggested:
MS Paint + Windows Photo Viewer (old one, color managed)
Switched calibration profile ON/OFF.
Lagom test with Windows Photo Viewer:Calibration ON = Darks become darker, compared to the sRGB profile.
Calibration OFF = Darks lighten up.
To my eyes = Calibration ON is correct.
MS Paint: Stays the same between ON/OFF. (no color managed, so..)My before-edited reference photo: (sRGB 8bit)
Same as above, calibration ON seems correct (I assume).
Then I switched back to macOS:
Lagom test:
Calibration ON (Profiled) = Same as calibration OFF on Windows! (Huh!). Darks are very, very bright like my first post.
Calibration OFF (sRGB profile) = A little darker than “Windows Calibration ON”.My before-edited reference photo:
Awful when I load the profile I made! Awful, much less contrast.
It seems like macOS doesn’t load the LUT? (two weeks ago updated to macOS 15.1)Attached photos:
1: Mac Profiled
2: Mac sRGB
3: Windows Profiled
4: Windows sRGB
Viewed with Windows Photo Viewer (old one, icc aware, not full screen)It seems that macOS is having issues. I was using it only because I thought the colors were accurate.
I think the healthiest solution would be to completely revert to the sRGB color profile and continue with Windows, right?
I’m curious about your choice and why you selected macOS Profiled. I definitely say that it’s calibrated on Windows, but on macOS, the darks really look way too bright.
What I find most interesting is that I’ve been following people online who have been doing serious work for over 20 years, especially one professional certified by Epson in color/printing. They even have the lighting in their entire room set up for this purpose, they print, shoot advertisements and photos for big brands. Now, they all say they’ve started to use Apple’s displays as a reference, claiming they’re very well-calibrated and they trust them blindly. But when I calibrate, the contrast basically collapses.
What I mean is that, when on all screens, 1,1,1 is almost indistinguishable, and 2,2,2 starts becoming visible and easier for the eyes to see, when I apply the profile on macOS with SpyderX, 1,1,1 literally shouts “I’m here!” — how can this be considered correct? I just can’t understand it.
On the Windows side, however, it’s calibrated, and the darks are more subtle. 1,1,1 is hard to see, and 2,2,2 gradually starts becoming more visible. Compared to sRGB, the transition balances are more consistent (on dark tones).
I can no longer decide what is right or wrong. I’ve been sitting in a dark room staring at the screen for a week, and it’s driving me crazy =)
was looking at the monitor in a completely dark room. Now, when I turned the lights on, I realized it’s not as bad as it appeared in the dark in macOS Profiles.
Now I’m totally out of it =)
But I like Windows in dark definitely more.
The only visual appraisal of grayscale calibration “JUST CALIBRATION” is MS paint. No other viewer with potential color managed canvas.
In other test you’ll see:
-calibration in VCGT (grayscale calibration loaded in GPU), which includes color tints in greys due to various factors
-mismatch between profile data and actual monitor behavior, by software simpolification, by user simplification
-color management engine limitations, treatment of out of gamut colors (all bellow actual black stored in profile)
-rounding errors…
That will explaing dark magenta tints…So instead wasting time and assuming “well behaved IPS-liek display”, use DisplayCAL not Spyder software. 2 profiles:
-matrix, SINGLE CURVE, black point compensation (fake infinite contrast), calibration speed low or medium : this works for macOS / Firefox/Edge
-matrix, SINGLE CURVE, NO black point compensation (actual contrast), calibration speed low or medium : this works for Adobe suite, won’t work properly on macOS dekstop due to macOS limitations.
(on macOS you need new DIsplayCAL-py3, for WIndows banilla one downloaded here).What I find most interesting is that I’ve been following people online who have been doing serious work for over 20 years, especially one professional certified by Epson in color/printing. They even have the lighting in their entire room set up for this purpose, they print, shoot advertisements and photos for big brands.
No.
If you print, you softproof. If you softproof L* of black goes up to 5-10 (typical) and even higher. Hence all those claims are nosense or misunderstood them.Now, they all say they’ve started to use Apple’s displays as a reference, claiming they’re very well-calibrated and they trust them blindly.
No.
Even on new ones like XDR ones… it works bc thet are color managed: primaries are known by technology (WLED PFS with exactly P3 display, no AdobeRGB green), grey is neutral (that is calibration) and close to reference TRC on brightness of eahc grey and white may be around D65. That’s all.
If you show an sRGB imae on photoshop on those displays, it show accurately bc that simplified factory profile matches display behavior.You should start taking a course or webinar on color management. Of if you do not have time, “matrix, SINGLE CURVE, black point compensation (fake infinite contrast), calibration speed low or medium” and use it, whatever it shows Adobe is OK. If dark greys have coloration buy an i1Displaypro.
Hi Vincent.
Yes, I’ve been reading and studying for quite some time. Following your advice, I found a few seminars, recorded online, and I will watch them.
I installed DisplayCAL, and after a bit of struggle, I started to understand it. I had used it a little before as well on Windows. Thanks to Erkan here for his contributions.
Unfortunately, I couldn’t find where the profile options you mentioned (matrix, SINGLE CURVE, black point compensation) are located. I searched in the forum and also Googled quite a bit. The only part that seemed relevant was the ‘3D LUT’ section. But I think we won’t be doing anything with that, am I right?
I calibrated as follows:
– Display & measurements
Dell, Spyder, White LED with Phos.-Calibration
Whitepoint, As measured
White level, As Measured (160cd)
Tone curve, As measured, sRGB, gamma 2.2
I have attached the reports for these below.– Profilling
Testchart, auto-optimized.Today, I noticed something else in Photoshop. When I don’t select the gamma 2.2 profile, the colors in Photoshop and macOS appear to match, and they correspond to what I edit.
Example:
Tone curve, as measured = The whole system works correctly.
Tone curve, sRGB = The whole system works correctly.
Tone curve, gamma 2.2 = The whole system works correctly, except for Photoshop (much brighter).
SpyderX profile = Same as DisplayCAL gamma 2.2.Also, when I switch the profile to gamma 2.2, I see a two-step contrast drop in Photoshop. This is something I hadn’t noticed before. I reset it to factory settings, and the result was the same.
What I don’t understand is, about a month ago I edited around 1500 photos in Lightroom, and this issue wasn’t something that stood out to me at all. I don’t understand why it’s happening now. Also, I’ll look into Photoshop compatibility with macOS 15.x. Because I edited the photos on 14.x, I want to make sure there are no issues before finishing the work.
Raports;
1: sRGB to sRGB
2: Tone curve: as measured
3: Tone curve: sRGB
4: Tone curve: gamma 2.2
5: Spyder calibration (maybe I checked this wrong, was first try)t’s a gamma conversation problem. I also found this option on the internet (see attachments).
When I choose Apple CMM, there’s no problem anymore, but Lightroom still has the issue, of course.The problem only occurs at the low low ends, when the gamma is targeted to 2.2.
First of all spyder is not very relaible since Wide LED setting trys to correct all kind of different widegamut led backlights and may be more limited that xrite i1d3 in low light
Tone as measured = no grey calibration.
-Profile-sRGB-to-sRGB.html default settings show a cool tint in blue with a mild green-pink cast
-Tone-curve-as-measured.html mearurement is done in a wrong way, so it’s useless
-Tone-curve-sRGB.html shows that you did not bother to correct white point… so whatever. (or maybe you are using the default “wrong” whitepoint from factory sRGB preset)
-Tone-curve-gamma-2.2.html same
-Spyder-Software-Gamma-2.2.html same
You can see actual TRC on the two last profiles by inspecting profile in DisplayCALAlso you cannot use apps using Apple CMM for comparisons since it has limitations regarding TRC near black and Apple-built profiles from edid have fake TRC near black.
Regarding DisplayCAL full configuration is accesible through options, shows advenced options. Matrix profile type, single curve, no BPC should be the default settings for macOS.
Usually is the preferered configuration for any well behaved >700:1 IPS display since “limited” CMM can work with them without issues.Anyway, with a >1000: display, *single curve* matrix profile with BPC (fake infinite contrast), Adobe PS should render things properly, AS PROFILE INSTRUCT IT to do. Same if profile has bno BPC but yu use adobe’s CMM, not apple’s.
This means that:
-any grey *coloration* (not “steps with same color) in greyscale is caused by bad calibration software (not displaycal) or by too few measurements (calibration speed best set to medium or low if thera are grey tint troubles), or by GPU limitations (nvidia, intel without dithering)
-any mismatch in L* for TRC is likely caused by measurement device, but you’ll need a better one to compare.Also this means that:
Tone curve, sRGB = The whole system works correctly.
Tone curve, gamma 2.2 = The whole system works correctly, except for Photoshop (much brighter).is false since whateverAppe CMM is showing si irrelevant by all its CMM simplificaions. The system is not working correctly (or more accurately “you do not know if it’s working”) because it was not the reference, never was.
****
If you thing that measured near black levels form SpyderX are not accurate, which may be true, get an i1displaypro from Xrite (or current calibrite counterpart, remember that Calibrite SL = old colormunki display, regarding future monitors with HW cal). Nobody buys spyders if done a previous reseach.
Otherwise, with your current gear, the only valid rendering comes form PS using ACE, not from Apple’s CMM or any of its tools, inlcuding desktop.
Also since you are on macOS it’s useless to use sRGB preset. Full native gamut and fix whitepoint (although SPyder generci correction may be innacurate).
Hi Vincent,
Thank you for your detailed explanation and patience.
I’m having trouble getting the white balance right with the SpyderX – it gives a green tint. I’m using my laptop screen as a reference for whites. But when I use the sRGB preset, the RGB sliders are locked. That’s why I keep it as the default and trust the factory calibration.
———————-
Rules:Don’t trust Apple’s CMM.
Trust Adobe’s ACE.———————————-
macOS Test:I tested with and without BPC (Black Point Compensation):
With BPC: Photos look okay on macOS, but blacks look bad in Photoshop.
Without BPC: Photos look almost okay on macOS, and blacks look good in Photoshop.The conclusion is the same as what you told me.
Without BPC, macOS showed almost the same results as Photoshop. The only difference was in the very dark blacks, but it was barely noticeable in normal dark photos. So, macOS did a decent job, not perfect, but it’s usable.
I can live with it because it doesn’t really affect much on photos, only the very deepest blacks.
———————
Windows test:
I also compared it with a Windows 10 machine, using the same monitor (without DisplayCAL this time):
– I just profiled the monitor without thinking much about it to 2.2 with white as measured (noob action).
– Compared it with Photoshop and the old Windows Photo Viewer (which is color-aware).Differences:
The differences were very small in the Lagom test. Only the first two blacks (1,1,1 and 2,2,2). But this was only noticeable in a very dark room, when pixel-peeping.So, what I see in Photoshop is the same in Windows Photo Viewer (or any other color-aware program).
——————————
I’m curious about your thoughts on Windows.
I’m asking because I’ve been using macOS for the last 2.5 years, and I’m getting tired of it. The hardware is great, but the software feels arrogant and slow. How scaling works drives me crazy every time.
What I want to do is stick with sRGB (or full gamut when editing) and switch back to Windows 11 for a better experience. (After testing on Windows 11)
Is there anything I might have missed or overlooked on the Windows side, in your opinion?
I added the last test rapport in attachments.
Hi Vincent,
Thank you for your detailed explanation and patience.
I’m having trouble getting the white balance right with the SpyderX – it gives a green tint. I’m using my laptop screen as a reference for whites. But when I use the sRGB preset, the RGB sliders are locked. That’s why I keep it as the default and trust the factory calibration.
If U2723QE + SPyder X + Wide LED/ PFS setting gives you a color tint aiming for D65 while in Dell’s “custom color” setting, adjusting only brightness & RGB gains… that you can try DisplayCAL visual match under “chromaticity coodintaes” settings for white, last icon with 3 balls. New popup will open. Drop sliders till you get a match on whatever display you use as reference (that does not mean that the other display is OK, just that it is “your reference”). At the end of the process it will measure visually matched white and set it as target CIE xy for white (xy won’t be D65). Then use RGB gains on your dell to get that alternative white.
———————-
Rules:Don’t trust Apple’s CMM.
Trust Adobe’s ACE.———————————-
macOS Test:I tested with and without BPC (Black Point Compensation):
With BPC: Photos look okay on macOS, but blacks look bad in Photoshop.
Without BPC: Photos look almost okay on macOS, and blacks look good in Photoshop.The conclusion is the same as what you told me.
Without BPC, macOS showed almost the same results as Photoshop. The only difference was in the very dark blacks, but it was barely noticeable in normal dark photos. So, macOS did a decent job, not perfect, but it’s usable.
That is caused by Apple CMM lacking of BPC, so when it sees that some image in an ideal colorspace like sRGB with infiniute contrast (RGB 0 -> L* 0) and your profiles is not able to show it, it “cuts” since it is out of gamut.
Adobe ACE if BPC isn enabled in their apps “deforms” near black RGB colors to preserve near black tonal separation.Note that Adobe ACE BPC and DisplayCAL BPC acts in different way with the same name
ACE: map image RGB 0 to display profile RGB 0, maybe lifting a little “in gamut” colors so RGB 0 to some RGB value in gamut can preserve tinal separation. There should be an official Adobe PDF about that, I do not remember link
DisplayCAL: BPC means “lying” in ICC profile about actual monitor black (RGB 0 -> L* 0) so all “less capable” CCM like Apple’s have no troubleIf display profile has BPC and you have ACE with BPC enable (by default) = do nothing, display porfile says “infinite contrast”
If display profile has actual nonzero L* for black RGB 0 => ACE kicks in and do that kind of precerptual rendering near black.Actual L* in profile is relevant for softproofing on low contast configuration, like the typical 100-120nit with K to 0.5nit. So when you softproof in ACE won’t lift simulated ink black in a not realistic way.
Otherwise for a well behaved IPS display >800:1, taht validates ok with a matrix profile… better to set RGB 0 -> L* 0 on display profile to avoid issues with “simple” CCMs.Windows test:
I also compared it with a Windows 10 machine, using the same monitor (without DisplayCAL this time):
– I just profiled the monitor without thinking much about it to 2.2 with white as measured (noob action).
– Compared it with Photoshop and the old Windows Photo Viewer (which is color-aware).Differences:
The differences were very small in the Lagom test. Only the first two blacks (1,1,1 and 2,2,2). But this was only noticeable in a very dark room, when pixel-peeping.So, what I see in Photoshop is the same in Windows Photo Viewer (or any other color-aware program).
IDNK if old windows image viewer has BPC, it’s unlikely since it did not support 3d mash profiles, just 3×3 matrix part of them but IDNK.
Hi, there.. I was reading through this thread and you made a comment “or by GPU limitations (nvidia, intel without dithering)”
Other than setting my Nvidia settings to the full DNR ie 0-255 from the TV settings I never thought about any settings or adjustments that I should make at the video card/driver level…
If you have a moment to clarify that comment you made …ie should there be some dithering enabled ?? I just wasn’t clear what you mean…
Thank you in advance.
If you map 256 to 256 there is only one combination that do not repeat values: the no calibration scenario where VCGT input equals output.
Although VCGT tables have 16bit data for each 256 input, this needs to be loaded to GPU LUT, then truncated to actual output bitdepth (8-10). Banding may happen.
AMD driver does dithering, hence visually equivalent to HW cal with no “non color managed” banding even over 8bit DVI channel. Dithering adds “decimals” (ivisually and for measurement devices)
Nvidia does not but AFAIK you can try to enable it with novideosrgb app. Contact developer for nvdia dithering support.
Intel GPU does not, hence banding induced calibration is common.Commandline uncalibrated screen report should be able to test this.
On any of these GPUs using DMWLUT with applies dithering on LUT3D calibration on shaders ( instead of dedicated VCGT hardware) solved these issues. But this has other implications: capture display behavior + VCGT for grey, create a synthetic version of it with nominal primaries, white and TRC, create LUT3D with embeded VCGT, load lut3d and use synth profile as display profile. And such LUT3D will have a greyscale corrected by interpolated 65 point instead of up to 96 interpolated to 256 from VCGT crated with ArgyllCMS.
But ofr a well behaved IPS, it works very well. - What is the difference between the background and level 1? Do you need to look carefully, or is it clearly visible?
