- This topic has 17 replies, 3 voices, and was last updated 5 months, 1 week ago by
.
-
Posts
-
Hello. This post has a few problems that I have tried to solve myself, but the more I dig deep the more I find wrong.
I have a Viewsonic VP2785-2K wide-gamut monitor that I calibrated with the Colorchecker i1 Display. I am calibrating for both video color grading in Resolve and photo editing via Lightroom.
Problem 1: I swear there is an issue with my levels, but I cannot confirm. The shadows are incorrectly displayed in Lightroom as opposed to my color-managed browsers and Instagram (also color-managed). The shadows only appear to match Lightroom when I open the image in ImageGlass, which uses my ICC profile. Then again, wouldn’t the image appear the same in my browsers if they too use that same ICC profile? This confuses me, but I took a screenshot to show you. See here. See how they’re darker for Instagram on the left? It appears that way in everything except ImageGlass and Lightroom, phone included. I copied and pasted all the information from the ICC profile it’s using right now in case it would tell you if something was wrong. That’s right here. I don’t remember being given the option to select full or video levels before, but now I have that option if I re-calibrate. Should I select one or the other? My monitor has an ‘auto’ option, but I am not sure when it goes to ‘full’ or ‘limited’ levels.
Problem 2: Did I miss something after calibrating? I thought that all I needed to do when the calibration was complete was to install it, but I have read on forums that I need to do something called verification? I still don’t understand what verification is? What does it do?
Problem 3: People have suggested a decklink or some sort of external card that holds LUTs and calibrations? Can you direct me to a good source to read about that? I thought all I needed was the ICC profile to interpret colors for me, and that it would be accurate enough.
I am very very confused. Any help would be appreciated.
Thank you.
“VP2785-2K #1 2022-09-23 18-47 120cdm2 D6500 2.2 F-S XYZLUT+MTX”
Very likely that your “fast” image viewer and browser only use matrix info of an XYZLUT profile. Or if you are using macOS same for OS engine.
If display is well behaved recalibrate (use medium or slow speed, not fast, if you have grey range issues) using a “matrix + single curve + BPC” profile. This will describe (create an ICC) the display in an idealized way and IF display is well behaved it will match actual calibrated display response.
If display is not well behaved and your browser/fast image viewer are only able to use matrix part of an XYZLUT profile… your issue is not solvable and the culprit are those applications (browsers + ImageGlass) or the low quality of display. 50%-50%.
In that situation (bad behaved display) you can try to “linearize” it using DWMLUT, using taht app with yiur custom XYZLUT profile simulating a “matrix idealized profile” with same primaries & white. Once LUT3D for “linearizing” display is loaded set that “idealized display profile” (matrix+singlecurve+BPC + no GPU calibration) as current display profile. Take a look on DWMLUT thread because there are examples.Sample if we assume that your display is bad beaved and a single curve matrix is not a good match for actual display respose:
-Your display profile: “VP2785-2K #1 2022-09-23 18-47 120cdm2 D6500 2.2 S XYZLUT+MTX” (slow)
-Create a synth profile with Displaycal apps (see displaycal folder), same primaries as above, gamma 2.2, same whitepoint (D65)
-Create a iridas .cube LUT3D with DIsplayCAL LUT3D creator (source : idealized disolay, destination : XYZLUT profile)
-Load that LUT3D in DWMLUT
-Assign idealized profile (the synth profile you created) as “default profile” in your OS for your display.Then your “limited color engine” apps like your browsers or ImageGlass will deal with a simple matrix porfile… and Lightroom too.
If display is well behaved recalibrate (use medium or slow speed, not fast, if you have grey range issues) using a “matrix + single curve + BPC” profile. This will describe (create an ICC) the display in an idealized way and IF display is well behaved it will match actual calibrated display response.
Should I adjust the levels settings in DisplayCal from Full to TV or Auto?
I have a correction for my monitor downloaded from the corrections database, and I’ve been using that but it automatically sets the instrument mode to Refresh (generic). Should I go with a Phosphor WLED IPS selection instead, which allows me to choose LCD? Wondering if that changes anything.
It won’t change your results because it has nothing to do with your issues.
The culprit is what I wrote above and if you wish to solve it you’ll have to choose one of the two approaches I wrote.
By default i’ll choose the 1st one (simple curve matrix BPC profile) because it’s an IPS for photo use. If that display behavior is not able to be matched with a matrix profile then it’s a sh*t monitor like most of these low cost widegamut models and you should return it an get an Eizo CS2731 for +300€ more (which should be every body’s first guess for an AdobeRGB 27″ QHD).
So let me get this straight.
I removed my V2 ICC profile and set Windows back to its default sRGB profile. I started Lightroom up again and immediately the images matched, pointing to my profile as the issue. I re-calibrated three times on medium speed and got identical results, but when I apply any of the profiles (some were curves + mtx, some were XYZ LUT + MTX) that DisplayCal put out, Lightroom once again displays shadows differently. The shadows are brighter again. EVERYTHING else I used for proofing except for ImageGlass, including my phone, shows darker shadows, including when I went back to the Windows default sRGB profile. ImageGlass runs off of my monitor profile and it seems to show an identical image to Lightroom. Brave and Instagram (powered by Edge) which are color-managed apparently, still display the darker shadows as indicated in the image I linked originally.
This means that my monitor is not “behaving”? Can you explain?
This means that my monitor is not “behaving”?
No, it does not mean that. It is not even related to that.
It was explained previously regarding each app color mamagement engine limitations.Also all your test with sRGB default profile are not valid, explained bellow:
Can you explain?
Assigning any profile to display and use it as diagnosis is useless because unless profile and display match results (the visual results) will be wrong. Hence what you seen in LR or any other tool with sRGB profile as display profile means nothing unless that profile matches display ( = validated with measurement).
Same apply to any visual test you make with a phone in a color managed app. No test, no trustable.Also you provide no data of any of your claims, and i mean claiming that this or that behavior is accurate (Measurement).
The accurate result is the profile that accurately describes display behavior (under certain x dE tolerance) and at the same time is supported by your apps.So again, calibrate using profile type single curve + matrix profile, then validate with measurements. ANY OTHER TEST, ANY OTHER PROCEDURE is useless and a waste of time.
If contrast was crippled due to bad panel or uniformirty compensation you can play disabling BPC on profile creation, just single curve + matrix without BPC. Since you provided no data we cannot know so you’ll have to test by yourself.
Some apps may not like a display profile that stores actual black color instead “fake infinite contrast” (BPC), and since it’s image viewer limitation there is little you can do about it, unless you avoid all profile type that are not properly supported by that app.FYI: Standard sRGB profile stores L* 0 for RGB input 0 (infinite contrast colorspace)
Assigning any profile to display and use it as diagnosis is useless because unless profile and display match results (the visual results) will be wrong. Hence what you seen in LR or any other tool with sRGB profile as display profile means nothing unless that profile matches display ( = validated with measurement).
Same apply to any visual test you make with a phone in a color managed app. No test, no trustable.I am trying very hard to understand. Maybe my brain is running out of space.
So you are saying that my tests are all inaccurate because I’m comparing my Lightroom issue to a Lightroom and ImageGlass that uses the default Windows profile? I’m not testing against that Windows profile to see how accurate my calibration is; I’m testing to see if the profile is causing a misrepresentation somewhere. See my first and my second images. The first is with the default Windows profile enabled- see how the shadows remain the same? Now look at the second image. See how the shadows on the left are darker than the ones in Lightroom (the right side)? Those darker shadows have appeared the SAME way across four different screens. They are all dark in the same way; it is only on my monitor with my monitor profile that they appear so light. THAT is the inconsistency I was testing for, and it was proven. I got the idea from multiple sources, including this one.
Also you provide no data of any of your claims, and i mean claiming that this or that behavior is accurate (Measurement).
The accurate result is the profile that accurately describes display behavior (under certain x dE tolerance) and at the same time is supported by your apps.Okay, how can I provide the data to you? Do you need profile info? A verification html?
I appreciate your answers. I am just trying to figure this out because DisplayCal forums are probably the only place where I can find the answer.
I have attached the measurement report from my most recent calibration. It’s a single curve + mtx + bpc profile.
Am I seeing the reason for my issue in the gamma? It looks like the gamma drops to 2.
Assigning any profile to display and use it as diagnosis is useless because unless profile and display match results (the visual results) will be wrong. Hence what you seen in LR or any other tool with sRGB profile as display profile means nothing unless that profile matches display ( = validated with measurement).
Same apply to any visual test you make with a phone in a color managed app. No test, no trustable.I am trying very hard to understand. Maybe my brain is running out of space.
So you are saying that my tests are all inaccurate because I’m comparing my Lightroom issue to a Lightroom and ImageGlass that uses the default Windows profile? I’m not testing against that Windows profile to see how accurate my calibration is; I’m testing to see if the profile is causing a misrepresentation somewhere. See my first and my second images.
These test are not valid because sRGB profile can only be assigned as defauyl display profile if display behavior matches sRGB profile: primaries & TRC.
Hence you cannot say that the proper rendering comes from ImageGlass.In addition to that, some applications do not support complex “3d mesh like” display descriptions (XYZLUT profile) and only read the “simplified description” of that which is a matrix (primaries location) and TRC.
In addition to that some very simplified color management engines do not like that a profile reports actual black brightness, expecting in their oversimplificated maths an infinite contrast display. When yiou make a profile with BPC active you are storing a fake infinite contrast 0nit black to overcome that limitations.
Adobe ACE color management enigne supports all kind of profiles. Other simplified engines, like the one in borwsers do not… hence you need tio apply what I ‘ve explained in previpus posts.
The first is with the default Windows profile enabled- see how the shadows remain the same? Now look at the second image. See how the shadows on the left are darker than the ones in Lightroom (the right side)? Those darker shadows have appeared the SAME way across four different screens. They are all dark in the same way; it is only on my monitor with my monitor profile that they appear so light. THAT is the inconsistency I was testing for, and it was proven. I got the idea from multiple sources, including this one.
And these test are not valid and you are wasting your time and ours.
You need to measure calibrated display behavior and see if porfile is a match. If profile is a match ACE engine will render image properly, unless you mess up with ist configuration, like softproofing a printer profile. That will be the accurate reference.
Since some other apps won’t suport complex 3d mesh profiles.. make a single curve matrix profile with BPOC and see it it matches = measure, DIsplayCAL has tools for taht. If that simplified profile is a match to your display, that simplified porfile will have the best compatibility across all sh*tty implementations of color management engines, like browsers.
Hence choosing a simplified porfile type by default is the best option for compatibility.If display cannot be matched by a custom single curve matrix profile due its irregular behavior then read whet I wrote about DWMLUT.
Also you provide no data of any of your claims, and i mean claiming that this or that behavior is accurate (Measurement).
The accurate result is the profile that accurately describes display behavior (under certain x dE tolerance) and at the same time is supported by your apps.Okay, how can I provide the data to you? Do you need profile info? A verification html?
I appreciate your answers. I am just trying to figure this out because DisplayCal forums are probably the only place where I can find the answer.
Validation tab, Measurement report, all options default or disabled. That will validate if display profile matches display behavior.
Am I seeing the reason for my issue in the gamma? It looks like the gamma drops to 2.
It does not matter at all as long as you are in a color managed enviroment and profile matches display behavior.
You can calibrate to 1.8 or L* and a color managed enviroment will show an sRGB image with sRGB TRC.Regarding your profile it seems ok excluding the unavoidable error in black sine porfile says 0nit and acrtual black has 0.12nit, expected for an IPS display at 120nit white. But that error is low, 2nd row 02. All other things seems OK, although grey color tint may be better (selecting slow speed), I see a [0.8, -1.x] range for a* tint… but thi is not related to your issues.
Hence opening the image in Photoshop should render the image properly (unles you missconfigured it). That rendering is your reference.
If other apps are not showing this they are
a) not color managed
b) configured in a wrong wayFor example default Windows image viewer is not color managed. Firefox needs to enable color management for untagged images or HTML colors. Edge should be color managed. Chrome has an history of unreliable color management. Old windows image viewer (whcih can be restored with a registry entry, google) is partiallly color managed, it supports matrix profiles but full screen F11 is not color managed… and so on with other apps.
Since you have a widegamut display is extremely easy to test this: Make an sRGB image with 255 color patches of R , G & B. Save it with embeded profile.
If you open it in Adobe ecosystem it will show prperly with “desaturated” 255 sRGB red and gree, not your native green and red.
If any of your browser do show oversaturated green or red (native colorspace of your display) it is niot coor managed or it has color management disabled, hence it is wroing and all images shown in that app are not accurate.If you did not know about color management before buying a widegamuyt display you can take some webinar about color management: What is a profile, what happens when you send raw RGB values to display comfigured in full native gamut and such.
If display cannot be matched by a custom single curve matrix profile due its irregular behavior then read whet I wrote about DWMLUT.
Your display CAN be described by a single curve matrix profile + BPC (from your Measurement report, it’s all OK) so unless you are making a LUT3D for video editing purposes, I’ll stick that that simplified profile type since it maximizes application compatibility.
These test are not valid because sRGB profile can only be assigned as defauyl display profile if display behavior matches sRGB profile: primaries & TRC.
I understand now why it’s inaccurate. Thank you very much. I think I can comprehend the issue now, for the most part. I’m grateful for your input on my measurement report.
The issue persists even when using the simple single curve + matrix profile. Edge, Chrome, or Instagram (all should be color managed) still show me an image that is darker in the shadows. This image problem shows up regardless of my profile, whether it’s a LUT + matrix or a curve + matrix. I have made calibrations with and without black point compensation as well. The issue remains 🙁 but if you have done all you can, I appreciate it.
Lastly, I have a question about Lightroom and changing profiles. I conducted a test where I exited Lightroom, changed my profile to a different one (default sRGB JUST for the effect), then booted Lightroom back up and changed the profile back again. It kept the contrast of the sRGB profile but changed the colors. What I want to know is- what part of the ICC profile am I seeing change? Is Lightroom keeping the curve or the matrix? Of course, I restart Lightroom and everything changes back to how it should be. I only conducted this test to dissect the profile parts so I could see visually what the different parts did. I’d just like to know what part I saw change.
The issue persists even when using the simple single curve + matrix profile. Edge, Chrome, or Instagram (all should be color managed) still show me an image that is darker in the shadows. This image problem shows up regardless of my profile, whether it’s a LUT + matrix or a curve + matrix. I have made calibrations with and without black point compensation as well. The issue remains
but if you have done all you can, I appreciate it.They may be color managed but “simplified”. You need to test.
Anyway single curve + matrix + BPC should render 99% equal on all color managed apps IF they are using it.Lastly, I have a question about Lightroom and changing profiles. I conducted a test where I exited Lightroom, changed my profile to a different one (default sRGB JUST for the effect), then booted Lightroom back up and changed the profile back again. It kept the contrast of the sRGB profile but changed the colors. What I want to know is- what part of the ICC profile am I seeing change? Is Lightroom keeping the curve or the matrix? Of course, I restart Lightroom and everything changes back to how it should be. I only conducted this test to dissect the profile parts so I could see visually what the different parts did. I’d just like to know what part I saw change.
Profile has 2 parts,
one mandatory: display description. “colorspace”. It can be descrived by primaries & TRC (matrix) or by a 3D mesh like xrite i1profiler “table” profiles or ArgyllCMS XYZLUT
a 2nd one OPTIONAL: grey calibration for GPU LUTs: even using HW cal they may be opresent with a null grey calibration (input=output)
Profile description (1st one) is created (measured) with 2nd optional part applied. Hence changing GPU LUT calibration invalidates the accuracy of profile (unless GPU LUT calibration was minimal… but I hope you get the point).
This optional part is applied system wide, color managed or not. It changes/corrects grey color & gamma.Color managed apps ONLY use 1st one, mandatory. Almost all apps request to OS display profile on app start, if you change it once it started it won’t notice it. You’ll need to restart app to notice the change.
If you change profile with a color managed app started, you flush GPU LUT grey calibration to whatever new display profile has (if any). That’s the change you see.
Viewsonic may have a HW cal app for that monitor although it has no valir correction for newer WLED PFS backlights. If you choose (if you can) a matrix profile it will be single curve matrix (+ BPC?). If you choose “table” (3D mesh) it will be like XYZLUT.
You can inspect TRC near black and presence of BPC by opening DIsplay cal profile info tool, look for Tone Response Curve combo.
I ran some tests with an AdobeRGB color spectrum and an sRGB color spectrum, viewed in Lightroom, Photoshop, ImageGlass (color managed viewer) and Brave (color managed browser).
The AdobeRGB in ImageGlass matched identically to Photoshop, but Lightroom was significantly lighter in the blacks and Brave was darker in the blacks.
The sRGB in ImageGlass matched closely to Photoshop with slightly lighter shadows. Again, Lightroom was significantly lighter and Brave was darker.
So now I see that Photoshop and Lightroom are not matching? I don’t understand what is going on; it seems to be more than an issue regarding some apps not using the profiles correctly.
