- This topic has 31 replies, 3 voices, and was last updated 3 years, 2 months ago by
.
-
Posts
-
Hi guys, i am new to this forum and displaycal, what a brilliant piece of software! So Hello!
I am a colourist and out of work due to corona lockdown. Since i have time i thought i could learn all of this, since i only have knowledge of basic calibration that some softwares do. Mostly using 100cd/m 2.4 and rec 709. But the more i see, the less i know! So please help a fellow displaycal user. But over the years i am used to seeing the right picture, so i can, not like a colourimeter but say when the picture is off, especially when viewing the ones i graded myself.
What i want to do – I have access to a recently calibrated sony grading reference monitor by a professional. It was calibrated to 100cd/m 2.4 and rec 709. holds a lut in itself so its independent of where the signal is coming from. This is my reference, and the picture on it looks good to my eye.
I have a benQ sw2700pt which i want to use as a GUI. For this, i want to create an icc for the windows environment, as well as an equivalent LUT in davinci which i can use in the color viewer LUT to match it to the reference monitor since davinci skips the windows icc. I could use the palette master software and get fine results, however since i have time, i thought i could study and learn when i can for my own future to be able to calibrate other displays. since i have to work on my own for sometime now.
Monitor Setup OSD
monitor is set to adobeRGB space since i didnt want to limit the calibration, brightness – 100(before i level it to 100cd at start gets about 33 in osd), contrast -50 (did not change this when leveling at start), gamma – 2.4
I used display cal with the following settings as attached along with verification report
Problem
After the calibration, i save the LUT from the popup and install the profile from the profiling tab and clicking install on the top right beside the delete bin. The icc that displaycal installs and sets as default in windows is fine by colour. however the darks and shadows are lifted.
The LUT when used in resolve for color viewer, makes the image extremely desaturated like and the darks are lifted heavily and nowhere matches how the icc working which has lifted darks as well but atleast the colour is fine
Please help me understand and what am i doing wrong here.
Some naive questions
what does the white level and black level drift compensation do?
The correction drop down, is this like the i1 profiler where we select the tech of the display? how does this work and what does it do?
in calibration tab, what is the difference between chromaticity coordinates and by temperature? does 0.3127 / 0.329 mean 6500?
tone curve rec 1886 and rec 709, arent they same? ive also tried selecting custom with 2.4 absolute and black output offset 0 and the problem persists of lifted darks. what is black output offset?
the options in 3d LUT tab looks about similar as in calibration tab, is there some difference in here which i should know of?
Update: I am reading about the questions I asked since yesterday and partially understand.
I was able to get a better black level by setting the tone curve to custom 2.4 relative with offset to 100% however the black is still not as black as the monitor can go without a calibration. Just to understand how this is working, I’ve created several LUTs changing the offset. I understand 0-100 is adjusting the curve from towards linear to more towards dark respectively. Is this how this works?
Also when I apply the lut with offset 100 relative to resolve scopes to see what it’s doing, I see that the levels at 0 are raised. Hence the lift I see in the dark I suppose compared to how dark this monitor can go. How can I make this lut go darker in the low levels?
About the colour space in windows environment. Before calibrating, I’ve set the colour space to Adobe RGB, and this makes a very good profile to my eye in terms of color. So anything sRGB viewed in a color managed environment looks very good. However as you all know, the non color managed look very saturated.
Is there a way to create a profile that limits the space of Adobe RGB in osd to sRGB? So the non color managed part looks ok? Or do I have to limit the space in osd itself? And create a profile for that?
Please help
update: applying the LUT to only scopes does nothing! applying it to color viewer does shift it up, including the scope, regardless if the scope is set to no lut.
I am not sure how this is working, please help!
1st error: “correction none” on device tab. It is a QLED: There are user made 3nm corrections for it but some of them were made in a wrong way, not native gamut.
Search in this forum, there should be a cleaned version somewhere because this question was answered some time ago.
Without a spectral power distribution sample (“power per wavelength”) for your display raw uncorrected readings from an i1d3 cannot be trusted. It is a extremely good device but it is not exactly as “CIE 1931 2 degree std observer”. Xrite provide a way to solve this: they store actual spectral sensivities of device in devce firmware. Argyll (or Xrite software) reads this and with display spectral power distribution sample it can correct device readinsg on the fly. That’s the reason i1d3 colorimeters are so cool.2nd error: do not use Rec1886 on a very low display like this(1000:1 is “low”). It won’t be 2.4 but something else (lower value near black)
Aim for 2.4. Google rec1886 why, there shoudl be a ITU-R paper and see how TRC changes for Rec1886 as you go for lower and lower contrast.
It work for an OLED or a VA panel with led backlioght and very high contrast… but it’s very likely that you do not want this for an IPS variant.3rd potential error. Do not limit yourselft to AdobeRGB, use full native gamut (more or less the union of AdobeRGB and P3 D65, covering both). Also since this setup have native red (~P3) and native green (~AdobeRGB green) is easy to spot visually with a cyan/green/red 100% saturation sample vidoe file if certain window/viewer is not applying a LUT3D (AFAIK thumbainls are not color managed but all video window will apply tjem if you configure it as suggested bu FAQ in GUI)
Calibrite Display Pro HL on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.1st revisited: if you do not get a visual match IN WHITE saturation 100% between SW2700PT and your reference it could be:
-innacurate correction: there is no official Xrite sample for those QLED (and Benq HW calibration software has not them) so you are limited to user made corrections. Some of them (some, not all) may be not as good as it shoudl be. See next bellow and act inthe same way.
-observer metameric failure. Even with lab grade equipment you get numeric match but to your eyes they do not. It is you. Do a visual white point match: 1st match numerically, then modify subjectively until visual match. If you do this, when you make LUT3D you should choose RELATIVE COLORIMETRIC = do not mess with my visually matched white.
-poor uniformity, device match numerically in the center and if you cover all other parts of screen it mat match visually too, but due to color cast nerar ceenter or sides because poor uniformity and bad quality on these low cost low quality disoplays like widegamuts from benq… they do not look equal.
This is not correcteable, the only solution is to get an Eizo CS that does not suffer from this. I hope that this is not your situation.Thank you Vincent for such detailed reply and for being so patient. I understand a bit more now and may have more questions. As I keep learning.
As for now, I do not have a problem with colour of the BenQ. It matches quite closely to the sony and since it’s a GUI display, if it’s close enough, it’s fine for me. When I shift eyes from Sony to benq I do not see a visual difference. That is, colour wise.
My problem is with the gamma I suppose which I’ll explain. Cuz it’s in the dark areas. I’ve read the tutorial again several times for the GUI calibration so that I am not missing something.
IMPORTANT If you have opted to do iterative gray balance calibration, enable advanced options in the “Options” menu, then go to the “3D LUT” tab and disable “Apply calibration (vcgt)”. This is very important because while Resolve does not make use of the display ICC profile, it’s user interface is still affected by the 1D calibration of the profile we’re going to install later, so applying it to the 3D LUT would result in the calibration being applied twice.
I want to understand this section. When resolve is not using the icc profile, user interface is affected by 1D calibration. This 1D is the icc it means? And this isn’t used in the colour viewer it means?
I understand the 1886 you explained and I get a very less contrast image with that. So not using that. The best I could get close to sony was by using custom gamma 2.4 relative and offset to 100%.
But still the black is not black as how it is before the calibration. For my tests, I put a black letterbox in resolve by adding output blanking, in the video made everything black. When no lut is selected, the output blanking is blacker than the video black and that’s the amount of black I want to accomplish. When I select the generated LUT for color viewer, it lifts the output blanking and there is no visual difference between the black of output blanking and the black in video. This is how it is on sony except, the black level itself on sony and what I want to achieve is the black level before applying the LUT if that makes sense. I don’t want the black to lift like that. And achieve Of 0.3cd that I can do with hardware controls before the calibration. As well as 100cd.
Maybe I am messing up the VC gamma table as mentioned in the guide. But I am not sure what wrong am I doing here.
As for now, I do not have a problem with colour of the BenQ. It matches quite closely to the sony and since it’s a GUI display, if it’s close enough, it’s fine for me. When I shift eyes from Sony to benq I do not see a visual difference. That is, colour wise.
No correction = i1d3 measurements are wrong (unless by chance i1d3 is identical to std observer whcih is very unlikely)
My problem is with the gamma I suppose which I’ll explain. Cuz it’s in the dark areas. I’ve read the tutorial again several times for the GUI calibration so that I am not missing something.
IMPORTANT If you have opted to do iterative gray balance calibration, enable advanced options in the “Options” menu, then go to the “3D LUT” tab and disable “Apply calibration (vcgt)”. This is very important because while Resolve does not make use of the display ICC profile, it’s user interface is still affected by the 1D calibration of the profile we’re going to install later, so applying it to the 3D LUT would result in the calibration being applied twice.
I want to understand this section. When resolve is not using the icc profile, user interface is affected by 1D calibration. This 1D is the icc it means? And this isn’t used in the colour viewer it means?
As long as you installed that ICC as DEFAULT display profile in OS and 1D calibration (VCGT data) is loaded on GPU (using DisplayCAL tray app / loader).
But still the black is not black as how it is before the calibration. For my tests, I put a black letterbox in resolve by adding output blanking, in the video made everything black. When no lut is selected, the output blanking is blacker than the video black and that’s the amount of black I want to accomplish. When I select the generated LUT for color viewer, it lifts the output blanking and there is no visual difference between the black of output blanking and the black in video. This is how it is on sony except, the black level itself on sony and what I want to achieve is the black level before applying the LUT if that makes sense. I don’t want the black to lift like that. And achieve Of 0.3cd that I can do with hardware controls before the calibration. As well as 100cd.
Maybe I am messing up the VC gamma table as mentioned in the guide. But I am not sure what wrong am I doing here.
(bold letters)
If profile is set as default display profile, measure resulting contrast, white, black (“measurement report”, no simulation profiles). That is what you get with 1D GPU calibration and nothing else
Load a black/white video sample on resolve (try to place video window near center), use ArgyllCMS command line “spotread” (see doc, colorimeter correction is provided with -X param). Does it match measured with only 1D calibration loaded on OS desktop?
If you meant “before 1D calibration” just do the same with Display cal menu, tools “uncalibrated screen report” (no VCGT data loaded into GPU LUT). If black or white do not match there is something on GPU lut, Displaycal profile info / calibration curves should show it too.
Thank you. I am outside right now and will measure all states as you suggested. And get back here.
Well to be honest I do see a slight difference in green and red/yellow area. But I don’t want to get overwhelmed by all this knowledge! So I thought I’d figure the gamma out first and then move on to colour.
What is the proper way to go to the non calibrated state? The state in which it measures in tools > uncalibrated report. Just switch off the use my settings from tray associations?
Also if I am calibrating again, do I have to go to uncalibrated state again before profiling or does displaycal do this automatically?
I wrote about measuring black and white, nothing else.
a) with 1D lut (from ICC)
b) with LUT3D applied in GUI (1D LUT + LUT3D without VCGT)
c) without 1d lut
Green, yellow etc it is done through LUT3D in GUI. It cannot match rec709 without LUT3D.Same for grey. 1D lut “gamma” (TRC) and resulting “gamma” (TRC) in Resolve does not need to be the same. Usually people aime for
-same white as video target (D65, with OSD controsl or HW calibration)
-near native TRC (usually 2.2, with 1DLUT or HW calibration), to minimize possible banding from GPU 1D calibration
-native gamut, so 1 display profile can be used for several LUT3D in MadVR/resolve (we assume same white point target, brightness anc contrast fro all of them)
Then on LUT3D you choose colorspace (rec709) and TRC (gamma 2.4)Uncalibrated state (1DLUT) can be set system wide with DisplayCAL tray app (reset VCGT), click on load calibration from current profile to load it again.
Displaycal does something equivalent for “uncalibrated display report”, you do not need to change things manually. It will do that sequence for you.hello again, thank you so much for the help!
i did the report > measurement report (deselected simulation profile) what does simulation profile do btw? is that the ICC over the 1d? and
report > uncalibrated report and here are the results. the black point seems off!
what does this imply?
also i tried the spotread, but not sure how to get the reading out of it. for the x parameter, i dont know where is that css file and how to generate the spectral samples for my i1 that you mentioned its important. Are spectral samples unique to every individual i1?
update : i got the clean correction ccss as you mentioned from this post. Thanks florian!
however i am not sure if i understand the other points that you mentioned in that thread.
a ) Move to OSD “User” or “Custom”, the one with RGB gains available in OSD. Use the “good” correction modified by you or RGBLED.
Get D65 white using gains with the help of DisplayCAL. Modify that white manually if it does not look “white” to you (specially if you use RGBLED correction which is not accurate for your backligt)
Once you have a visual white near D65, calibrate and profile, make LUT3D for Resolve.. etc… but if you modified white to get a “whiter visual match” you must set white target “as measured / native”, not D65 coordinates (and relative colorimetric approach)firstly there is no native space in the OSD of this monitor. the largest would be Adobe RGB. When you say get d65 using gains with help of displaycal do you mean the interactive in the beginning? and then cancel it out and not profile? If yes, how is the LUT created without profiling?
b ) Use Benq’s Palette Master Elements, get a native gamut D65 hardware calibration (if you can).
Then go to DisplayCAL, use RGBLED correction or the “Good one” edited by you, verify white.
If white is OK in numbers and visually, use DisplayCAL to get LUT3D… etc.
If white is not OK visually or numerically apply the same as in a ) but here you cannot use RGB gains in OSD so any attempt to get a visual match to D65 should be done with “visual white point editor”.This measuring white is again in the interactive in the beginning and cancel out?
Device link is an ICC equivalent of a LUT3D. Transforms some input colorspace RGB values (like simulation profile rec709) to RGB values in other colrospace (display profile) that have the same coordinates.
When you make a LUT3D for Resolve or madVR, a device link ICC is computed too in the same folder (bigger one if there are 2).
Using a simulation profile and a device link you can simulate an equivalent to a LUT3D (device link) that simulates some colorspace (simulation) in your display.Plot that CCSS correction, i button next to it. Myabe it is a bad one. Avoid CCSS that store channels that look like linear combination of others… but some people find it difficult to spot visually. IDNK about you. A QLED should be 3 lonely peaks for each channel in their respective wavelengths. If you see some amount/peak of green in red wavelengths for example, it means that it is not native, it’s likely that “red 255” in the display that was measured to create that CSS is a simulation of sRGB/AdobeRGB red. CCSS should be measured (created by people who upload them) at native gamut no matter what you want to do with it.
Uncalibrated report says that your white (on OSD, monitor standalone without computer or GPU calibration) is “not white” BY FAR using that correction.
Measurement reports that your contrast is totally destroyed, 3xx:1. It should be 800-1000:1 and uncalibrated is in that range, so your VCGT calibration is faulty.
Redo with suggested values, my previous post: D65 white (set on OSD), yu cant set brightness “as measured” and set manually 100cd/m2 on OSD (as measured just means DisplayCAL don’t care, whatever you put manually is fine), gamma 2.2 (uncalibrated repot say a mean value, but IDNK actual values for each grey, set a value 2.2 or close to actual mean-native value, maybe you can twaelk it on OSD), calibration speed low or medium (depending on grey issues on uncalibrated screen IDNK yours)Device link is an ICC equivalent of a LUT3D. Transforms some input colorspace RGB values (like simulation profile rec709) to RGB values in other colrospace (display profile) that have the same coordinates.
When you make a LUT3D for Resolve or madVR, a device link ICC is computed too in the same folder (bigger one if there are 2)Thank you for the explanation, i get this now.
Plot that correction, i button next to it. Maybe it is a bad one. Avoid CCSS that store channels that look like linear combination of others… but some people find it difficult to spot visually. IDNK about you. A QLED should be 3 lonely peaks fro each channel. If you see some amount/peak of green in red for example, it means that it is not native, it’s likeli that red 255 is a simulation of sRGB/AdobeRGB red.
Ok, but how do i generate or measure this correction and spectral data in the first place for my specific display and i1? sorry if this is a stupid question, and thank you for being so patient! really appreciate it. i am not completely getting it
Uncalibrated report says that your white (on OSD, monitor standalone without computer or GPU calibration) is “not white” BY FAR using that correction.
Measurement reports that your contrast is totally destroyed, 3xx:1. It should be 800-1000:1 and uncalibrated is in that range, so your VCGT calibration is faulty.
Redo with suggested values, my previous post: D65 white (set on OSD), yu cant set brightness “as measured” and set manually 100cd/m2 on OSD (as measured just means DisplayCAL don’t care, whatever you put manually is fine),I was thinking setting this to 100 is a setting for interactive in the beginning, and that it doesnt really effect the profile or lut. Does it?
gamma 2.2 (uncalibrated repot say a mean value, but IDNK actual values for each grey, set a value 2.2 or close to actual mean-native value, maybe you can twaelk it on OSD), calibration speed low or medium (depending on grey issues on uncalibrated screen IDNK yours)
Plot that correction, i button next to it. Maybe it is a bad one. Avoid CCSS that store channels that look like linear combination of others… but some people find it difficult to spot visually. IDNK about you. A QLED should be 3 lonely peaks fro each channel. If you see some amount/peak of green in red for example, it means that it is not native, it’s likeli that red 255 is a simulation of sRGB/AdobeRGB red.
Ok, but how do i generate or measure this correction and spectral data in the first place for my specific display and i1? sorry if this is a stupid question, and thank you for being so patient! really appreciate it. i am not completely getting it
Somebody has to make it with an spectrophotometer and share it with you/us.
Her you have several, CCSS, IDNK / I do not remember which one is good. Download all (try first 3nm ones), plot with “i” button, if there are no signs of gamut simulation.. that one is good. Maybe it is the one you have, INDK, plot it.
Uncalibrated report says that your white (on OSD, monitor standalone without computer or GPU calibration) is “not white” BY FAR using that correction.
Measurement reports that your contrast is totally destroyed, 3xx:1. It should be 800-1000:1 and uncalibrated is in that range, so your VCGT calibration is faulty.
Redo with suggested values, my previous post: D65 white (set on OSD), yu cant set brightness “as measured” and set manually 100cd/m2 on OSD (as measured just means DisplayCAL don’t care, whatever you put manually is fine),I was thinking setting this to 100 is a setting for interactive in the beginning, and that it doesnt really effect the profile or lut. Does it?
Yes, it is a helper for interactive display. It is fine if you use it.
What I said is that since interactive display adjustment provides cd/m2 feedback you can set it to “as measured” and set manually on display to whatever value you want, including that 100cd/m2. You have visual feedback to see value, set manually the value you want. “It is not required to use it”.To Vincent: Is my PV270 correction usable here (both ColorMunki and non-NIST i1p2 should be avaible)? The PV270 seems to have the same IPS panel.
To Irfan: BenQ SW2700PT comes with Palette Master Elemets. This tool is limited and produces poor results cause of few pathces used, I’ve tried it with BenQ SW240 a month ago. DisplayCAL makes excelent calibration for SW240, but you can’t use native display superb gamut (SW240 does not have Rec.2020 setting to get it) and you can’t use your display for AdobeRGB or sRGB emulation (i.e. setting AdobeRGB as display ICC profile in system). This shouldn’t be a problem for modern software, but some hoary tools like 3ds Max are better with standard ICC emulation. To make colors limited to some standard gamut you should set it in display menu, then start calibration and profiling process. …Hm, try to download Palette Master (non-Elements) out of PV270 support pages. Does it recognize your display? Note: Argyll/DisplayCAL supports ICC v.2 format only, select it in other software to make 3D LUT later.
Note on BenQ: I’ve met SW240 and PV270 twice. Old ones (including my display) have non-homogenous panels, but new ones (2019-2020 mfg years) are pretty good. The most homogeous panel I’ve met is 4K LG 27UL850-W that is sRGB+ display (good for Rec.709) with internal LUT. Tomorrow I will calibrate the second unit and I will test its proprietary software. I’ve never met Eizo here (it’s a dream).
