- This topic has 19 replies, 3 voices, and was last updated 5 years, 5 months ago by
.
-
Posts
-
Hello Florian & Others.
My configuration is the following :
- Macbook Pro mid 2012
- Eizo CS2420
- DeckLink Mini Monitor
- Probe : I1d3 or i1Display Pro
A few questions for starters : i have been trough many subjects on which you answered, i read the wiki, and still, i’m wondering which kind of correction (matrix) i should use ?
I have downloaded one from the Argll database, the last one from the link down below, but i’m not quite sure that i should be using this one.
Vincent recommended LCD RG Phospor, there is this new categorie LCD CCFL IPS (Eizo)… also, i don’t have access to a spectrometer, only colorimeter.
So, here is my question : this is not the first time i’m using display cal : i would like to be able to work on different field :
- Web delivery : color grading for vimeo, youtube, to be viewed on computer and phone displays.
- Instagram & Photo delivery
- Broadcast compliant delivery : BT 1886 / Rec 709
I’d would love to set my EIzo to be as accurate as possible, in a dark dimmed environnement, and my macbook screen to be as close as it can reasonnably be.
From my understanding of the way DisplayCal work, the best way to achieve this would be :
- Calibrate and profile my Eizo using DisplayCal, with color navigator off, trough the resolve preset :
- white level drift off / black drift off / output level on auto / color correction (up to what you’ll answer)
- Interactive display adjustement : CIE 1931 / 6500 K / 100 or 120 cd/m2 / Black Level as mesured / sRGB Gamma, Tone Curve
- Profiling on auto optimized 1553 patches
- 3D Lut REC 709 / Gamma 2.2 / absolute colorimetric with withe point scaling / full range input output
Trough these, i could work on my Resolve trough a decklink on regular srgb photography, but also on content for vimeo by loading the corresponding 3D LUT on resolve
Then, i would go back to display cal, uncheck interactive display and put everything to as measured (accept that the 3D Lut will use the measured white point as well), go back to 3D LUT :
- Same correction as before (won’t matter am i right ?)
- everything as measured / no interactive
- REC 709 / Rec 1886 / Relative colorimetric (?) / Full range input output
And then, i’ll load the 3D LUT when i want to work on Broadcast compliant project..
For the macbook screen, i’ll try to match the Eizo screen by using the laptop preset..
Would that be a logic way of thinking ? I won’t have to mess with color navigator, i won’t have to mess with the ICC profile once done (one for the macbook screen, nothing really for the EIZO (thanks to the resolve decklink) except the one i’ll do with colornavigator for using photoshop or watching video files, not sure about this), i won’t have to mess with the hardware settings on the eizo once done, the only thing that i will have to set up would be the 3D LUT in Resolve..
So, could i make it work with a screen calibrated to display srgb color and gamma and using a 3D LUT in resolve for Broadcast BT 1886 content ?
Anyway, thanks for all the thoughts behind the software. Take care,
Matthieu
Calibrite Display Pro HL on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.Hi,
Vincent recommended LCD RG Phospor, there is this new categorie LCD CCFL IPS (Eizo)… also, i don’t have access to a spectrometer, only colorimeter.
LCD RG Phosphor is the correct choice for the CS2420. LCD CCFL IPS is a different type altogether (CCFL backlight).
Trough these, i could work on my Resolve trough a decklink on regular srgb photography, but also on content for vimeo by loading the corresponding 3D LUT on resolve
So, repeating in my own words to make sure I got your intent, you want to:
- Use the Eizo exclusively through Resolve via decklink (not part of desktop)
- You want an sRGB 3D LUT to edit sRGB material (photos),
- and also a BT.1886 3D LUT to edit video footage
The two 3D LUTs will probably be very similar, because on a (roughly) 1000:1 contrast display, sRGB and BT.1886 match very closely.
- So, my first suggestion would be to create only one 3D LUT, the BT.1886 one, to be used for both photo and video editing (alternatively, use the standalone 3D LUT maker with a sRGB source profile to create a “true” sRGB 3D LUT after creating the BT.1886 one, no need for additional measurements – but as said, I would expect the visual difference to be tiny).
- Don’t set a white level target, instead, set brightness using monitor controls during interactive adjustment (ignore the white bar, just look at the numbers).
- Don’t set a calibration tone curve (use “As measured”), the 3D LUT tone curve will override this anyway.
For the macbook screen, i’ll try to match the Eizo screen by using the laptop preset.
Sounds reasonable. You can use the visual whitepoint editor to help match the white (which is crucial for a good visual match).
Hello Florian,
Thanks for the answer. Is there a way to send a few bucks to help ? I’ll gladly send a little something in return. You’ll see 2/3 questions inside the post for you. This will be made as a little guide for helping those that will come after me asking always the same things as i did.
Anyway :
Thanks for the answer. I tried yesterday the following, it would be a sort of guide for those wanting to do the same things, i tried sending a few images back to my cellphone, friends laptop and displays, they all look pretty consistent, and the report and verifications were good, backed by the approval of Florian and others on most of the process.
Erase all of previous custom personal profiles from the macbook and the EIZO (the one you created before) and hard reset on the EIZO.
Also, deactivate completely ColorNavigator
- Calibrate my macbook screen first : preset laptop 2.2 – Correction used : AppleMacbookPro 9,1 i1displaypro CCMX from Argyll community BUT should have used AppleMacbookPro Antiglare display MD104 CCSS from the Argyll community.
- Interactive display checked with 6500 K as a whitepoint and 100 cd/m2 as target – Tone Curve Gamma 2.2 first, but, once the good nits reached, closed the calibration, then launched back with Interactive display UNCHECKED, 6500 K, 100cd/m2 and Gamma 2.2
- Profiling AutoOptimized 1553 Patches / High Profile Quality
- Verification WITHOUT use simulation as display profile
- You would have a hard time having the same image on the RESOLVE GUI and the RESOLVE trough DeckLink monitor for different reasons BUT : you can check use mac display color in RESOLVE to help OR create a GUI 3D lut following Florian advices OR simply hope that you laptop screen is decent enough to, when you export footage or stills from RESOLVE to your laptop, and playback the footage, RESOLVE opened againt your laptop, to see the same thing almost. Not the case for me, old macbook display. Florian, is there any way to improve instead of the visual white point editor ? (difference seems to be in contrast)
THEN
EIZO Screen : Custom profile on the EIZO screen, all profil reset, THEN : in SIGNAL : RGB / Input Range on LIMITED / Noise reduction OFF
In Color Mode Custom : Color Gamut Native / Gamma 2.2
Then, go back to displayCal :
- Preset Resolve D65 Rec 709 / Correction used : https://colorimetercorrectio ns.displaycal.net/?get&type=ccss&manufacturer_id=ENC&display=CS2420&instrument=i1%20DisplayPro%2C%20ColorMunki%20Display%2C%20Spyder4&html=1
- Last one from this list BUT SHOULD HAVE USED : LCD RG Phosphor is the correct choice for the CS2420. ((LCD CCFL IPS is a different type altogether (CCFL backlight).)) Florian, does that make a big difference or not ? Should i start again because of that ?
- White Drift uncheked / black drift unchecked / no advanced options (not needed)
- Calibration set on Interactive display : color temperature 6500 K / White level 100 cd/m2 / Tone Curve to ‘as measured’
- Alternatively, you can follow Florian advice AND instead : Color Temp to 6500K and let white level target to as measured and adjust trough the numbers (100 cd/m2 still…)
- Profiling to High : 1553 patches
- 3D LUT : Create 3D LUT after profiling then REC 709 ITU R / Gamma 2.2 / Absolute colo with white point scaling
- Verification with use simulation profile as display CHECKED
- Calibrate
- Then, go back to 3D LUT : uncheck “Create 3D LUT after profiling” and only change Gamma 2.2 to REC 1886 in 3D LUT then click create 3D LUT
- Ignore the prompt about vcgt
- Alternatively acoording to Florian, you can ignore creating the 2.2 gamma 3D LUT and directly create a REC 1886 3D LUT
- Use the 3D LUT created accordingly in RESOLVE
From my experience, using the different 3D Lut on my EIZO set as i did help a fair bit : the black would be a little bit too crushed on the web by using 1886 3D LUT (visually, using the 3D LUT 1886 give you lifted shadows, as it was intended by the people behind it, resulting in crushed black after the export of course). I hope that you can give this long post a look Florian, to help erase any doubts and that it will help others.
Best,
Matthieu
From my experience, using the different 3D Lut on my EIZO set as i did help a fair bit : the black would be a little bit too crushed on the web by using 1886 3D LUT (visually, using the 3D LUT 1886 give you lifted shadows, as it was intended by the people behind it, resulting in crushed black after the export of course). I hope that you can give this long post a look Florian, to help erase any doubts and that it will help others.
Best,
Matthieu
From my experience, I would avoid the 1886 curve and choose gamma 2.4, and then change 2.4 to 2.2. This should turn “absolute” to “relative” and turn off the lifted blacks.
I’ve just run a DCP test from my display set to gamma 2.2. On the big calibrated screen at the post house, all looked correct. If I had used 1886, then I would have crushed my blacks to overcome the lift. 1886 curve looks different on displays having different black levels. It’s not good for conversion to cinema where the software can’t account for your display type. For web distribution it might be kind of ok, but I think the implementation in DisplayCal is a little bit too aggressive. You can try to tame it with the provided slider, but it’s difficult to be subtle I’ve found with the slider control.
Not the case for me, old macbook display. Florian, is there any way to improve instead of the visual white point editor ? (difference seems to be in contrast)
As Bruce has pointed out, when using BT.1886 on different displays, black level should be equalized (to the one of the lesser display) so that contrast is the same (EBU recommends a below 0.05 cd/m2 black level for grade 1 monitors used for evaluation of images, but most graphics arts computer monitors cannot achieve that).
LCD RG Phosphor is the correct choice for the CS2420. ((LCD CCFL IPS is a different type altogether (CCFL backlight).)) Florian, does that make a big difference or not ?
Probably, as the spectra are completely different (CCFL tended to be spikier).
I don’t use BT.1886 on different displays, i created a 3D LUT only for my EIZO, for Resolve.
My macbook only used a 2.2 icc profile, and i checked use mac display color in Resolve.
I did not used the CCFL as correction, i used this one :
instead of the Phosphor one ? i’ll use the right one next time, but, is that a real problem until the next calibration ?
All right Bruce, i’ll think about it for my calibration to come.
I don’t use BT.1886 on different displays, i created a 3D LUT only for my EIZO, for Resolve.
My macbook only used a 2.2 icc profile, and i checked use mac display color in Resolve.
If you don’t use a 3D LUT with your MacBook as well, Resolve will do its own color management (set under “Timeline Color Space” in project settings, note that there is no BT.1886 option, and note that Resolve does not support LUT profiles).
https://colorimetercorrectio ns.displaycal.net/hash/aafcd9bebae57317fcf7 088df53b276b/Eizo%20CS2420%20%28ColorMunki%29.ccss
That’s probably fine, just keep in mind the caveats that apply to user-provided corrections.
I don’t use BT.1886 on different displays, i created a 3D LUT only for my EIZO, for Resolve.
My macbook only used a 2.2 icc profile, and i checked use mac display color in Resolve.
If you don’t use a 3D LUT with your MacBook as well, Resolve will do its own color management (set under “Timeline Color Space” in project settings, note that there is no BT.1886 option, and note that Resolve does not support LUT profiles).
https://colorimetercorrectio ns.displaycal.net/hash/aafcd9bebae57317fcf7 088df53b276b/Eizo%20CS2420%20%28ColorMunki%29.ccss
That’s probably fine, just keep in mind the caveats that apply to user-provided corrections.
I don’t understand this discussion. Maybe a case of ambiguous language writing here…
For your Eizo, connected through your decklink card, in Resolve settings, you need to, in “settings/color management” set your output video display to use the 3d LUT you created with displaycal. This will have no effect on your MBP screen. To avoid all confusion, all the other video outputs and scopes should be set to “no LUT selected”. If not, you will find inaccurate scopes when you color grade.
To keep things simple in Resolve, don’t use, in preferences “use mac display profile” (or what ever it’s actually called). I think there is something wrong with this implementation and it’s best to leave it off. Treat Resolve as if it’s a non-color managed application where everything must be set manually. Now, this has little to do with “Resolve color management” or ACES grading modes. For these, I strongly suggest reading the Resolve manual, taking a course, or other education about them.
If you really desire to see a proper image in Resolve, on your MBP screen, I suggest making a separate 3D LUT for your MBP screen. In Resolve “settings/color management”, you can now assign this 2nd 3D LUT to effect only your GUI display. But, this can be tricky, as any .icc profile set in your OS will have an effect on making this 3D LUT. If your display, OS level, .icc profile gets changed from what it was when you created the MBP 3D LUT, then your MBP screen will no longer be accurate. It is for these reasons that the Resolve manual strongly suggests only using the output through the decklink device to an external display for color evaluation. So, it’s best not to worry about the appearance of color on your MBP screen when color grading and always rely on your external display, connected through your “mini monitor” device, for color evaluation.
And lastly, a MBP for using Resolve is already on the edge of not enough “horsepower” to run resolve. Adding a 2nd 3D LUT for viewing on your MBP will only slow things down further. I don’t really recommend it.
For your Eizo, connected through your decklink card, in Resolve settings, you need to, in “settings/color management” set your output video display to use the 3d LUT you created with displaycal. This will have no effect on your MBP screen. To avoid all confusion, all the other video outputs and scopes should be set to “no LUT selected”. If not, you will find inaccurate scopes when you color grade.
That is exactly what i did, i select the 3D LUT 2.2 or 1886 depending on my need.
And lastly, a MBP for using Resolve is already on the edge of not enough “horsepower” to run resolve. Adding a 2nd 3D LUT for viewing on your MBP will only slow things down further. I don’t really recommend it.
That is the reason why i did not make the 3D lut for my macbook screen. First of all, i calibrated my screen trough the display cal laptop preset, thus not allowing me to get a 3D LUT.
I don’t think that the implementation is that wrong.. actually, my macbook screen is closer when checking the option because my .icc profile is created trough display cal on the same parameter. But, i would have needed to create others 3D LUT for 2.2 and 1886 for my macbook screen, eating power from the macbook, so i decided not to do it, and only use the option “use mac display” whatever it called to not be disturbed by the gui being a little too different.
Florian, i did not understood your answer, i must admit. I do use a 3D LUT, for my Eizo screen only, but not for the GUI viewer, i rely solely on my eizo.
I did some test for the color management settings, and for grading Adobe RGB material and export to SRGB, the RESOLVE YRGB option seemed the best idea, because with RCM, i can’t point out to resolve that my footage are in Adobe RGB (because resolve does not have that option, Baselight does, but i don’t have the aja card). If i could have used RCM, i would have put Input in Adobe RGB, timeline in ACESCCT (for the grade ‘feeling’) and output to Srgb.
And of course, for video footage on a different set up.
So Bruce, not a fan of 1886 ?
Best, Mat
So Bruce, not a fan of 1886 ?
No 🙂
It wouldn’t be needed if every display were the same contrast ratio/black level. Since it effects every display differently, I find that problematic. Look at the calibration report of your display using 1886 black level compensation. What is the gamma of your display?
2.4 in the highlights, but 2.2 overall, but something else in the blacks…
A DCI cinema projection doesn’t have “zero” blacks either, but the standard is 2.6, power law gamma. Not some different adjustment depending on the projector/venue combination.
I don’t have the report, because i created the 3D lut after creating the 3D LUT in 2.2, by unchecking “create 3D lut after profiling”.
My eizo is set on 2.2 on the screen itself.
The curve on 2.2 seems to be on point, except for the extreme right of the curve.
I do understand why you might prefer 2.4. But why is it adopted by so many post houses ?
I don’t have the report, because i created the 3D lut after creating the 3D LUT in 2.2, by unchecking “create 3D lut after profiling”.
My eizo is set on 2.2 on the screen itself.
The curve on 2.2 seems to be on point, except for the extreme right of the curve.
I do understand why you might prefer 2.4. But why is it adopted by so many post houses ?
You can still run a report after you’ve made the LUT. But, how to do it, is a bit confusing.
I don’t personally prefer 2.4 over 2.2. But, I’m grading for cinema release, so I only need to know the gamma of my rendered images for conversion to DCI/P3 gamma 2.6 release. As long as I know what it is, it doesn’t make a difference at all.
Personally, I’ve found gamma 2.2 to be a better match for online release.
The TV release has been a mushy subject for the past few years. And, the 1886 curve has made it more mushy as one sets the software (in this case, displaycal) to gamma 2.4 and 1886 to create the 1886 LUT. But, when you look at the report from the calibration, you will see that the 1886 calibration is mostly…. gamma 2.2! That is, unless your display has a zero black level, in which case it will indeed be gamma 2.4 overall. How’s that for mushy and confusing!!!????
I only have one TV at home (a plasma with very near zero black) that is set to gamma 2.4 in the factory menu. And it looks pretty good for almost all TV broadcasts. But, if I grade to gamma 2.4 on my Eizo, with it’s lifted blacks, the same material will be a little bit too light on my plasma. Grading to 2.2 is closer, but maybe 2.3 would be better.
That said, most home TV’s are LCD with lifted blacks, which are closer to my Eizo. So which to choose? I don’t know. And what is the typical home TV set to for gamma? It’s a mystery as most people don’t even know what gamma is, or how to use the TV menu.
It’s the wild west out there I tell you! 🙂
And… beware of a lot of misinformed discussion online about this very subject.
The only way to solve this would be to standardize all displays, to the same black and white levels and colorspace… and that won’t happen in my lifetime I’m afraid.
see Steve Shaw’s response in this thread:
https://www.avsforum.com/forum/139-display-calibration/2649073-rec-709-vs-1886-a.html
To keep things simple in Resolve, don’t use, in preferences “use mac display profile” (or what ever it’s actually called). I think there is something wrong with this implementation and it’s best to leave it off.
The Resolve ICC color management implementation is flawed in the sense that it does not support LUT-based profiles (same bug exists in macOS itself, affecting applications that come with the OS, like Preview and Quick Look. Third party applications that implement their own color management are not affected, which makes me think that Resolve is probably relying on the OS to do the color management in that case). I too would recommend not relying on it much, but as long as you’re aware of its limitations, it might be better than nothing.
Since it effects every display differently, I find that problematic.
Dominic Chan explains some of the reasoning behind that design in the thread you linked, and effects of different approaches.
A DCI cinema projection doesn’t have “zero” blacks either, but the standard is 2.6, power law gamma. Not some different adjustment depending on the projector/venue combination.
Yes, that standard is somewhat less sophisticated (probably because there is less variation expected among cinema projectors than home TVs), but it does not mean that (cinema) projectors follow a power law gamma 2.6 function near black perfectly. I think in reality, it’s probably much closer to a roughly 2.2-ish gamma close to black.
I do use a 3D LUT, for my Eizo screen only, but not for the GUI viewer, i rely solely on my eizo.
That’s what I meant. I.e. it is expected that Resolve color management when enabling use of the display ICC profile won’t match a Rec. 1886 3D LUT (or a 3D LUT using anything other than what is set under Resolve timeline colorspace), because Resolve doesn’t offer more sophisticated transfer function choices. If you wanted to have the same result (as far as possible) on the MacBook, you’d need to use a GUI 3D LUT with Resolve (which you have decided against, which is fine).
But, when you look at the report from the calibration, you will see that the 1886 calibration is mostly…. gamma 2.2! That is, unless your display has a zero black level, in which case it will indeed be gamma 2.4 overall. How’s that for mushy and confusing!!!????
It’s not really all that confusing. You only set the exponent for the transfer function, this is only loosely related to the effective gamma you’re getting on screen (and “gamma” is often [mis-]used as an oversimplified term for transfer functions generally). It is the same with the Rec. 709 or sRGB functions, which both use a linear segment near black and exponents of 2.22 and 2.4 respectively (on the decoding side), yet, their effective “gamma” isn’t anywhere near those values (well, sRGB does come close at overall roughly 2.2).
That said, most home TV’s are LCD with lifted blacks, which are closer to my Eizo.
Most home TVs have a black level that is an order of magnitude lower than that of the Eizo (or any other IPS-panel-based display).
So which to choose? I don’t know. And what is the typical home TV set to for gamma?
But you don’t grade to individual unknown home TVs. You grade for a standard, and then hope for the best. That is literally the only thing you can reasonably do, and it allows those that actually care about color to come very close to what you saw on your grading monitor (and even those who are unaware of it will still get a pleasing picture on most home TVs, as their display characteristic usually doesn’t differ too much from a roughly 2.2.-2.4 “gamma” transfer function – HDR is a different matter though).
To keep things simple in Resolve, don’t use, in preferences “use mac display profile” (or what ever it’s actually called). I think there is something wrong with this implementation and it’s best to leave it off.
The Resolve ICC color management implementation is flawed in the sense that it does not support LUT-based profiles (same bug exists in macOS itself, affecting applications that come with the OS, like Preview and Quick Look. Third party applications that implement their own color management are not affected, which makes me think that Resolve is probably relying on the OS to do the color management in that case). I too would recommend not relying on it much, but as long as you’re aware of its limitations, it might be better than nothing.
Since it effects every display differently, I find that problematic.
Dominic Chan explains some of the reasoning behind that design in the thread you linked, and effects of different approaches.
I think I understand the argument here, I just don’t agree with the results 🙂
A DCI cinema projection doesn’t have “zero” blacks either, but the standard is 2.6, power law gamma. Not some different adjustment depending on the projector/venue combination.
Yes, that standard is somewhat less sophisticated (probably because there is less variation expected among cinema projectors than home TVs), but it does not mean that (cinema) projectors follow a power law gamma 2.6 function near black perfectly. I think in reality, it’s probably much closer to a roughly 2.2-ish gamma close to black.
I do use a 3D LUT, for my Eizo screen only, but not for the GUI viewer, i rely solely on my eizo.
That’s what I meant. I.e. it is expected that Resolve color management when enabling use of the display ICC profile won’t match a Rec. 1886 3D LUT (or a 3D LUT using anything other than what is set under Resolve timeline colorspace), because Resolve doesn’t offer more sophisticated transfer function choices. If you wanted to have the same result (as far as possible) on the MacBook, you’d need to use a GUI 3D LUT with Resolve (which you have decided against, which is fine).
In Resolve, “Resolve Color Management” refers to an in app workflow that is separate from OS color management. It’s confusing that they’ve used this term as it has little to do with OS color management and does not communicate with the OS color management at all. So, it would be best not to use this term when discussing OS /Resolve color management, as there really isn’t any in Resolve.
But, when you look at the report from the calibration, you will see that the 1886 calibration is mostly…. gamma 2.2! That is, unless your display has a zero black level, in which case it will indeed be gamma 2.4 overall. How’s that for mushy and confusing!!!????
It’s not really all that confusing. You only set the exponent for the transfer function, this is only loosely related to the effective gamma you’re getting on screen (and “gamma” is often [mis-]used as an oversimplified term for transfer functions generally). It is the same with the Rec. 709 or sRGB functions, which both use a linear segment near black and exponents of 2.22 and 2.4 respectively (on the decoding side), yet, their effective “gamma” isn’t anywhere near those values (well, sRGB does come close at overall roughly 2.2).
I’m just going by the gamma section of the calibration report in DisplayCal here. Maybe you can write a more clear explanation of this Florian? I know this stuff is not so easy to explain 🙂
That said, most home TV’s are LCD with lifted blacks, which are closer to my Eizo.
Most home TVs have a black level that is an order of magnitude lower than that of the Eizo (or any other IPS-panel-based display).
Don’t most home TVs use IPS panels? (Excluding plasma or OLED)
So which to choose? I don’t know. And what is the typical home TV set to for gamma?
But you don’t grade to individual unknown home TVs. You grade for a standard, and then hope for the best. That is literally the only thing you can reasonably do, and it allows those that actually care about color to come very close to what you saw on your grading monitor (and even those who are unaware of it will still get a pleasing picture on most home TVs, as their display characteristic usually doesn’t differ too much from a roughly 2.2.-2.4 “gamma” transfer function – HDR is a different matter though).
2.4 to 2.2 is a significant difference though. And the black level compensation of 1886 is quite different as well. So, my point is, that it’s hard to really pin down THE standard here. I guess the closest would be a display with near zero black (OLED) and gamma 2.4 at 100nits, with a 5% surround or bias light surround, as that will look the same in either 1886 or 2.4 power law gamma, no?
Ironically, my Eizo, with it’s lifted blacks, very much resembles cinema projection, but an OLED would look different with it’s much higher contrast ratio. Of course it is best to grade for cinema on a DCI projector, but this is not always possible for cost reasons.
