- This topic has 4 replies, 3 voices, and was last updated 1 year, 7 months ago by
.
-
Posts
-
Hi there,
I’ve been trying to calibrate my Samsung G8 (QD-OLED) monitor for Rec. 709 and have come away with a number of questions about the process. I’ll run through my setup and what I’ve done so far below, but firstly pardon my ignorance on the topic – this isn’t my wheelhouse and I’m just trying to piece together a ton of technical info from the internet =]
In terms of hardware I just have the ColorMunki Display, though if it might help I’d consider getting a spectro (hopefully something affordable like the Calibrite ColorChecker Studio).
My first attempt at calibration using a user-shared .ccss and observer 1931_2 threw up a huge red (green?) flag – white didn’t look white, it looked green. Reading through this forum suggested it might be metameric failure and that I could try 2012_2 instead. Switching to this observer fixed the green cast – white looked white – though it did look a little cool to my eye. I then recalibrated the display, targeting a warmer daylight temp, using dispcal to get the coordinates (dispcal -Q 2012_2 -X … -t 6100 -> 0.3198x 0.3360y). The resulting whitepoint looks pretty neutral to me – no cast, not too cool, not too warm. (Curiously DisplayCal’s hardcoded-to-1931_2-HTML report pegs this temp at around 6500K, though I don’t know how significant an observation that is.) Seems all is well but I have a number of questions:Firstly my colorimeter. Is it suitable for use with observer 2012_2? I’ve read conflicting takes on this on the DisplayCal forums (namely this thread).
I’m also a bit confused as to which .ccss to use. Most of the shared .ccss files seem to capture this monitor running under its native color space, but I noticed one user uploaded a .ccss for the AW3423DW and AW3423DWF (same panel as the G8) running in sRGB mode. I’ve attached two examples, but in short the red spike in ‘sRGB’ .ccss is a mix of red and green, whereas that’s not the case for the ‘native’ .ccss. Should I be using the ‘sRGB’ .ccss when running the G8 in Rec. 709 SDR mode? And how accurate is it likely to be given it’s technically for another monitor which may implement its sRGB clamp differently to the G8?
And if I want to cut out the guess work and make a colorimeter correction myself, would buying something like the Calibrite ColorChecker Studio give me more accurate results than using a shared .ccss?
I also noticed ccxxmake takes an observer arg (-o observ) and that the .ti3 files accompanying the shared .ccss files all seem to contain the OBSERVER “1931_2” tag. If I’m using 2012_2 in my calibration, should the colorimeter corrections be using it as well?Thanks for your help
Calibrite Display SL on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.Hi there,
I’ve been trying to calibrate my Samsung G8 (QD-OLED) monitor for Rec. 709 and have come away with a number of questions about the process. I’ll run through my setup and what I’ve done so far below, but firstly pardon my ignorance on the topic – this isn’t my wheelhouse and I’m just trying to piece together a ton of technical info from the internet =]
In terms of hardware I just have the ColorMunki Display, though if it might help I’d consider getting a spectro (hopefully something affordable like the Calibrite ColorChecker Studio).
My first attempt at calibration using a user-shared .ccss and observer 1931_2 threw up a huge red (green?) flag – white didn’t look white, it looked green. Reading through this forum suggested it might be metameric failure and that I could try 2012_2 instead. Switching to this observer fixed the green cast – white looked white – though it did look a little cool to my eye. I then recalibrated the display, targeting a warmer daylight temp, using dispcal to get the coordinates (dispcal -Q 2012_2 -X … -t 6100 -> 0.3198x 0.3360y). The resulting whitepoint looks pretty neutral to me – no cast, not too cool, not too warm. (Curiously DisplayCal’s hardcoded-to-1931_2-HTML report pegs this temp at around 6500K, though I don’t know how significant an observation that is.) Seems all is well but I have a number of questions:Firstly my colorimeter. Is it suitable for use with observer 2012_2? I’ve read conflicting takes on this on the DisplayCal forums (namely this thread).
Choose the whitepoint that looks white. Once you have that calibrate & profile to a white point with those coordinates (1931 2degre). That’s all. It won’t change your visually validated white.
When making a LUT3D use relative colorimetric (recomended for newbies, it’s the easiest) or if want abs colorimetric then you need to simulate an alternative version of rec709 with that alternative whietpoint.I’m also a bit confused as to which .ccss to use. Most of the shared .ccss files seem to capture this monitor running under its native color space, but I noticed one user uploaded a .ccss for the AW3423DW and AW3423DWF (same panel as the G8) running in sRGB mode. I’ve attached two examples, but in short the red spike in ‘sRGB’ .ccss is a mix of red and green, whereas that’s not the case for the ‘native’ .ccss. Should I be using the ‘sRGB’ .ccss when running the G8 in Rec. 709 SDR mode? And how accurate is it likely to be given it’s technically for another monitor which may implement its sRGB clamp differently to the G8?
I’ll say the native gamut with highest resolution (1, 3nm) but there is a thread in AVS forum about that with other opinions.
And if I want to cut out the guess work and make a colorimeter correction myself, would buying something like the Calibrite ColorChecker Studio give me more accurate results than using a shared .ccss?
Maybe, but I would buy that.
I also noticed ccxxmake takes an observer arg (-o observ) and that the .ti3 files accompanying the shared .ccss files all seem to contain the OBSERVER “1931_2” tag. If I’m using 2012_2 in my calibration, should the colorimeter corrections be using it as well?
Thanks for your help
I’ve not tested it, maybe other person can answer you.
For me it would be easier to configure 3nm in that i1Studio, choose whatever oberser, and calibrate display to whetever whgitepoint you want, let’s say 2012_2_D65. If there is a huge diference between that white and what I see as white, fine tune visually.
Then create CCSS. Measure white with you i1d3 + CCSS 1931_2. Then set thos ecoordinates as taregt whitepoint, Calibrate & profile and same warnings when making LUT3Ds.Thanks for the advice Vincent. I think it’s worth getting the spectro given how time I spend looking at screens all day…
When making a LUT3D use relative colorimetric
Good point, I forgot to mention rendering intent. Relative colorimetric is certainly the easiest to enable in the apps I use (a small line of text in mpv for example).
but there is a thread in AVS forum about that with other opinions.
I’ll wander over and see if I can find any more info.
Cheers
My first attempt at calibration using a user-shared .ccss and observer 1931_2 threw up a huge red (green?) flag – white didn’t look white, it looked green. Reading through this forum suggested it might be metameric failure
Seems normal https://youtu.be/t22CvXeTRdA?t=1086
I observed the same problem, D65 measured with an i1 pro 2 does look too greenish compared to a D65 adjusted wled lcd.
But according to an avsforum calibrator equipped with a <1nm spectro white on a D65 adjusted qd oled looks similiar to a D65 white adjusted sony BVM hx mastering monitor.
https://www.avsforum.com/posts/63610004/
“I’m simply providing accurate information as my results are different with my meter that has a wavelength resolution of 0.9nm. Colourists also recognise the QD-OLED as a direct match with a BVM, and they are the ones creating the movie/TV content you’re watching.”
vincent, can you confirm this?
The greenish white of my qd-oled tv is unbearable for me, but if that equals the white of the mastering monitor used by many studios it would be the “right one”.
Do measurements changein RGB balance when you compare 10nm vs 3nm?
On White LED (regular sRGB screens) they almost mach, but on GB-LED @10nm measurements was moved to yellow vs i1d3 ccss corected or i1pro2 3nm with argyll driver ( = visually looked white because it was still in daylight curve)
I do not have access to a QDOLED to compare.
