- This topic has 17 replies, 6 voices, and was last updated 5 years, 8 months ago by
.
-
Posts
-
I don’t understand this statement, but after my post last night I thought this might be related to the display firmware and/or Icon Mode the user is using.
It means what spectral power distribution plots show.
All colors that your WOLED can show are (should be) a linear combination of native gamut. Becasue of that is useless to measure a dsiplay in an emulates gamut (an “artificially limited gamut because som kind of user/factory calibration”) like in Zoyd CCSS.
He made that CCSS in the “wrong way”. He should have measure with his JETI that WOLED TV in one of the widest gamut possible. That way his CCSS would be useful to any OSD from almost any current consumer WOLED.
“His” WOLED CCSS just works for WOLED displays with similar Rec709 to his… be it good factory Rec709 emulation or a bad one. Not good.
Do yo you see red line (Green channel)? This green channel is mabe up by mixing diffret weights of native blue, green and red in order to limit green coordinates to Rec709 green instead of near P3 red.
The one taken with a lesser grade instrument linke a Munki is better IMHO because it will work for all OSD modes from consumer WOLED:
If you know him and you are worried about helping community of consumer WLOED, then ask him to re-measure his WOLED in the right way: native gamut. I’ve no account in AVS.
Same goes for people that upload CCSS from monitors with emulate gamuts, like the guy who uploaded a QLED SW2700PT at AdobeRGB gamut, or the guys that upload W-LED PFS AdobeRGB99% displays like Dell’s UP 25″ & 27″ or newer Eizos CG.
With a little effort from “final user” (CCSS consumer) and some spreadsheet cooking that “wrong” CCSS could be converted to a native gamut CCSS… but it would be better if final users (who do not need to know these things IMHO) could just download and use CCSS without doube checking if the guy who uploaded those CCSS did it in the right way.
I mean their measurements (Zoyd & others) are accurate… but are not useful as it could have been.
I don’t understand this statement, but after my post last night I thought this might be related to the display firmware and/or Icon Mode the user is using.
It means what spectral power distribution plots show.
All colors that your WOLED can show are (should be) a linear combination of native gamut. Becasue of that is useless to measure a dsiplay in an emulates gamut (an “artificially limited gamut because som kind of user/factory calibration”) like in Zoyd CCSS.
He made that CCSS in the “wrong way”. He should have measure with his JETI that WOLED TV in one of the widest gamut possible. That way his CCSS would be useful to any OSD from almost any current consumer WOLED.
“His” WOLED CCSS just works for WOLED displays with similar Rec709 to his… be it good factory Rec709 emulation or a bad one. Not good.As far as my limited knowledge goes, WOLEDs are just white pixels with RGB subpixel filters in front of them where as LG’s WRGB OLEDs are seperate white, red, green, blue subpixels without filters. Both have 4 subpixels (W/R/G/B), but very different spectral response between the implementation.
Just to be clear, what do you mean when you say “WOLED”? Refering to the 4 subpixel types (both) or literally the WOLED type (the former which is NOT LG’s WRGB type)?
Regardless these types of OLEDs are not linear at all. You have to consider TWO different whites at the same time. One from the white subpixel, and the other from RGB subpixels. Both will be completely different, and changing one (RGB) changes the other (W). I think the W subpixel is likely something simple like voltage offsets from blue subpixel. On my E6, blue can reach insanely higher luminance levels than the others combined when all are pushed to the limit (over 1000 nits, others are ~700-750 nits).If you know him and you are worried about helping community of consumer WLOED, then ask him to re-measure his WOLED in the right way: native gamut. I’ve no account in AVS.
I take it you don’t recognize his name then. Zoyd is the current developer of HCFR, actual display calibration software. I don’t think he would have made a mistake, but hey we’re all human… Except Florian. No human could maintain software like DisplayCAL for so many years by themselves, the stress would kill anyone (not to mention making hair fall out).
Lol. 😛I sent him a message about this, but I think the file is fine and there is a good reason why it looks the way it does that you don’t know (I know I don’t know). I might simply ask in the HCFR thread later to see if anyone else has comments on this. I am curious as to why the curves have the same general shape but with a different color (I’m assuming this is a visual/report oddity and data is actually fine).
Another user, sillysally, made a ccss for his EF9500 (LG’s 2015 OLED) which was generally recommended back around 2016 as it was the same display panel type, which has a very similar response… I think. Not really sure how to interpret spectral response charts. I’ve attached it if anyone is curious. I don’t usually use correction files and such made by others, but perceptually it was much better than raw readings (I have no spectro).
. I’ve not seen any other reports of it saying otherwise, but I don’t monitor the 2017/2018 calibration threads at AVSforum.
Although I am interested in why the WRGB order is different as shown in the plots you attached. There is probably a reason for why the spectral correction file is the way it is, but I certainly don’t know.Same goes for people that upload CCSS from monitors with emulate gamuts, like the guy who uploaded a QLED SW2700PT at AdobeRGB gamut, or the guys that upload W-LED PFS AdobeRGB99% displays like Dell’s UP 25″ & 27″ or newer Eizos CG.
With a little effort from “final user” (CCSS consumer) and some spreadsheet cooking that “wrong” CCSS could be converted to a native gamut CCSS… but it would be better if final users (who do not need to know these things IMHO) could just download and use CCSS without doube checking if the guy who uploaded those CCSS did it in the right way.
I mean their measurements (Zoyd & others) are accurate… but are not useful as it could have been.
I don’t understand this statement, but after my post last night I thought this might be related to the display firmware and/or Icon Mode the user is using.
It means what spectral power distribution plots show.
All colors that your WOLED can show are (should be) a linear combination of native gamut. Becasue of that is useless to measure a dsiplay in an emulates gamut (an “artificially limited gamut because som kind of user/factory calibration”) like in Zoyd CCSS.
He made that CCSS in the “wrong way”. He should have measure with his JETI that WOLED TV in one of the widest gamut possible. That way his CCSS would be useful to any OSD from almost any current consumer WOLED.
“His” WOLED CCSS just works for WOLED displays with similar Rec709 to his… be it good factory Rec709 emulation or a bad one. Not good.As far as my limited knowledge goes, WOLEDs are just white pixels with RGB subpixel filters in front of them where as LG’s WRGB OLEDs are seperate white, red, green, blue subpixels without filters. Both have 4 subpixels (W/R/G/B), but very different spectral response between the implementation.
Just to be clear, what do you mean when you say “WOLED”? Refering to the 4 subpixel types (both) or literally the WOLED type (the former which is NOT LG’s WRGB type)?
Regardless these types of OLEDs are not linear at all. You have to consider TWO different whites at the same time. One from the white subpixel, and the other from RGB subpixels. Both will be completely different, and changing one (RGB) changes the other (W). I think the W subpixel is likely something simple like voltage offsets from blue subpixel. On my E6, blue can reach insanely higher luminance levels than the others combined when all are pushed to the limit (over 1000 nits, others are ~700-750 nits).It seems that you are not able to catch what is in plotted this two graphs:
Zoyd’s
DisplayCAL user:
If you were we would not be having this conversation again and again. But.. one last time.
In that WOLED spectral power distribution you can see that WHITE is not R+G+B but WOLED white.
In that WOLED SPD you can see that Zoyd’s green channel (plotted with another color beacause of order) is a linear combination of R, G and B from OLED. Why? Because that TV’s have a gamut bigger that Rec709.
Your particular factory calibration to Rec709 would be done with a LUT3D or lux matrix… and that calibration will store Rec709 calibration as a LINEAR COMBINATION of RGB native primaries. In a numeric way. I does not matter if OLED is not linear… firmware will store any kind of gamut emulation as this as “weights”, digital numbers to do a mixing.
So the CCSS from an annonymous DisplayCAL user with a lesser spectrofotometer like a Munki Photo is better than Zoyd’s.
If you know him and you are worried about helping community of consumer WLOED, then ask him to re-measure his WOLED in the right way: native gamut. I’ve no account in AVS.
I take it you don’t recognize his name then. Zoyd is the current developer of HCFR, actual display calibration software. I don’t think he would have made a mistake, but hey we’re all human… Except Florian. No human could maintain software like DisplayCAL for so many years by themselves, the stress would kill anyone (not to mention making hair fall out).
Lol.
I sent him a message about this, but I think the file is fine and there is a good reason why it looks the way it does that you don’t know (I know I don’t know). I might simply ask in the HCFR thread later to see if anyone else has comments on this. I am curious as to why the curves have the same general shape but with a different color (I’m assuming this is a visual/report oddity and data is actually fine).
It is wrong, or if you want another equivalent sentence… it is “worse” than the one made with a Munki Photo. Reasons explained above. It does not matter who made it.
Plots are spectral power distribution. X axis is nanometer wavelength and Y axis is “how much energy” (I think that they are mW/(m^2.sr.nm) but scaled).
VioletBlue in rainbow is on the left, red is on the right.OLED green SPD gives you CIE XYZ coordinates that are beyond sRGB/Rec709 gamut (2º plot)… that’s why you can see P3 content on an OLED display. If you want to show Rec709 content in that display, “green” should be sent to “panel” not as just green “0,255,0” (asume PC levels for simplification in explanation) but as (a, b,c) because Rec709 “green” is INSIDE OLED gamut.
That’s exactly what Zoyd’s CCSS captured, the linear combination of native R, G and B to make each of the Rec709’s emulated gamut R, G and B (and THESE ones are the ones stored in that CCSS).If he had taken native WOLED readings all of you will have an CCSS suiteable for ALL osd modes in your WOLED TVs: Rec709, P3, “photo” or whatever each manufacturer decided to include in that kind of TVs.
But he didn’t, so you are struck at Rec709 emulation with your particular i1d3 IF your TV’s factor calibration have similar weights for that RGB linear combination for each channel (IF your Rec709 factory calibration is as accurate has his).
That CCSS is not just “device dependent” (from Zoyd’s TV actual WOLED spectral behavior) but also “factory/user calibration dependent”. This last one (“factory/user calibration dependent”) is what makes that CCSS an unsuitable as “generic correction” (which is the purpose of a CCSS)… or if you want to put it in a more friendly sentence, that is what makes that CCSS a worse generic correction than 2nd one.
It’s a pity that with such anvanced measuremednt hardware he didn’t take SPD readings “in a good way” like that ColorMunki DisplayCAL user did.So that CCSS is made in a wrong way. And this also applies to other DisplayCAL’s user like the examples I wrote before.
I hope that now you understand why Zoyd’s CCSS is not as good as you thought.
PS: your attached CCSS is also “wrong” because it seems to have (order is not WRGB, so look at nanometer position in “blue-red” X axis, not actual line’s color)
-blue “limited” to a linear combination of native blue and red (very small amount of red)
-green “limited” to a linear combination of native blue and green.
-red “limited” to a linear combination of native blue and red (very small amount of blue)
So that CCSS does not just store that LG TV WOLED emited spectra but it is also dependent on A PARTICULAR CALIBRATION (user or factory made) with gamut emulation. Not very suitable for use as generic CCSS correction for “all users” with that LG model.Why? Because gamut accuracy from factory calibrations are at least “questionable”. For example frank’s Rec709 mode in his TV seems to have a blue that is slightly wrong placed because of factory calibration’s slightly wrong mixing of native R, G and B… and his TV’s gamut is able to do 100% Rec709… but Rec709 “factory calibration” failed to do that (by a very small amount)
And that’s is a very good reason to look for native gamut CCSS if you just have an i1d3. If they are not available… you may try some “cooking” in an spreadsheet by manually erasing the mixing.
Calibrite Display Pro HL on Amazon
Disclosure: As an Amazon Associate I earn from qualifying purchases.Good thread everyone, thanks for documenting your findings. I’m about to attempt a more accurate calibration for my B7 and was hoping anyone could perhaps point out if I’m making any mistakes.
I’ve assumed that colour gamut set to “wide” on a B7 is the TV’s native gamut, but I can’t even find a straight answer indicating this is actually correct. I suspect that the TV is staying locked to Rec709 emulation when set to “auto”, based on some banding I can see in test patterns.
Signal chain is: Decklink 4K SDI -> Blackmagic SDI to HDMI box with calibration LUT -> HDMI to LG B7. Producing content for the web and other mobile devices, so sticking with rec709 gamut + gamma 2.2 target.
- TV picture profile set to “ISF Expert (Bright Room)”, with all extra processing disabled
- OLED light 85, contrast 100, brightness 50, saturation 65
- Gamma 2.2, Colour Temp Warm2, Black Level low
- Resolve’s data levels set to “video” under project settings.
- Test pattern is one I found with black frames inserted, but can’t recall the original author. Filename is “Precon-rec709-OLED-A1.ti1” and attached.
It would be great to find some definitive answers on what the optimal configuration is here, I’m surprised it feels so much like some sort of black art.
Have we agreed on what CCSS is most appropriate for the B7 OLED and this sort of workflow?
What’s the correct colour gamut to select on the TV itself, Wide?
