Originally Posted by ebr9999
[email protected]ConnecTEDDD and all:
I have a B6:
- Not clear to me whether the suggested LG method for calibrating grey scale at 540 nits (resulting Oled light for me is 76), is only for calibration or is also for vision (i.e. 100 should be taken back after calibration phase).
- It looks to me that for calibrating colours (to the little extend that B6 allows), rec 2020/P3 should be preferred instead of rec 2020 (better colour space coverage). I understand it also a matter of choice. But what puzzles me is that those two have (slight) different white point, so somehow affecting the Gray scale measure.
PS: my assumption is that various grey scale patterns (e.g. Masciolla) are based on rec 2020.
Hi, when you calibrate for HDR10, you have to use patterns with metadata which will emulate how the LG's internal tone mapping will work when you will watch HDR10 movie content. So the 540nits entry the LG is providing (totally useless) it's not emulating any real movie, since the HDR10 movies have mastering display luminance 1000/1100/4000/10000 nits.
You have to choose between 1000 or 4000 nits (of mastering display peak) metadata patterns, not for 540 nits.
Also the OLED Light and Contrast (and brightness) has to stay to the default position (100/100) and not reduce OLED Light according to LG's instructions, the internal baked PQ tracking and tone mapping is based to these 2 default settings (work like bypass settings for internal mapping to work as expected), once you will reduce OLED Light you will see that your color errors will be increased a lot, when you will take gamut measurements.
Even the percentage numbers of LG 2016/2017 documents are not matching the digital levels of the Dolby's golden reference ST.2048 targets, so there is mismatch of 10bit values represent the percentages. I have checked all digital levels compared to Dolby's Golden Reference values and the percentages don't match the digital values (some are correct, only a few). LG's are 0.1% off.
If you reduce Contrast/OLED from LG, you will reduce the peak output, OLED/Contrast @ 100 provides better color tracking at these default values.
Using LG instructions for HDR calibration it will increase your color errors and provide lower peak output but better Grayscale and gamma tracking, but the color errors are more important from grayscale errors in that case.
After some testing performed using a 65E6 @ HDR mode, lowering the OLED Light, can improve the Grayscale but increase the color errors. To compare what is happening you have to take 5-Point Saturations (or 10-Point)
If you keep the Contrast/OLED @ 100, and calibrated the RGB levels for HDR, then you get an average Grayscale of about 1.6dE2000 and the errors are coming from gamma (brighter) at mid range while the RGB balance is near perfect....you get about 670nits calibrated.
If you run a 5-Point Saturation with targets DCI P3 inside a REC2020 (look CalMAN HDR workflow) then you get about ~2.0dE2000
If you keep the Contrast @ 100 and reduce OLED about 80? (I don't remember the exact value...to get 540nits) and calibrated the RGB levels for HDR, then you get an average Grayscale of about ~1.0dE2000 (or lower)and the errors of gamma are reduced at mid range while the RGB balance is near perfect....you get about 540nits calibrated.
If you run a 5-Point Saturation with targets DCI P3 inside a REC2020 (look CalMAN HDR workflow) then you get about ~4.0dE2000
About mastering display black/white levels the movies are sending to the LG, the E6 (EU) I have checked (before long time) with 50-Point Grayscace runs, sending any metadata (10000 nits or 4000 nits etc) it was not affecting anything the PQ tracking, I have checked about 60-80 different metadata, even different min (black) numbers from 1 nits up to 6.5nits, or every combination of MaxCLL and MaxFALL. (TV's tone mapping is not affected by any MaxCLL and MaxFALL value)
The only metadata was affecting the PQ tracking was the 540nits max display, which was changing a bit the RGB balance and PQ so I don't recommend anyone to calibrate using 540nits mastering peak becasue no movie has that metadata, with 1000nits it will be ok to use.(even if you send custom metadata with 650nits you will have the same output as 1000nits metadata)
2) About CMS in HDR, leave it untouched, only setting the correct colorspace setting is required, which is the normal gamut for 2016 models.
Here is the 5-Point Saturation of DCI-P3 inside a REC.2020:
DCI-P3 is not a consumer colorspace, so you should not aim for that colorspace when you calibrating any consumer device.
REC.2020 is the only colorspace you have to aim for HDR10 calibration, because displays can't cover that colorspace completely, we are using lower Saturation level (50%) and lower Stimulus Level (50%) patterns.
REC.2020 is being used as container, most of current UHD movies has been mastered using DCI-P3 primaries with REC.2020 D65 White Point calibrated mastering monitors, so colors beyond that gamut coverage are not being used (currently) to most of the movies.
Because most of the consumer HDR displays are not covering 100% the DCI-P3 primaries (inside REC.2020 container), this is why we are using REC.2020 with 50% Saturation patterns for HDR CMS, because xy of 50% Saturation REC.2020 is undersaturated from 100% Saturation of DCI Primaries (inside REC.2020 container), so it's a target that you can reach when the CMS controls of the display are working as expected.
You can use DCI-P3 Saturation tracking inside a REC.2020 container to evaluate the display tracking of DCI-P3 inside REC.2020 (where previously have calibrated the display with REC.2020 as target colorspace)