- This topic has 23 replies, 6 voices, and was last updated 1 year, 1 month ago by
.
-
Posts
-
There is many monitors reviewers that use sRGB as target and not BT1886. Is there a reason behind this ? Or should they target the BT1886 EOTF instead ?
There is an sRGB gamut boundaries and sRGB TRC. Also I see no reason to use Rec1886 on 1000:1 displays.
Yeah I noticed that a 1000:1 display calibrated to BT1886 is more or less sRGB curve. But for a VA monitor, like my primary monitor, is BT1886 objectively better ? Rtings uses sRGB for all monitors even for OLED and I think it’s wrong but I could have missed something. And also for TVs they always target gamma 2.2 where I think it’s wrong too. Could my 10 years into calibration confirm my feelings about these test methologies ?
BT1886 has higher blacks and looks less contrast on a 1000 contrast display. It is a challenge for good contrast and good blacks on 1000 contrast display. I like native contast on my Vizio V 5-555-J01 tv. HCFR will show you the native contrast. If you have 10 or 20 point white balance you can get good white balance without crushing black by not turning down 5,10,15 white balance rgb. You need to use a meter and have a well behaved screen. Check the 1% black. On mine it is raised compared to the 2%. It probably done on purpose at the factory. I see 1% and less on native gamma 2.2 but on displaycal gamma 2.2 it is very hard to see 1% or less. http://www.lagom.nl/lcd-test/black.php . Difference is 1st row is dark vs 1 dark only. On native I can switch to 2.4 gamma and see in a dark room. The 1st row is very dark on 2.4 2.4 gamma is like bt1886 on a oled screen.
SRGB is not wrong. It is has less raised blacks than bt1886 on the 1000 contrast displays. Oled can have black crush worse and raised blacks is something you can see on them versus black crush. 2.2 gamma is good for a lighted room. People watch tv in a lighted room sometimes.
A factor is also the nit brightness. 2.4 is better at higher nits.
In fact I calibrate monitors to BT1886 for quite a long time so in fact they don’t calibrate with the good target. In fact for a well lit room changing the black levels is better cause it matches more cones’ threshold. Like making a SDR version of Dolby IQ or HGiG. There is informations there.
Rtings uses sRGB for all monitors even for OLED and I think it’s wrong but I could have missed something. And also for TVs they always target gamma 2.2 where I think it’s wrong too. Could my 10 years into calibration confirm my feelings about these test methologies ?
Rtings hardly knows what they are doing. The post calibration image is like using on DisplayCAL simulate sRGB on a custrom matrix + 3curves profile with no grey calibration at all.
Mostly useless as a source of computer display reviews.Best use pard.de but they review less displays.Download one of their custom ICC profiles and inspect them with ICC profile isnoecto or “DisplayCAl-profile-info.exe path_to_icc\rtings.icc”
Just my memory that gamma 2.2 2% hcfr input offset matched my native 2.2 gamma. But displaycal has been make luts starting at 145-170 instead of native 256 and around point 40 of 255 they start to match very closely. It makes my blacks crushed visualy. Well my screen is wierd and the red and green cross from 1 to 40. Fixed with raising green at 5% to match red almost. In the end displaycal is better. It is just harder see. however each number in lagom lcd test a different brightness visually and can see on contrast the white is not blueish at 1 , 2 and 3 in the contrast chart and is smoother increase in brightness. Still get errors like 2024-05-18 23:19:30,696 Failed to meet target 0.400000 delta E, got worst case 1.727526 . and 2024-05-18 23:19:30,684 Patch 96 of 96 DE 0.001315, W.DE 1.314525, W.peqDE 1.314525, Fail ( > 0.250000) . Those are very rare not to get and usually only dont get any if I do not touch and white 20 point controls and that is only with fast speed quality. The good qualiy speed always fails something.
I’m convinced that using BT1886 for display calibration is the best approach we can have for SDR. But L* with a black output offset, or input, may be better. But here I can’t find any sources to know exactly how the eyes luminance response looks like in the best viewing conditions. So I stay with BT1886.
Old post. I find that setting black point to lower than actual and using bt1886 as gamma guide works good. Netflix uses a 2000 contrast ration for gradeing sdr and I only have 1000 contrast so I half the native black point. At 1/2 it matches my red and green and calibrates bt1886 to a 2000 contrast screen. It is very close to 2.2 gamma past 50% . 2.2 is brighter on brights past 50 percent bt1886 and not brighter on the blacks 15 precent and less. It is good to keep a flat curve for amount above and below the 100% calibration nits level at the near nits. Luminance going up and down looks crazy if it is not for gamma.
I set gamma 2.2 with 100% black output offset in novideo_sRGB. My profile is made using a custom testchart that has 256 neutral patches for accuracy. On my 1500:1 IPS panel other TRC like sRGB and BT1886 makes the shadows look washed out.
Technically with my 1500:1 contrast panel I could set white level to 75cd/m2 and be BT.1886 compliant by EBU Tech 3320. But I prefer a brighter white level between 100-150cd/m2, so I don’t bother evaluating. With a more limited 1000:1 panel neither sRGB or BT.1886 are worth using.
Thanks MW. It is so hard getting a good picture. Somedays i like something, somedays i do not like the same thing. I try to improve it. i took what I did before and matched red and green luminance and then looked better but then i could notice the washed out bt1886. Blue was not tracking the luminance. bt1886 set the luminance of red and green or maybe just blue different than black compensated gamma 2.2. I looked and saw missing details in windows dehighlighting instead of highlighted over a click on picture on the web. Trying increaseing red and green to match the blue gamma graph. Funny the gamma graph changings by the setting of tone curve in hcfr. Novideo_sRGB does work. Not needed on my not over rec709 colorspace green. I try to increase the colorspace to 91 or 92. Bad days its 89 or 90. New it was 92. 9000hrs on a screen does things.
novideo_srgb stated purpose is gamut clamping. But it’s more capable than that, ends up being better in a number of ways vs. Windows profile loader, DisplayCal profile loader, and MadVR 3DLUT loader. I’ve nerver seen novideo_srgb introduce color banding with temporal dither enabled, whereas multiple apps with color management systems do cause banding. MadVR is equal to novideo_srgb with regard to banding-free output. novideo_srgb only supports Curves+matrix is which is a one up for MadVR, but my display is well behaved and the effect on DeltaE is imperceptible, frankly the granger rainbow looks visually smoother with simpler curves+matrix profiles. novideo_srgb lets me switch on the fly between gamma values or specific TRCs like BT1886, with other tools this requires regeneration of the profile. That’s why I nerver looked back since I started using it. I cant speak about everyone’s specific setup but don’t sleep on novideo_srgb is good general advice.
novideo_srgb only supports Curves+matrix is that the real purpose? I’m using it with xyz+lut profile.
This is why my statment is true: It’s only optimized for curves+matrix profiles. It has functionality to load XYZLUT-profiles but will not apply as a XYZLUT-profile, instead it will derive a curves+matrix profile given a XYZLUT-profile.
