- This topic has 23 replies, 6 voices, and was last updated 1 year, 1 month ago by
.
-
Posts
-
Put it in a different way, for XYZLUT in novideo_srgb there is an internal conversion from XYZLUT to curves+matrix. The conversion results in a loss of accuracy. When given a curves+matrix profile instead there is no conversation and accuracy is maintained, because it’s 1:1.
Put it in a different way, for XYZLUT in novideo_srgb there is an internal conversion from XYZLUT to curves+matrix. The conversion results in a loss of accuracy. When given a curves+matrix profile instead there is no conversation and accuracy is maintained, because it’s 1:1.
<svg style=”display: none;”><filter id=”_videostyler_filter_”></filter></svg>
I guess it may be, but for my display curves+matrix is not working well.
Using sRGB transfer function as a target for a display makes no sense. Why?
- If you watch a movie, blacks will be washed out.
- If you view an image using a color managed application, it doesn’t matter which transfer function the monitor is calibrated to as long your ICC profile TRC tag specifies the same one. The color managed app will decode the sRGB image to linear space and encode it back using the TRC function making it look the same on a 1.8 or 2.6 gamma screen.
Keep in mind that EBU Tech 3320 mandates a contrast ratio of at least 2000:1 to use BT.1886 at 100cd/m2. And for 120cd/m2 at least 2400:1 contrast ratio.
sRGB TRCs inherent gamma lift ner black gets compounded with the black output offset required for non-zero black displays to avoid black crush, that’s why it looks washed out.
Keep in mind that EBU Tech 3320 mandates a contrast ratio of at least 2000:1 to use BT.1886 at 100cd/m2. And for 120cd/m2 at least 2400:1 contrast ratio.
sRGB TRCs inherent gamma lift ner black gets compounded with the black output offset required for non-zero black displays to avoid black crush, that’s why it looks washed out.
It still bothers me how can you view content mastered on a 100 nits 2000:1 contrast on a 100 nist 1000:1 contrast…I don’t think it will look the same. Even with 2.4 gamma.
Keep in mind that EBU Tech 3320 mandates a contrast ratio of at least 2000:1 to use BT.1886 at 100cd/m2. And for 120cd/m2 at least 2400:1 contrast ratio.
sRGB TRCs inherent gamma lift ner black gets compounded with the black output offset required for non-zero black displays to avoid black crush, that’s why it looks washed out.
It still bothers me how can you view content mastered on a 100 nits 2000:1 contrast on a 100 nist 1000:1 contrast…I don’t think it will look the same. Even with 2.4 gamma.
I agree it won’t look the same. And EBU Tech 3320 does not consider those two conditions(100 nits on 2000:1 vs. 1000:1) equally as far as BT.1886 compliance.
Keep in mind that EBU Tech 3320 mandates a contrast ratio of at least 2000:1 to use BT.1886 at 100cd/m2. And for 120cd/m2 at least 2400:1 contrast ratio.
sRGB TRCs inherent gamma lift ner black gets compounded with the black output offset required for non-zero black displays to avoid black crush, that’s why it looks washed out.
It still bothers me how can you view content mastered on a 100 nits 2000:1 contrast on a 100 nist 1000:1 contrast…I don’t think it will look the same. Even with 2.4 gamma.
I agree it won’t look the same. And EBU Tech 3320 does not consider those two conditions(100 nits on 2000:1 vs. 1000:1) equally as far as BT.1886 compliance.
So what is the solution for those with 100 nits and 1000:1 displays?
AFAIK you can go for 2.4/2.35, relative, 100% black output offset (displayCAL). This, with gentle bending of gamma vs input 2D plot validates Ok in CalMAN “power curve reference” and will validate OK in HCFR for 100% input offset reference.
In the same way as print black point compensation, or Adobe black point compensation for non 0 L* at RGB 0 TRC curve in ICC profiles, these “out of gamut” blacks can be displayed with some “perceptual-like” deformation. That setting will do the job and can be verified as OK with 3rd party software.
Something near to black compensated gamma 2.2 . I am renegade and use HCFR and see the native gamma. and note how Red Green and Blue change with the 2point white point adjustment. Blue is way low on RGB 16 for its correct lumance on the non-factory gamma correction in the service menu. I set blue 5% white with the 10% luminance of blue in Calmans 10 point luminance test in HCFR to be 15 blue instead of 2 blue and not its expected 16 on 16 since other colors are 15 of 16 on 16. Set it to bt1886 not 2.4 or 2.2 . Blue matches the 5% so make 5 to 100 % match in 100% in red and green. 5% is bad delta e but 10 and up is good. 10% is raised blue to match it to the blue number of white 10% with 1/2 the difference of blue white at 5%. The white balance is wrong but the blue is right. Red and Green is right too but they read 0 at 0% and blue reads 8 or 9 at 0%. So, 15 reds, greens and blues to 5% is great. I really cannot get it exact. Display Cal gets it exact with auto black correction on but it gives less contrast.
I do not know what the solution is. 2.2 black compensated or bt 1886. Try them and see. Sources change and to much detail in the shadows shows black bars in movies as not black so too much is bad.
