Dolby Vision now possible through MP4 Mux.

[RESET_9999]

I ran some tests on my Sony A95L, and it seems tone mapping starts at:

• avg_pq 600, MDL 1000, CM v2.9: 1150 nits
• avg_pq 600, MDL 1000, CM v4.0: 1150 nits
• avg_pq 600, MDL 4000, CM v2.9: 150 nits!
• avg_pq 600, MDL 4000, CM v4.0: 1150 nits

How can one be sure that the old algorithm is being used?
Does having empty L2 trims guarantee the use of the legacy algorithm, or is there another way to confirm it?

thanks, I will add your TV to the chart. hard clipping point 1100nits?

Any cmv2.9 rpu made from a cmv4.0 workflow will always have an avg_pq floor of 10nits with L2 trims at 100,600,1000nits.
see this comparison, I've compared all the Dolby algo version: https://slow.pics/c/OA0pupEN

[bbeny123]

thanks, I will add your TV to the chart. hard clipping point 1100nits?

yep, 1100nits.

Damn, now I’m really torn about whether to generate DV for The Return of the King...
The original DV was already created using the new algorithm, with an MDL of 4000 nits, and it actually has really rich L2 trims.
Unfortunately, it seems the original CM v4.0 version isn’t available anywhere online.

L1_plot.png (227.42 KiB) Viewed 7020 times

L2_plot-100.png (270.99 KiB) Viewed 7020 times

L2_plot-600.png (255.4 KiB) Viewed 7020 times

[RESET_9999]

I would generate cmv4.0. The whole movie L1 is under your TV tone mapping target nits.
It doesn't matter how much editing they did in the trims... Unless they deliver a trim, the HDR10 layer always has the original HDR master brightness/grade and should be the reference, the intended representation.

[daffie]

I ran some tests on my Sony A95L, and it seems tone mapping starts at:

• avg_pq 600, MDL 1000, CM v2.9: 1150 nits
• avg_pq 600, MDL 1000, CM v4.0: 1150 nits
• avg_pq 600, MDL 4000, CM v2.9: 150 nits!
• avg_pq 600, MDL 4000, CM v4.0: 1150 nits

How can one be sure that the old algorithm is being used?
Does having empty L2 trims guarantee the use of the legacy algorithm, or is there another way to confirm it?

Any cmv2.9 rpu made from a cmv4.0 workflow will always have an avg_pq floor of 10nits with L2 trims at 100,600,1000nits.
see this comparison, I've compared all the Dolby algo version: https://slow.pics/c/OA0pupEN

What is the easiest way to check this ?

[RESET_9999]

What is the easiest way to check this ?

Check the L1 plot for the 10nits avg_pq floor

[daffie]

Is there some list of movies in old/original cmv2.9 algo (the ones with no automatic trims, no cmv4.0 to 2.9 conversion)?

I read some pages back that this is only about 10% of the movies anyway. Is that correct?

[RESET_9999]

Probably all the bluray released before 2020.

Just look at the plot here, it has practically all the movies plotted:
https://drive.google.com/drive/u/1/fold ... KFJwFxnDt-

[bbeny123]

I would generate cmv4.0. The whole movie L1 is under your TV tone mapping target nits.
It doesn't matter how much editing they did in the trims... Unless they deliver a trim, the HDR10 layer always has the original HDR master brightness/grade and should be the reference, the intended representation.

So the artistic impact of those L2 trims would be minimal?
In your opinion, is it better to go for higher brightness by eliminating the too early tone mapping issue in CM v2.9, rather than keeping the version with manually tuned L2 trims?

[RESET_9999]

So the artistic impact of those L2 trims would be minimal?
In your opinion, is it better to go for higher brightness by eliminating the too early tone mapping issue in CM v2.9, rather than keeping the version with manually tuned L2 trims?

It depends on how strongly the colorist adjusted them. I’ve seen many 100 nits trim passes that make the image brighter than the master (HDR10 encode). It seems some colorists are not aware and/or are not monitoring the interpolation to the 600–1000 nits target when they do the SDR trim. The trim’s purpose is to adjust Dolby’s default tone mapping (L1) when it deviates from the artistic intent, which is the 12-bit master.
Making the trims different/brighter from the master doesn’t make much sense and is not recommended in Dolby best practice because this trim will affect more displays with lower targets, and higher-nits displays won’t even see that trim’s effect.

There is a way for the colorist to be more aggressive on the 100 nits trim without affecting the other targets. When you see an RPU with a blank 600 and/or 1000 nits trim, this is exactly what the colorist did.
When I create trims for my own work, I always start by making the 600 nits trim blank, and then I go through all the shots and adjust my 100 nits trim. This way, I know that whatever I do in 100 nits will not affect my other targets. When I’m working on a shot that exceeds 600 nits in the scopes, then I can either unblank the 600 nits trim and let the 100 nits interpolation do its work, or better, adjust the 600 nits trim.

FYI, you can use my script workflow 7-4 to make L1 vs L1-L2 comparison of any frame: I already did many here:
https://drive.google.com/drive/folders/ ... dMiy2ka0mf

[deadchip12]

@RESET9999

Could you let me know which rpu I should use for Presence (2024)? There are 3 of them in your folder. Thanks.

Presence.2024.2160p.AMZN.WEB-DL.DDP5.1.HDR.H.265 Generated T3.xml
Presence.2024.BD Generated T3.xml
Presence.2024.BD_Generated_T3 (frame by frame).xml

[coopzr]

@RESET9999

Can you please explain how FEL interacts with generated DV?

I assume that if you generate your own DV the FEL gets discarded. However, some of your RPU releases say "movie name P7 FEL Only Generated T3 OSC" and now has me second guessing on how it works.

[RESET_9999]

@RESET9999

Could you let me know which rpu I should use for Presence (2024)? There are 3 of them in your folder. Thanks.

Presence.2024.2160p.AMZN.WEB-DL.DDP5.1.HDR.H.265 Generated T3.xml
Presence.2024.BD Generated T3.xml
Presence.2024.BD_Generated_T3 (frame by frame).xml

one is the usual shot by shot for amazon
one is the usual shot by shot for the bluray disc
and the last one is frame by frame because this movie has very long shots so you should use this one

[RESET_9999]

@RESET9999

Can you please explain how FEL interacts with generated DV?

I assume that if you generate your own DV the FEL gets discarded. However, some of your RPU releases say "movie name P7 FEL Only Generated T3 OSC" and now has me second guessing on how it works.

Nothing gets discarded. I just replace the L1-L2 and add L3 L8 L9

[bbeny123]

Nothing gets discarded. I just replace the L1-L2 and add L3 L8 L9

So it's possible to replace all levels in a P7 FEL RPU? The FEL layer isn't "tied" to L1?

Interesting detail: based on my tests, cm_analyze is deterministic - for the same input and arguments, the generated RPU is always identical.

[RESET_9999]

As long as you reconstruct the master for the DV analysis.
FEL contains the difference between the master and the HDR10 encode, which can be fully restored with the dovi_baker.