Dolby Vision now possible through MP4 Mux.

[skull88]

Let's say one makes a content with 1000nits master display and does 100nits trim for SDR. When the content gets playback on a 1000nits display, it will need to use 100nits trim? This does not make sense to me.

Re-read my previous post and the following section in Dolby docs. You're focusing too much on the numbers and labels, and not how Dolby designed their pipeline. They interpolate from the 100-nit SDR trim to higher nit targets, and they recommend monitoring the SDR trim preview on the same HDR display or display with similar backlight technology for this reason. Obviously if 100-nit SDR trims were not improving the image (for SDR deliverable or not), they would just be static or removed, but they're almost always present, whether automatically generated or, as is almost always the case, manually adjusted by the colourist, which means Dolby Vision processing does some math and interprets these values for all possible end-user setups. They are also usually required by studios for delivery, although not 100% of time, as Netflix says they are optional for example. There are still questions over implementation, cmv2.9 vs. cmv4.0, but this is a major point of confusion now resolved, at least in my mind. Finally, as RESET_9999 pointed out, you really need to watch his comparison and testing clips, or just test yourself on your setup.

[susanstone2022]

trims are not just for diming the picture to the TV target brightness... They can use the trims to lift the brightness and they do that very often. I literally made a bunch of test file with real content that show exactly that.
sorry but It sounds to me that you only read stuff online and you don't actually check out how it behaves on your TV.

yes. i know it can do stuff other than diming pictures. my question is why a 1000nits tv need to use trim that target to 100nits display? the trim is created by referencing 100nits target. it does not make sense to use it on a 1000nits tv.

[RESET_9999]

Well if TV-LEd works the same as LLDV for the target brightness, an internal TV config or an edid of 1000+nits with a 1000nits RPU would make the trims to be ignored which makes less sense to me because then the intentional positive lift would be ignored by brighter TV? I know the trims are ignored in LLDV with as shown here but my tv internal config is only 810nits(assuming my calib guy didn't change it) so I don't know if it's the same in TV-LEd (probably)

https://www.youtube.com/watch?v=29ed65Aq5K0

[skull88]

yes. i know it can do stuff other than diming pictures. my question is why a 1000nits tv need to use trim that target to 100nits display? the trim is created by referencing 100nits target. it does not make sense to use it on a 1000nits tv.

I hate to do this, but please see Interpolation. This is literally built into the spec and Dolby's conception of how their technology works. As explained before, the 100-nit SDR trim is already used for interpolation to generate higher-nit trim level data, automatically, by default. It does not mean it's linear and likely Dolby extensively tested this internally over the years to ensure it does "make sense" when TVs or devices tonemap accordingly -- this is important, as not all devices and situations seem to follow the spec and bugs do exist in different versions of the Dolby Vision engine/SDK in devices like the Shield Pro 2019 etc. It is important to separate these two distinct topics: 1) how Dolby Vision (metadata, RPU, engine, spec, algorithms, tools) is designed and works in line with Dolby's suggested best practices and 2) how displays and devices process, pass, and receive Dolby Vision data that is generated by studios/colourists/software.

The logical leap, if any, is that if only the 100-nit L2 (or L8) trim exists, the engine still uses that to interpolate/interpret and adjust the image for higher-nits and variety of displays. If it didn't do this, it would be hard to sell everyone on Dolby Vision as future-proof, given the huge difference in nits-output between an HDR DV capable consumer display in 2016-2018 and 2023-2025, or even mastering and colourist habits and mistakes (e.g. some UHD BDs had ridiculously strange and seemingly invalid DV metadata even as late as 2019). Case and point, some HDR colourists are putting an SDR grade inside an HDR container, which literally means that Dolby Vision would have to interpolate up from an SDR grade and any 100-nit SDR trims, there's nothing else there! Hopefully that clears things up a little, I'm off for the weekend. I strongly recommend reading and testing more on the side to see things for yourself.

[susanstone2022]

Well if TV-LEd works the same as LLDV for the target brightness, an internal TV config or an edid of 1000+nits with a 1000nits RPU would make the trims to be ignored which makes less sense to me because then the intentional positive lift would be ignored by brighter TV? I know the trims are ignored in LLDV with as shown here but my tv internal config is only 810nits(assuming my calib guy didn't change it) so I don't know if it's the same in TV-LEd (probably)

https://www.youtube.com/watch?v=29ed65Aq5K0

do you know the lift is targeted 100nits TV or 1000nits TV? I bet it is the former.

[susanstone2022]

yes. i know it can do stuff other than diming pictures. my question is why a 1000nits tv need to use trim that target to 100nits display? the trim is created by referencing 100nits target. it does not make sense to use it on a 1000nits tv.

I hate to do this, but please see Interpolation. This is literally built into the spec and Dolby's conception of how their technology works. As explained before, the 100-nit SDR trim is already used for interpolation to generate higher-nit trim level data, automatically, by default. It does not mean it's linear and likely Dolby extensively tested this internally over the years to ensure it does "make sense" when TVs or devices tonemap accordingly -- this is important, as not all devices and situations seem to follow the spec and bugs do exist in different versions of the Dolby Vision engine/SDK in devices like the Shield Pro 2019 etc. It is important to separate these two distinct topics: 1) how Dolby Vision (metadata, RPU, engine, spec, algorithms, tools) is designed and works in line with Dolby's suggested best practices and 2) how displays and devices process, pass, and receive Dolby Vision data that is generated by studios/colourists/software.

The logical leap, if any, is that if only the 100-nit L2 (or L8) trim exists, the engine still uses that to interpolate/interpret and adjust the image for higher-nits and variety of displays. If it didn't do this, it would be hard to sell everyone on Dolby Vision as future-proof, given the huge difference in nits-output between an HDR DV capable consumer display in 2016-2018 and 2023-2025, or even mastering and colourist habits and mistakes (e.g. some UHD BDs had ridiculously strange and seemingly invalid DV metadata even as late as 2019). Case and point, some HDR colourists are putting an SDR grade inside an HDR container, which literally means that Dolby Vision would have to interpolate up from an SDR grade and any 100-nit SDR trims, there's nothing else there! Hopefully that clears things up a little, I'm off for the weekend. I strongly recommend reading and testing more on the side to see things for yourself.

I think the major selling point of DV previously or at the moment is L1 data compare to HDR10.

[RESET_9999]

do you know the lift is targeted 100nits TV or 1000nits TV? I bet it is the former.

just find out by playing these:
https://drive.google.com/file/d/1NM2Vjz ... sp=sharing
https://drive.google.com/file/d/1Qfa2zo ... sp=sharing
https://drive.google.com/file/d/1TBXGnp ... sp=sharing
https://drive.google.com/file/d/1oN1uV8 ... sp=sharing

As i said earlier the L8 100nits trim is a noticeable positive on my 810nits C2 which matches the positive lift of CMV2.9 (600+1000).
I also tested L2 with a 100nits trim and it also raised brightness but not as much as 600+1000nits.

Same story on my PC monitor(samsung 400nits) with the Energy player.

[skull88]

I think the major selling point of DV previously or at the moment is L1 data compare to HDR10.

Obviously that's always been the baseline marketing, the "dynamic" part of it. However, that's been there since the initial release and HDR10+ also has this, so what has changed, especially in cmv4.0 is the added creative controls and automatic (now tunable) analysis and generation of trims! If you play around with cm_analyze for example, you'll note that it generates the 3-levels of L2 trims by default for backwards compatibility. Why would this be the case, if L1 was enough? The default is not to have ONLY L1+L3, right? This is precisely because it allows a colourist to run the automatic algo and then only make adjustments to the shots or trims that they actually feel need them. To me, that's precisely why Dolby Vision is the metadata Apple and Netflix chose for their platforms, and not HDR10+, aside from business reasons, it gives a wild amount of control (especially in cmv4.0), but also works quite well if you just let it do its thing automatically, despite bugs and issues with devices/partner implementation. Oh, and forgot to add that you actually don't even need a paid license to generate L1, but you do for adjusting any L2/L8 trims, that tells the whole story in my view. In any case, back to actual work lol. Cheers!

[RESET_9999]

I think the major selling point of DV previously or at the moment is L1 data compare to HDR10.

Actually the best thing about DV is the trim pass because they can control how it will look on more display.
L1 only is like HDR10plus or madvr HDR dynamic tonemaping, it basically just compress the highlights.

[RESET_9999]

The logical leap, if any, is that if only the 100-nit L2 (or L8) trim exists, the engine still uses that to interpolate/interpret and adjust the image for higher-nits and variety of displays.

yep, I think that is what's happening. The 100nits trim response is clearly different on my 400 and 800 monitor

[susanstone2022]

The logical leap, if any, is that if only the 100-nit L2 (or L8) trim exists, the engine still uses that to interpolate/interpret and adjust the image for higher-nits and variety of displays.

yep, I think that is what's happening. The 100nits trim response is clearly different on my 400 and 800 monitor

Have you asked Dolby directly on how the trim data are used or supposed to be used during playback? Can you please ask if you have not?

[RESET_9999]

They haven't responded to my L2 trim offset being ignored in TV-LED yet so I don't want to open a new ticket already but I do want to ask them what happens with positive lift for 1000nits RPU with 1000+nits TV.
This is the only remaining thing that doesn't make sense in my mind. Everything else is pretty clear.

[RESET_9999]

1. Does MakeMkv rip DV tracks as dual track dual layer, or single track dual layer profile 7?
2. Related to above, are there options for how MakeMKV rips DV layers I should be aware of?
3. I believe that to improve compatibility of DV files, I need to convert profile 8 and there are tools available to do this (dovi_tool)
4. I've read that removing FEL in that process is a good thing as it isn't used by players?

1- single track dual layer P7
2- idk
3- yes if your device doesn't support fel, P8 is better.
4- same as response as 3 but be aware that some movies require FEL to be displayed correctly. see: https://docs.google.com/spreadsheets/d/ ... edit#gid=0

[RESET_9999]

I tested the AppleTV , Shield and x700, the behavior was the same:

on the C2, 100nits L2 or L8 has an effect if 600-1000trims is not 2048 and the effect is much more pronounced with lower nits TV.

[susanstone2022]

I tested the AppleTV , Shield and x700, the behavior was the same:

100nits L2 or L8 has an effect if no 600-1000trims exist and the effect is much more pronounced with lower nits TV.

Does the effect ever phase out on higher nits TV?