Spears & Munsil 4K UHD HDR Benchmark (Test Discs) May 15, 2023

[Geoff D]

Originally Posted by SeeMoreDigital

When spinning discs encoded with 2-layer Dolby Vision, the 4K UHD player does the video decoding, not the TV

It does beg the question though, given all the stuff we hear about player led vs TV led: if the TV isn't actually doing the decoding of the DV after all, then what's the difference between player led and TV led? With FEL is the TV only handling the application of the dynamic metadata (which is done upstream in player led) and not the actual recombination of the two layers?

[wxman2003]

Originally Posted by Geoff D

It does beg the question though, given all the stuff we hear about player led vs TV led: if the TV isn't actually doing the decoding of the DV after all, then what's the difference between player led and TV led? With FEL is the TV only handling the application of the dynamic metadata (which is done upstream in player led) and not the actual recombination of the two layers?

Is decoding different than tone mapping? If so, then with tv led, it may not decode, but it could still tone map. Or is that one in the same?

[SeeMoreDigital]

Originally Posted by wxman2003

Is decoding different than tone mapping? If so, then with tv led, it may not decode, but it could still tone map. Or is that one in the same?

Yes...

When spinning discs (in a 4K UHD player), in the case of 2-layer Dolby Vision, its 'tone mapping' meta-data can be analysed by either the player (player-led) or the TV (TV-led).

Our old friend Vincent Teoh explains the differences in one of his instructional videos, along with why TV-led meta-data analysis is more accurate

Dolby Vision TV-Led vs Player-Led Comparison: Which is Better? - YouTube

Cheers

[Balling]

Originally Posted by mrtickleuk

* a generic peak luminance value from the HDMI EDID which is the same for every TV,

There is no any generic luminance value in EDID of LG C9, CX, C1. EDID ommits it, some other TVs do have it. This: https://github.com/linuxhw/EDID/blob...0CF86F7EC0#L99

But there is one inside dolby vision vendor data block in EDID. Including primaries with very high precision. Use edid-decode to decode it. So it is very much possible to do it, except there is a bug in all Blu-ray players in LLDV implementation. So that is it, hdtvtest was wrong here (he is correct about Dolby IQ though). Actually I think we learnt about the bug in SDK due to scandal from his video. I think Stacey Spears was the first to talk about it?

[gotmule]

Originally Posted by Stacey Spears

Only 8? Poor soul, living in squalor, not sure how they make it through a movie.

Which AVR/SSP? Are they using a Trinnov? Curious how they are configured.

When I redo everything in a couple of years I am looking at going Trinnov, B.A.T. and Perlisten or Revel.

I have been using Meridian SSPs since the 565 came out in 94/95. The 861v8 has not been updated in years, so its time to move on.

Hey Stacey. REL markets the setup as their 3D system with a line array of 3 in each of the front corners, 1 in each of the rear corners, and a center effect sub hooked up to the center channel output. I got bored during COVID and went for it. I don’t know about the marketing behind it as it really is just a way to sell more subs to insane people like me. Each stack has the same signal run to it, and there is an output on each that feeds the next one up in the stack. Setup was tricky as crossovers and gains change within each stack as you go up, but that kind of tricky is a damn fun way to spend a couple of hours. Rear subs share the same signal between them and are crossed over and volume set the same as the bottom subs in the stacks up front.

I currently have the Marantz 8805, but have started scratching the itch again about the Trinnov and Storm Audio offerings. I blame Wendell for that.

It may seem like overkill, but the room has never sounded better as you basically sit in the middle of a jelly donut of perfect bass.

You can see some of the pictures in my gallery.

By the way, thank you for everything you do to make this glorious hobby of ours a better one.

[Geoff D]

Originally Posted by SeeMoreDigital

Yes...

When spinning discs (in a 4K UHD player), in the case of 2-layer Dolby Vision, its 'tone mapping' meta-data can be analysed by either the player (player-led) or the TV (TV-led).

Our old friend Vincent Teoh explains the differences in one of his instructional videos, along with why TV-led meta-data analysis is more accurate

Dolby Vision TV-Led vs Player-Led Comparison: Which is Better? - YouTube

Cheers

Tone mapping being done upstream or downstream is understood, I just never knew that the player also handles the FEL rebuild itself in a TV-led setup.

[mrtickleuk]

Originally Posted by Geoff D

Tone mapping being done upstream or downstream is understood, I just never knew that the player also handles the FEL rebuild itself in a TV-led setup.

Yes, it's new to me too. So if I've understood it, and I'm happy to be corrected, the only reason TVs up to now had two decoders was so that they could handle dual layer (more than 10 bit) decoding for Dolby Vision in their own internal media players and their own internal apps.

Since providers like Netflix seem to have decided never to stream 12 bit Dolby vision ever again (which I think is a bitter shame and it logically means we'll never see the full power of what Dolby Vision is capable of outside of spinning discs), it seems that someone has given all the manufacturers the "green light" to downgrade the capabilities of their TVs for the "new future" which will be limited to 10-bit Dolby Vision, except on a handful of spinning discs which have a FEL, on a format which is itself under threat.

[Balling]

Originally Posted by mrtickleuk

Since providers like Netflix seem to have decided never to stream 12 bit Dolby vision ever again

You're wrong, Netflix and Display+ (the last one is being ripped as we are talking) are using IPTPQc2 with reshaping in RPU that in by itself can provide for (just like ICtCp in zimg) 11.5 bit while encoded in 10 bits. Disney+ is using 20 Mbit/s video, while Netflix only 15.25 Mbit max.

AFAIK, when Blu-rays with FEL (NLQ+MMR) is being reconstructed IPTPQc2 is used too, even though base layer is just YCbCr.

Profile 7 can have NLQ+MMR but it can just have poly stuff.

[Geoff D]

Originally Posted by mrtickleuk

Yes, it's new to me too. So if I've understood it, and I'm happy to be corrected, the only reason TVs up to now had two decoders was so that they could handle dual layer (more than 10 bit) decoding for Dolby Vision in their own internal media players and their own internal apps.

Since providers like Netflix seem to have decided never to stream 12 bit Dolby vision ever again (which I think is a bitter shame and it logically means we'll never see the full power of what Dolby Vision is capable of outside of spinning discs), it seems that someone has given all the manufacturers the "green light" to downgrade the capabilities of their TVs for the "new future" which will be limited to 10-bit Dolby Vision, except on a handful of spinning discs which have a FEL, on a format which is itself under threat.

Pretty much a legacy thing, yeah. As UHD disc is the only current Dobly delivery system that uses the dual-layered approach AND we now know that TVs don’t even decode & rebuild the FEL themselves anyway it does seem rather redundant to put a dual decoder inside the TV. If streamers don’t want to use 12-bit streams then that’s their business, but as they wouldn’t need a second layer to make that happen (remember, Dobly is only like that on disc because of the need for a mandatory HDR10 base layer) then the lack of a dual decoder makes no difference there.

[Geoff D]

Originally Posted by Balling

You're wrong, Netflix and Display+ (the last one is being ripped as we are talking) are using IPTPQc2 with reshaping in RPU that in by itself can provide for (just like ICtCp in zimg) 11.5 bit while encoded in 10 bits. Netflix is using 20 Mbit/s video, while Netflix only 15.25 Mbit max.

AFAIK, when Blu-rays with FEL (NLQ+MMR) is being reconstructed IPTPQc2 is used too, even though base layer is just YCbCr.

Profile 7 can have NLQ+MMR but it can just have poly stuff.

But if these pre-shaped 10-bit streams can basically be reprocessed back to 11.5 bits then why bother with FEL at all for disc? Not having a go at you, just aksing a question, as encoding Dobly for disc appears to be a pain the ass at the best of times.

[edit] Oh, I see: the might wiki sez that usage of ITP for 10-bit encoding is regarded to be equivalent to 11.5 bits of YCbCr.

[Balling]

Originally Posted by Geoff D

But if these pre-shaped 10-bit streams can basically be reprocessed back to 11.5 bits then why bother with FEL at all for disc? Not having a go at you, just aksing a question, as encoding Dobly for disc appears to be a pain the ass at the best of times.

[edit] Oh, I see: the might wiki sez that usage of ITP for 10-bit encoding is regarded to be equivalent to 11.5 bits of YCbCr.

I wrote that on wikipedia, yeah. As almost all the ICtCp article.

As for why, FEL can reconstruct 4000 nits (not always), not just hillarious 12 bit (as mezannine master is 16 bit, you know). For obvious reasons (ICtCp does not require reshaping of encoding matrices on the fly to get 11.5 bits result) Dolby wants to move to ICtCp, but because the standard for ICtCp in VESA EDID was only finalised in January 2021, and no updates to HDMI are there yet... Of course AV1 with film grain and ICtCp with reshaping can do even more bits in YCbCr bits.

[FilmFreakosaurus]

Why the hell don't they have a true 12 bit depth video standard and get away from these proprietary jury-rigged approaches?

If 12 bits is really the minimum for HDR grading to pretty much get rid of visible banding and other anomalies exacerbated by the wider dynamic range, then move to an industry wide 12 bit ICtCp consumer standard (with hopefully better than 4:2:0 sampling matrices)!

The studio bean counters and lawyers can still breathe a sigh of relief that the unwashed masses still won't have access to the pristine 16 bit master archival files, and the studios won't have to cough up a bunch of royalties to Dolby or any other entity for proprietary tech.

[Geoff D]

Originally Posted by Balling

I wrote that on wikipedia, yeah. As almost all the ICtCp article.

As for why, FEL can reconstruct 4000 nits (not always), not just hillarious 12 bit (as mezannine master is 16 bit, you know). For obvious reasons (ICtCp does not require reshaping of encoding matrices on the fly to get 11.5 bits result) Dolby wants to move to ICtCp, but because the standard for ICtCp in VESA EDID was only finalised in January 2021, and no updates to HDMI are there yet... Of course AV1 with film grain and ICtCp with reshaping can do even more bits in YCbCr bits.

The FEL doesn’t matter much w/ref to 4000 nits when studios are already mastering the base layer to 4000 nits, Sony and Warners have never used FEL for that reason. But even for the studios that do use FEL with 1000-nit base layers I’ve never seen more highlight info in the final Dobly output than on the HDR10 base layer, though I have seen more lowlights: the dreadful black crush on the base layer of Annihilation is cured by the DV.

Originally Posted by FilmFreakosaurus

Why the hell don't they have a true 12 bit depth video standard and get away from these proprietary jury-rigged approaches?

If 12 bits is really the minimum for HDR grading to pretty much get rid of visible banding and other anomalies exacerbated by the wider dynamic range, then move to an industry wide 12 bit ICtCp consumer standard (with hopefully better than 4:2:0 sampling matrices)!

The studio bean counters and lawyers can still breathe a sigh of relief that the unwashed masses still won't have access to the pristine 16 bit master archival files, and the studios won't have to cough up a bunch of royalties to Dolby or any other entity for proprietary tech.

Uh…but that’s the point: ITP *is* proprietary to Dobly, isn’t it? Good luck getting the whole industry on board with that. Samsung would rather commit the Korean equivalent of hari-kari than pay a penny to them. Even 12-bit PQ-based HDR is behind that paywall, its name rhymes with Grolby Pigeon.

[David M]

Originally Posted by Geoff D

Tone mapping being done upstream or downstream is understood, I just never knew that the player also handles the FEL rebuild itself in a TV-led setup.

I'm sure it does. Doing base layer decoding on one brand of HEVC decoder and then passing an actual HEVC bitstream out over HDMI (if there is even a way of doing that) for decoding on a different decoder, then combining them in the display, probably wouldn't be a good idea. Better to keep it all in one box.

The TV's MPEG2/AVC/HEVC decoders are only used for in-TV apps like the tuner, streaming apps etc.

[GREM]

Originally Posted by Geoff D

And yet while I see some of those artefacts on the ZD9 the noise in the centre of the feather is cleaned up nicely in Dobly (either flavour) vs the HDR10.

Please take a look at this compairson https://slow.pics/c/Eyy9YjPX. Both shots are UHD DV versions, same HDMI input with identical settings, Bravia XF90, P-led. Sources are UBP-X700 and Shield TV.
Wich one is correct?
Maybe X700 is darker, maybe Shield is clipped.
Maybe both are wrong.
I dunno.
And if this frame has variations so do the others. Some noticable, some negligible.
Thing is if Dolby Vision promises to deliver picture as close to creators intend as possible than with Player-led this idea is ruined. It is not supervised DV (TV-led) or HDR10 verison, it is something crippled.
As end user I just can't be sure this or that device can output correct DV image in P-led mode. Shame on Dolby cause they just unable to do proper QC of their tech on consumer devices.
In this case I prefer* HDR10 on Bravia
*except StudioCanal disks

[Geoff D]

Originally Posted by GREM

Please take a look at this compairson https://slow.pics/c/Eyy9YjPX. Both shots are UHD DV versions, same HDMI input with identical settings, Bravia XF90, P-led. Sources are UBP-X700 and Shield TV.
Wich one is correct?
Maybe X700 is darker, maybe Shield is clipped.
Maybe both are wrong.
I dunno.
And if this frame has variations so do the others. Some noticable, some negligible.
Thing is if Dolby Vision promises to deliver picture as close to creators intend as possible than with Player-led this idea is ruined. It is not supervised DV (TV-led) or HDR10 verison, it is something crippled.
As end user I just can't be sure this or that device can output correct DV image in P-led mode. Shame on Dolby cause they just unable to do proper QC of their tech on consumer devices.
In this case I prefer* HDR10 on Bravia
*except StudioCanal disks

The Shield is *definitely* clipping some highlights. And yet in a world where people can choose from Dolby Dark, Dolby Bright and Dolby IQ I’m not sure that everyone’s looking at the same thing anyway. Not to excuse Dolby’s incompetence at releasing this half baked LLDV mode, but for me it still has more benefits than drawbacks. The day that changes (i.e. when StudioCanal stop using drunken marmosets to encode their product ) then I may stop using it, but for now it stays - and bear in mind I’m a picky bastard when it comes to picture quality.

Originally Posted by David M

I'm sure it does. Doing base layer decoding on one brand of HEVC decoder and then passing an actual HEVC bitstream out over HDMI (if there is even a way of doing that) for decoding on a different decoder, then combining them in the display, probably wouldn't be a good idea. Better to keep it all in one box.

The TV's MPEG2/AVC/HEVC decoders are only used for in-TV apps like the tuner, streaming apps etc.

Oh, the logic of it is impeccable. But given all the guff about player-led vs tv-led I just assumed that more was taking place downstream than just the tone mapping.

[mrtickleuk]

Firstly, thanks for all the replies which have helped my understanding! Really appreciate it!

Originally Posted by Geoff D

But if these pre-shaped 10-bit streams can basically be reprocessed back to 11.5 bits then why bother with FEL at all for disc?
[...]
[edit] Oh, I see: the might wiki sez that usage of ITP for 10-bit encoding is regarded to be equivalent to 11.5 bits of YCbCr.

But imagine how good it would look if they used ITP with 12-bit encoding?

Originally Posted by FilmFreakosaurus

Why the hell don't they have a true 12 bit depth video standard and get away from these proprietary jury-rigged approaches?

One reason, for 4k blu-ray at least (desperate attempt to drag back to topic!) is that the hardware HEVC decoders being used in the consumer devices chipsets are all 10bit, so that's why it was done as 10bit in one chunk and the other 2 bits in the other chunk. I personally think that's an ingenious way around that problem!

[KC-Technerd]

Originally Posted by Stacey Spears

You may have to power cycle the OPPO before playing a Dolby Vision disc if you had played HDR10 before hand with the player on. If a disc has an HDR10 menu and not Dolby Vision, then you may be out of luck.

I recently purchased Blue Underground's 4K UHD BD of "The Final Countdown" and found it to have an HDR10 menu with the main program in Dolby Vision. In case no one else has posted this yet, I think a possible workaround is to power off the Oppo while the desired DV title is playing, power the Oppo back on and restart play of the BD, then simply allow or select the "Resume from saved position" or equivalent option. Playback then begins in the Dolby Vision program without any HDR10 content being played first. I also found this to work on the UHD HDR Benchmark disc's Dolby Vision Demo Material.

[Katatonia]

Originally Posted by Stacey Spears

Just an update. We may end up moving to 4-discs. It appears that the file size has ballooned because of the M2Ts container, minimum disc block size and number of files. What should take ~51 GB is consuming ~91 GB of disc space.

Right now we are trying to figure out if we should cut some montage versions and merge what does not fit on disc 1 onto the disc with the montage videos keeping a 3-disc set or just moving to 4-discs and not cutting anything.

Stacey, I wanted to buy the current Benchmark UHD release for a friend. Has it gone temporarily out of print recently...or should I perhaps just wait until the updated set is released? I don't suppose there is an expected release date for that yet?

[Penton-Man]

Originally Posted by mrtickleuk

One reason, for 4k blu-ray at least (desperate attempt to drag back to topic!) is that the hardware HEVC decoders being used in the consumer devices chipsets are all 10bit, so that's why it was done as 10bit in one chunk and the other 2 bits in the other chunk. I personally think that's an ingenious way around that problem!

correct