Dolby Vision now possible through MP4 Mux.

Please post here for issues related to UHD discs
daddy
Posts: 77
Joined: Sat Jul 06, 2019 8:17 am

Re: Dolby Vision now possible through MP4 Mux.

Post by daddy » Tue Dec 22, 2020 6:29 pm

Pacer wrote:
Mon Dec 21, 2020 5:28 pm
Hello everyone,

first I wanted to say a big THANK YOU to everyone so dedicated to get the DV playback riddle solved, first of all to yusesope, whose tool I've tried to use.

I've just had one UDH Bluray Disc (MI: Fallout), which I tried to get to run in DV after ripping it. With the help of yusesope's tool I was able to create an mp4 file which my LG OLED65GX9LA liked to play. I used the software and procedure described on p199 in this forum:

- ffmpeg to demux from BD m2ts file
- MUX_AND_CONVERT_TO_PROFILE_81.bat
- mp4muxer to get an .mp4 file

The file triggered DV on my LG TV, but I am not sure if it displayed the content correctly. That's why I would like to ask you more experienced people to take a look at my photo comparison (DSLR in manual mode). From left to right you see:

- MKV ripped with 1.15.3 from BD disc, played on my HiMedia Q10 Pro (which plays the file only in HDR10)
- same file, but played from my LG OLED65GX9LA with internal player via DLNA (which also plays it in HDR10)
- mp4-File created with the above procedure, played from my LG's internal player with active DV

Bildschirmfoto 2020-12-12 um 19.55.51.jpg

To me, the colors look quite different compared to the HDR10 variants of the movie. I do not have a hardware BD-UHD player to see how the movie would look like when played from there :-/

What are your thoughts on this?
In my expectation the colors should not differ between HDR10 and DV, right?
The EL is a 10-bit video bitstream that carries the residual between the source and the BL and the “dynamic” metadata for DV. An enhancement layer that carries a residual signal is = 0 (zero), that is, the decoder does not need to process the residual signal, is called minimum enhancement layer (MEL). If the residual signal is > 0, it is called full enhancement layer (FEL)

MI Fallout is a FEL Dolby Vision movie. (BL + EL (>0) + RPU)
Making DV files in single layer profile works only with MEL (EL=0).
The only correct method to have dual layers files (or dual tracks of single layer) is with mp4muxer with Dolby Vision in profile 7.

agneva
Posts: 61
Joined: Mon Jan 06, 2020 11:42 am

Re: Dolby Vision now possible through MP4 Mux.

Post by agneva » Tue Dec 22, 2020 9:54 pm

Did you try the ape?

pete19
Posts: 31
Joined: Thu Dec 03, 2020 12:37 pm

Re: Dolby Vision now possible through MP4 Mux.

Post by pete19 » Tue Dec 22, 2020 10:19 pm

lexyz wrote:
Tue Dec 22, 2020 1:57 pm
Do you suggest me to search for a proof for your statement? Really? :D
I would rather not spend time on this misconception.
you don't seriously think anybody here will spend more time educating about something that you've not done your research for.

but you can always "believe" what you want :)

like I said, this is years old. You spread misinformation, I'm sure not on purpose. :roll:

two main things about DV that are the reasons it is the premiere HDR format:
(1) ability to fully re-create the 12bit 422 master at home in FEL dual layer releases
(2) dynamic metadata

quietvoid
Posts: 373
Joined: Sun Apr 19, 2020 4:15 pm

Re: Dolby Vision now possible through MP4 Mux.

Post by quietvoid » Tue Dec 22, 2020 10:49 pm

pete19 wrote:
Tue Dec 22, 2020 10:19 pm
you don't seriously think anybody here will spend more time educating about something that you've not done your research for.

but you can always "believe" what you want :)

like I said, this is years old. You spread misinformation, I'm sure not on purpose. :roll:

two main things about DV that are the reasons it is the premiere HDR format:
(1) ability to fully re-create the 12bit 422 master at home in FEL dual layer releases
(2) dynamic metadata
You're the only one posting conflicting statements without providing proof. :?:

As I've mentioned already:
quietvoid wrote:
Mon Dec 21, 2020 5:15 am
420 is not wrong, the patent specs also only mention 12 bit YCbCr 420 reconstructed output.
As for the HDMI transmission, it is detailed as 12 bit YCbCr 422 because the metadata is also transported into the least significant bits.
No where is there a mention that the output video is actually 12 bit 422.

There's no place for full 4:2:2 video:
Image

Manixx2020beyound
Posts: 127
Joined: Thu Oct 08, 2020 5:19 pm

Re: Dolby Vision now possible through MP4 Mux.

Post by Manixx2020beyound » Wed Dec 23, 2020 3:43 am

C006C359-D5D4-4621-B8B6-27FC3EDF7B99.png
C006C359-D5D4-4621-B8B6-27FC3EDF7B99.png (687.86 KiB) Viewed 17681 times
05A0F454-1A78-430E-BD42-5F969372BDDC.png
05A0F454-1A78-430E-BD42-5F969372BDDC.png (444.74 KiB) Viewed 17681 times
Might be some confusion where the mezzanine is concern.

pete19
Posts: 31
Joined: Thu Dec 03, 2020 12:37 pm

Re: Dolby Vision now possible through MP4 Mux.

Post by pete19 » Wed Dec 23, 2020 6:03 am

quietvoid wrote:
Tue Dec 22, 2020 10:49 pm
420 is not wrong, the patent specs also only mention 12 bit YCbCr 420 reconstructed output.
As for the HDMI transmission, it is detailed as 12 bit YCbCr 422 because the metadata is also transported into the least significant bits.
first off, nobody said 420 is "wrong" - that is your wording. 420 simply has less color resolution, hence 422 is more desirable.

Second, both STD and LLDV output a 12bit 422 signal yet only STD contains in addition the dynamic metadata because it's tv-led. so the player-led stream (already DV decoded) which also contains 12bit 422 signal requires less bandwidth, hence it provides low latency...

again, BOTH DV stream types contain 12bit 422 signal (because that's the Dolby spec), yet LLDV has less data than the other... only STD carries metadata in the stream, not LLDV.... hence your 2nd statement is also wrong.

to recap:
all DV streams are 12 bit 422.
only tv-led contains the full dynamic metadata, so full stream requirements, hence RGB tunneling in 8bit float container is used.
player-led requires less bandwidth because it contains less data (still 12 bit 422 signal).
MEL is also 12 bit 422 but only contains a 10bit 420 signal, padded.

and to stress this part:
quietvoid wrote:
Tue Dec 22, 2020 10:49 pm
because the metadata is also transported into the least significant bits.
you make it sound like that "explains" a 12bit 420 final output. no, the metadata allows to reconstruct the full 12bit 422 master, which is why it is a 12bit 422 signal.

if final output was 12bit 420 (so EL only contains luminance, Y difference) for FEL STD, they wouldn't need a 12 bit 422 stream, they'd use a 12bit 420 stream. Metadata is the mere difference, it doesn't need "extra" space on top of the difference, especially since it's encoded as well.

DaMacFunkin
Posts: 311
Joined: Tue Oct 30, 2018 4:17 pm

Re: Dolby Vision now possible through MP4 Mux.

Post by DaMacFunkin » Wed Dec 23, 2020 6:58 am

🥊
I’m off to get some popcorn for round 2, this is freakin’ awesome 🤩

pete19
Posts: 31
Joined: Thu Dec 03, 2020 12:37 pm

Re: Dolby Vision now possible through MP4 Mux.

Post by pete19 » Wed Dec 23, 2020 7:40 am

DaMacFunkin wrote:
Wed Dec 23, 2020 6:58 am
I’m off to get some popcorn for round 2, this is freakin’ awesome
you want more ? here is an encore... :D

most folks here are forgetting that 12bit 422 ONLY ever comes into play for profile 7 FEL - not even profile 7 MEL (10bit 420).

All the gazillion other DV profiles are all 420 drained sauce for streaming. THIS is easily 90+% of DV content.

Then, for disc releases (--> profile 7), (unfortunately) the majority of releases are MEL... so 12bit 422 (--> p7 FEL) only rarely ever comes into play.

This may be another reason why in the many Dolby docs this is not explicitly stated, to not confuse content creators who in 9/10 cases will use single DV layer BL (for garbage compressed streaming), and even if authoring disc deliverables, then often use MEL.

Even some of the largest disc releases do not get DV or DV FEL (--> SW, only EVIII from Ryan Johnson is DV, but MEL). There are authoring issues with FEL when one gets into the 2nd and 3rd layers on the UHD disc, so that seems to be another huge obstacle.

again: p7 FEL tv-led is as good as it gets right now.

pete19
Posts: 31
Joined: Thu Dec 03, 2020 12:37 pm

Re: Dolby Vision now possible through MP4 Mux.

Post by pete19 » Wed Dec 23, 2020 8:23 am

and here's more...

there are only two (2) DV streams:

1) 8bit RGB via tunneling (for STD tv-led, as it contains both BL+EL data)
2) 12bit 422 for everything else (LLDV player-led, MEL, or single layer profiles)

IF (!) the actual max chroma subsampling was 420, then the 2nd stream variant above would only need to be 12bit 420 - there would be no need to use 12bit 422 (+ HDCP, hence you'd never use more bandwidth than needed).

MEL is just 10bit 420 anyways, all single layer profiles as well.
IF (!) profile 7 LLDV was 12bit 420 for FEL releases you could just use a 12bit 420 stream as LLDV carries no metadata (DV was already encoded by player).

But they specc'ed a 12bit 422 stream as LLDV for p7 FEL is a 12bit 422 signal.

Pacer
Posts: 2
Joined: Mon Dec 21, 2020 5:09 pm

Re: Dolby Vision now possible through MP4 Mux.

Post by Pacer » Wed Dec 23, 2020 8:47 am

daddy wrote:
Tue Dec 22, 2020 6:29 pm
Pacer wrote:
Mon Dec 21, 2020 5:28 pm
Hello everyone,

first I wanted to say a big THANK YOU to everyone so dedicated to get the DV playback riddle solved, first of all to yusesope, whose tool I've tried to use.

I've just had one UDH Bluray Disc (MI: Fallout), which I tried to get to run in DV after ripping it. With the help of yusesope's tool I was able to create an mp4 file which my LG OLED65GX9LA liked to play. I used the software and procedure described on p199 in this forum:

- ffmpeg to demux from BD m2ts file
- MUX_AND_CONVERT_TO_PROFILE_81.bat
- mp4muxer to get an .mp4 file

The file triggered DV on my LG TV, but I am not sure if it displayed the content correctly. That's why I would like to ask you more experienced people to take a look at my photo comparison (DSLR in manual mode). From left to right you see:

- MKV ripped with 1.15.3 from BD disc, played on my HiMedia Q10 Pro (which plays the file only in HDR10)
- same file, but played from my LG OLED65GX9LA with internal player via DLNA (which also plays it in HDR10)
- mp4-File created with the above procedure, played from my LG's internal player with active DV

Bildschirmfoto 2020-12-12 um 19.55.51.jpg

To me, the colors look quite different compared to the HDR10 variants of the movie. I do not have a hardware BD-UHD player to see how the movie would look like when played from there :-/

What are your thoughts on this?
In my expectation the colors should not differ between HDR10 and DV, right?
The EL is a 10-bit video bitstream that carries the residual between the source and the BL and the “dynamic” metadata for DV. An enhancement layer that carries a residual signal is = 0 (zero), that is, the decoder does not need to process the residual signal, is called minimum enhancement layer (MEL). If the residual signal is > 0, it is called full enhancement layer (FEL)

MI Fallout is a FEL Dolby Vision movie. (BL + EL (>0) + RPU)
Making DV files in single layer profile works only with MEL (EL=0).
The only correct method to have dual layers files (or dual tracks of single layer) is with mp4muxer with Dolby Vision in profile 7.
Interesting. Your statement stands in contrast to what Yusesope has written in his post:

UPDATE 02/10/20
I added the possibility to "convert" a UHD-BD (MEL or FEL, it doesn't matter) into a profile 8.1 file (Single Track Single Layer)


2. UHD-BD TO PROFILE 8.1
You can use both UHD-BD MEL and FEL


Can you explain, why this should not be possible with this tool?

Thanks a lot!
Pacer

MastaG
Posts: 91
Joined: Thu Aug 08, 2019 11:40 am

Re: Dolby Vision now possible through MP4 Mux.

Post by MastaG » Wed Dec 23, 2020 9:12 am

So which players are capable of playing FEL remuxes from network sources (Plex/Emby/SMB), e.g. the full deal retaining the 12bit 4000nits final picture like when playing the original disc?
e.g. remuxed into a TS or MKV container so I can add a high def audio track?

shawnc22
Posts: 637
Joined: Tue Jan 21, 2020 7:40 am

Re: Dolby Vision now possible through MP4 Mux.

Post by shawnc22 » Wed Dec 23, 2020 9:16 am

MastaG wrote:
Wed Dec 23, 2020 9:12 am
So which players are capable of playing FEL remuxes from network sources (Plex/Emby/SMB), e.g. the full deal retaining the 12bit 4000nits final picture like when playing the original disc?
e.g. remuxed into a TS or MKV container so I can add a high def audio track?
From network? The oppo and that's it.

pete19
Posts: 31
Joined: Thu Dec 03, 2020 12:37 pm

Re: Dolby Vision now possible through MP4 Mux.

Post by pete19 » Wed Dec 23, 2020 9:40 am

MastaG wrote:
Wed Dec 23, 2020 9:12 am
So which players are capable of playing FEL remuxes from network sources (Plex/Emby/SMB), e.g. the full deal retaining the 12bit 4000nits final picture like when playing the original disc?
e.g. remuxed into a TS or MKV container so I can add a high def audio track?
Oppo with (free) Russian firmware hack can do all of that. And you can mux to m2ts or BDMV folder structure and it'll play it (and keep HD audio). As long as you don't trim the movie, tsmuxer creates a perfect DV mux.

aboulfad
Posts: 111
Joined: Sat Feb 09, 2019 12:51 pm

Re: Dolby Vision now possible through MP4 Mux.

Post by aboulfad » Thu Dec 24, 2020 11:29 am

Hi,
Can someone elaborate on yusesope´s comment copied below. Would (and how) DV streaming providers (NF, Apple) have to deal with the lack of EL in a DV profile 5 and potential « issues » such as chromatic aberrations. Or are those issues a side effect of yusesope method for creating a single track single layer DV media ?

One point that seems not so obvious, is that a Profile 5 can make use of dynamic metadata carried in the RPU- Reference Processing Unit, what is this? Thank you.
If instead I choose to play my files with the profiles dvhe.05 and dvhe.08 ..., the hardware decoder will try to process the Base Layer (BL) and to improve it through RPU metadata by discarding the EL layer.
however, the Base Layer is non-standard (it has a different color space than the IPTPQc2/IPT one) as extracted from a Bluray disk.
It should appear better than an HDR stream (since metadata is dynamic and non-static) but will not reach its maximum splendor due to the lack of data contained in the EL layer.
The lack of this information, however, could lead, in some cases, to evident chromatic aberrations.
[Oppo UDP-203, ATV 4K] —> Anthem MRX-720 —> LG OLED65E7P

aboulfad
Posts: 111
Joined: Sat Feb 09, 2019 12:51 pm

Re: Dolby Vision now possible through MP4 Mux.

Post by aboulfad » Thu Dec 24, 2020 11:56 am

After processing some old broken links I only found this excerpt from THIS on RPU and EL- it is unclear what are the differences in metadata between RPU vs EL ?

When encoded using the dual-layer scheme, the Dolby Vision stream consists of a base layer, an enhancement layer, and a reference processing unit:
...
• The enhancement layer carries the color and brightness differences between the graded source (that is, a source that is graded to Dolby Vision standards) and base-layer version of the source.
• The reference processing unit is a special Network Abstraction Layer (NAL) unit that contains the Dolby Vision metadata. It is also a complete Dolby Vision metadata access unit for the current picture.
...
[Oppo UDP-203, ATV 4K] —> Anthem MRX-720 —> LG OLED65E7P

Post Reply