[DDVT Tool] Dolby Vision RPU Demuxing / Injecting / Editing.

[quietvoid]

The issue is most likely that the original MKV had the transfer set in the container.

[Hansolo]

Originally Posted by Hansolo

Hello!

I did some test injecting with original metadata and modified metadata (removing some frames), and i cant see any difference seeing the film.

I thought that using modified metadata, the image will flash or similar when changing scene but film goes well.
What is supposed to be?

Regards!

Nobody cant tell me what will happen when injecting differente metada?

[GodzilaAvenger]

You've given us very little information, as the answer to your question is completely dependent on the DV RPU, and your eyes. E.g. if brightness across your RPU is mostly uniform, there is a high chance you won't notice the difference. If there is e.g. one scene that has 10 nits brightness and the next 1000 nits or the other way around and you drop a few frames, you MAY notice weird brightness changes for a brief period before a scene cut.

[Hansolo]

Originally Posted by GodzilaAvenger

You've given us very little information, as the answer to your question is completely dependent on the DV RPU, and your eyes. E.g. if brightness across your RPU is mostly uniform, there is a high chance you won't notice the difference. If there is e.g. one scene that has 10 nits brightness and the next 1000 nits or the other way around and you drop a few frames, you MAY notice weird brightness changes for a brief period before a scene cut.

Thanks for your reply!

This is the info regarding the file...

ídeo
ID                                       : 1
Formato                                  : HEVC
Formato/Info                             : High Efficiency Video Coding
Formato del perfil                       : Main 10@L5@High
HDR format                               : SMPTE ST 2094 App 4, Version 1, HDR10+ Profile B compatible
ID códec                                 : V_MPEGH/ISO/HEVC
Duración                                 : 1 h 48 min
Tasa de bits                             : 24,5 Mb/s
Ancho                                    : 3 840 píxeles
Alto                                     : 1 634 píxeles
Relación de aspecto                      : 2,35:1
Modo velocidad fotogramas                : Constante
Velocidad de fotogramas                  : 23,976 FPS
Espacio de color                         : YUV
Submuestreo croma                        : 4:2:0 (Type 2)
Profundidad bits                         : 10 bits
Bits/(píxel*fotograma)                   : 0.163
Tamaño de pista                          : 18,6 GiB (77%)
Título                                   : The Thing
Idioma                                   : Español
Default                                  : Sí
Forced                                   : No
Rango de color                           : Limited
Colores primarios                        : BT.2020
Características transferencia            : PQ
Coeficientes matriz                      : BT.2020 non-constant
Principios color monitor masterización   : Display P3
Luminancia monitor masterización         : min: 0.0050 cd/m2, max: 1000 cd/m2
Nivel ligero contenido máximo            : 1034 cd/m2
Nivel ligero fotograma medio máximo      : 794 cd/m2

I dont know what you told me about RPU.
The last test i did was to remove 60 frames from the metadata.json and injected into a 1minute clip with brighter scenes and darker scenes for seeing the difference as you told me, but i dont see anything.

Im starting to thing that the hdr10+ is a fake. Not sure, so i upload the 1minute clip with original metadata, not modified.

https://mega.nz/file/201QnYZD#HQd9z8-kFHJ47HpgsCDjOkiOirKi7SF8JFs9X96RGR0

Maybe you can tell me anything.
Regards!

[Amateur]

Is it possible to run the Hybrid tool to work in a batch to make hybrids from TV episodes? One folder of HDR files and another folder of DV files?

Originally Posted by puremind

You can use the tool in posted above, very simple instructions. You don't need to do much for the file you have. It is more complicated to transform HDR layer into DV readable, but here you already have the DV layer baked into.

This is possible? I've had a couple instances where the Dolby stream is higher bitrate than HDR and wondered if I could add the HDR info to the Dolby stream.

[GodzilaAvenger]

Originally Posted by Hansolo

Thanks for your reply!

This is the info regarding the file...

ídeo
ID                                       : 1
Formato                                  : HEVC
Formato/Info                             : High Efficiency Video Coding
Formato del perfil                       : Main 10@L5@High
HDR format                               : SMPTE ST 2094 App 4, Version 1, HDR10+ Profile B compatible
ID códec                                 : V_MPEGH/ISO/HEVC
Duración                                 : 1 h 48 min
Tasa de bits                             : 24,5 Mb/s
Ancho                                    : 3 840 píxeles
Alto                                     : 1 634 píxeles
Relación de aspecto                      : 2,35:1
Modo velocidad fotogramas                : Constante
Velocidad de fotogramas                  : 23,976 FPS
Espacio de color                         : YUV
Submuestreo croma                        : 4:2:0 (Type 2)
Profundidad bits                         : 10 bits
Bits/(píxel*fotograma)                   : 0.163
Tamaño de pista                          : 18,6 GiB (77%)
Título                                   : The Thing
Idioma                                   : Español
Default                                  : Sí
Forced                                   : No
Rango de color                           : Limited
Colores primarios                        : BT.2020
Características transferencia            : PQ
Coeficientes matriz                      : BT.2020 non-constant
Principios color monitor masterización   : Display P3
Luminancia monitor masterización         : min: 0.0050 cd/m2, max: 1000 cd/m2
Nivel ligero contenido máximo            : 1034 cd/m2
Nivel ligero fotograma medio máximo      : 794 cd/m2

I dont know what you told me about RPU.
The last test i did was to remove 60 frames from the metadata.json and injected into a 1minute clip with brighter scenes and darker scenes for seeing the difference as you told me, but i dont see anything.

Im starting to thing that the hdr10+ is a fake. Not sure, so i upload the 1minute clip with original metadata, not modified.

https://mega.nz/file/201QnYZD#HQd9z8-kFHJ47HpgsCDjOkiOirKi7SF8JFs9X96RGR0

Maybe you can tell me anything.
Regards!

I thought you were talking about Dolby Vision, the RPU I was refering to was for that.

For your case, go into the tools folder of DDVT on cmd or Powershell and run

hdr10plus_tool plot metadata.json

This will plot the min/avg/max brightness for each frame and should give you an idea of what the metadata contains.

[hidef_rec]

My very small contribution to helping less techie DV/HDR fans use this great tool to make hybrid videos:
https://youtu.be/FMqhb7ppGWY

[asarian]

I'm getting the following error on a movie with the following characteristics: HDR Info = [HDR10, Dolby Vision Profile 7 (FEL)]. I am trying to create an RPU file:

Please wait. Demuxing EL...
frame=197674 fps=2047 q=-1.0 Lsize=49958925KiB time=02:17:24.52 bitrate=49640.6kbits/s speed=85.4x
EL Done.

Please wait. Demuxing DV Reference Processing Unit...
Error: No frames parsed!
Done.

[Amateur]

Originally Posted by hidef_rec

My very small contribution to helping less techie DV/HDR fans use this great tool to make hybrid videos:
https://youtu.be/FMqhb7ppGWY

The other guy hasn't replied but maybe you can help me with this

http://forum.doom9.org/showthread.ph...61#post2017261

I know how to convert DV5 to DV8 and inject that into an HDR stream. But the above post that I quoted makes it sound like you can use a DV stream as your base and inject the HDR info into that maybe. Am I understanding that correctly?

Also, are there any drawbacks in creating DV info from an HDR10+ file? Is that considered fake DV and not quality?

[GodzilaAvenger]

The color space of the video stream of a P5 DV is different from P8. P8 is backward compatible with HDR, P5 is not, and injecting DV metadata will not change that.

[asarian]

Originally Posted by asarian

I'm getting the following error on a movie with the following characteristics: HDR Info = [HDR10, Dolby Vision Profile 7 (FEL)]. I am trying to create an RPU file:

Please wait. Demuxing EL...
frame=197674 fps=2047 q=-1.0 Lsize=49958925KiB time=02:17:24.52 bitrate=49640.6kbits/s speed=85.4x
EL Done.

Please wait. Demuxing DV Reference Processing Unit...
Error: No frames parsed!
Done.

Anyone have an answer to this? Thanks.

[Amateur]

Originally Posted by puremind

A while ago, I had shared a mass convert script adapted from yours so that it would do all files in a given folder sequentially...

Is it possible to apply this method to the Hybrid feature of the tool using two folders? When it first asks for the HDR file, a folder can be dropped into the window (hdr1.mkv, hdr2.mkv, hdr3.mkv). Then when it asks for the DV file, a second folder can be dropped into the window (dv1.mkv, dv2.mkv, dv3.mkv). Then the tool will sequentially convert the DV profile to 8.1 and inject into the HDR file automatically? (The automatic starting of each process might also need to be changed because it currently needs user input to type S to start)

This would really help me for TV series.

Thank you

[ed_co]

Hello guys,

I am fairly new with adding DV/HDR10+ to encodes (I did just around a dozen now). I wanna thank the creator/s of the scripts for such a nice tool.

I have 2 files of the same movie.

The file 1.mkv:
- HDR10. Mediainfo:
SMPTE ST 2086, HDR10 compatible
- Size/quality is good. A little bit more yellowish tinted than the 2.mkv
- Resolution: 3840x1608
- Framerate: 23.976
- Has a delay of 198 frames respecting the following movie file

The file 2.mkv:
- HDR10+, DV. Mediainfo:
Dolby Vision, Version 1.0, Profile 8.1, dvhe.08.06, BL+RPU, no metadata compression, HDR10 compatible / SMPTE ST 2086, Version HDR10, HDR10 compatible / SMPTE ST 2094 App 4, Version HDR10+ Profile B, HDR10+ Profile B compatible
- Size is the double.
- Resolution: 3840x1606 (2 pixels less in height than the previous one)
- Framerate: 23.976

That 2 pixels height difference is weird. There are no black bars to any of them, and is imperceptible to know what the exact mismatch is. I don't know if it is:
- A crop thing: one was cropped more than the other.
- Aspect ratio management: they were cropped the same but the ratio was modified.
- Maybe the two movies, as they are probably from different releases of different countries, they came cropped differently.
In any case I don't know which of the three possibilities is.

As you can guess, I want to put DV and HDR10+ of 2.mkv in 1.mkv, so I proceeded in the following way:

1) Extract DV and HDR10+ from 2.mkv, steps:
- Drag and drop 2.mkv to DDVT_DEMUXER.cmd
It found: HDR Info = [HDR10, HDR10+, Dolby Vision Profile 8]
- So I guess, I must choose: SAVE BL[NO], SAVE RPU [YES], CONVERT RPU [PROFILE 8.1 HDR10], CROP RPU [NO], Remove HDR10+ Metadata from BL [NO], SAVE HDR10+ [YES]
- Then I started the process and I get the .bin and .json files.

Are these options correct? Do I have to change anything in my selection? Please, let me know.

2) Inject DV and HDR10+ in 1.mkv
- Drag and drop 2.mkv to DDVT_DEMUXER.cmd
- RPU file: Drag and drop the .bin file
- HDR10+ file: Drag and drop the .json file
- CUSTOM EDIT FILE: I just press ENTER (I don't even know what it is)
- Now I get this.
DDV Tool cleaned.png

The only thing I have clear is that I have to put 198 frames in the delay. Then I don't know what more to do and haven't continued.

- Don't know what is RPU L6, and don't know if I have to change anything in options 3 or 4.
- Don't know if I have to EDIT ACTIVE AREA as I don't have to crop it and the area of the target movie is 2 pixels higher than the DV/HDR10+ movie. It just detects a crop 1 in the bottom in RPU input... (which IMHO I don't think it should be cropped in any way...)

Maybe could be stupid, but I don't know how to proceed.

Can anyone help me?

Thank you very much,

E.

[GodzilaAvenger]

Hi,

Generally speaking, you don't need to do anything about L6 metadata, because the one in the RPU usually doesn't even matter. But if you want, you can use the FILEINFO tool with the RPU you have (enter a random frame number, I usually go for 12345) to see what the L6 values for the RPU are, and you can open the mkv in mediainfo to see what those values are for the video (frame average light level, etc.) Then you can see if they are different and decide if you want to match them.

As for cropping, based on what you have said, you're good to go, you don't need to change anything.

[ed_co]

Originally Posted by GodzilaAvenger

Hi,

Generally speaking, you don't need to do anything about L6 metadata, because the one in the RPU usually doesn't even matter. But if you want, you can use the FILEINFO tool with the RPU you have (enter a random frame number, I usually go for 12345) to see what the L6 values for the RPU are, and you can open the mkv in mediainfo to see what those values are for the video (frame average light level, etc.) Then you can see if they are different and decide if you want to match them.
.

I have no idea about what L6 metadata is.
I don't know what is matching what.
Could you please develop a little bit, so I could understand and I could decide what to do?

Originally Posted by GodzilaAvenger

As for cropping, based on what you have said, you're good to go, you don't need to change anything.

The problem I see is that even though does not need any crop, it crops RPU, even for 1 pixel height (which it shouldn't crop anything at all as the RPU source file is shorter than the target file, and I think is a bug).

I kept going with the script and continued the injection, adding the 198 frames delay. And it crops the RPU based on what I see in the script:

injection.png

So, I guess if this is a bug, it should be great to be fixed.

[ed_co]

Apart from my previous reply (please @GodzilaAvenger or any other, reply if you can), I was thinking about another plan to avoid the RPU not match with VIDEO.

I just got another 3.mkv file, which is a UHD remux, and contains DV (not HDR+ though).
Obviously the resolution is not cropped: 3840x2160
So, I could add it to the 1.mkv, which has a resolution of 3840x1608.

So the steps for new plan could be:
1- Extract DV from 3.mkv (with CROP enabled or disabled? I did it with enabled)
2- Use the HDR10+ JSON file that I already extracted from 2.mkv
3- Inject DV from 3.mkv and HDR+ from 2.mkv to 1.mkv

By the way, extracting the 3.mkv I realised the size of the RPU.bin is much different from 2.mkv (28.3 vs 22.1), and then I realize the
2.mkv --> RPU .bin file size 28.3Mb --> RPU Info = [Dolby Vision Profile 8] [DM = CM v4.0]
3.mkv --> RPU .bin file size 22.1Mb --> RPU Info = [Dolby Vision Profile 8] [DM = CM v2.9]

What is this CM v4.0 vs v2.9?? Can anyone explain to me? The more size is better/more quality?

Thanks

[GodzilaAvenger]

CM refers to Dolby Vision Content Mapping, v4.0 is a newer, expanded version of v2.9, so it makes sense for the file to be larger.

As for L6 (and DV metadata labels in general), see here.

As for the cropping, first of all, I doubt being 1 pixel off on one side would make any difference noticeable to the human eye. Second, in general you want the RPU borders to match the video. Your remux probably has ~280 for the top and bottom borders, when you are injecting it you would want to set all those to zero.

[ed_co]

Originally Posted by GodzilaAvenger

CM refers to Dolby Vision Content Mapping, v4.0 is a newer, expanded version of v2.9, so it makes sense for the file to be larger.

As for L6 (and DV metadata labels in general), see here.

Thanks for the info.

Originally Posted by GodzilaAvenger

As for the cropping, first of all, I doubt being 1 pixel off on one side would make any difference noticeable to the human eye.

Still a bug: if we are injecting RPU from a 3840x1606 video into a 3840x1608 video, the RPU shouldn't be cropped in any way, right?

Anyway, there is something more I will comment later, as this bug looks to be more complicated.

Originally Posted by GodzilaAvenger

Second, in general you want the RPU borders to match the video. Your remux probably has ~280 for the top and bottom borders, when you are injecting it you would want to set all those to zero.

I don't understand what you mean with this one, as I never wanted to inject anything in the remux.

Just wanted to extract the RPU from the remux, which the RPU was cropped at RPU extraction, and no problems at injecting time in 1.mkv having a message of [MATCH WITH VIDEO], and not like the RPU from the 2.mkv video (3840x1606) as was saying [NOT MATCH WITH VIDEO].

I have some new findings that I will post later.

[GodzilaAvenger]

Does the 1608 video fill out all pixel rows? Or are the top and bottom pixel rows black? If it is the former, RPU top & bottom border should be 0, if the latter, 1.

As for the remux, what I meant was that when you extract the remux RPU the borders in it are probably set to ~280 (if the film has standard cinematic aspect ratio). When you want to inject said RPU into a cropped video (e.g. 1606) then you should set the borders to 0.

I recommend looking at the L5 description in the link in my previous post. The top/bottom borders are basically the amount of "Letterbox/Blanking".

[ed_co]

Originally Posted by GodzilaAvenger

Does the 1608 video fill out all pixel rows? Or are the top and bottom pixel rows black? If it is the former, RPU top & bottom border should be 0, if the latter, 1.

Is totally filled out, as I explained in detail in a comment before:

Originally Posted by ed_co

That 2 pixels height difference is weird. There are no black bars to any of them, and is imperceptible to know what the exact mismatch is. I don't know if it is:
- A crop thing: one was cropped more than the other.
- Aspect ratio management: they were cropped the same but the ratio was modified.
- Maybe the two movies, as they are probably from different releases of different countries, they came cropped differently.
In any case I don't know which of the three possibilities is.

Originally Posted by GodzilaAvenger

As for the remux, what I meant was that when you extract the remux RPU the borders in it are probably set to ~280 (if the film has standard cinematic aspect ratio). When you want to inject said RPU into a cropped video (e.g. 1606) then you should set the borders to 0.

I guess it depends how you extract the RPU:
- If you extract it cropped you add it -in a cropped video- with 0
- If you extract it not cropped you add it -in a cropped video- with 280
This is quite clear to me. I extracted both ways and I think it is how it works.

As a reminder, the bug [NOT MATCH WITH VIDEO] just sets bottom 1, and not top.

Originally Posted by GodzilaAvenger

I recommend looking at the L5 description in the link in my previous post. The top/bottom borders are basically the amount of "Letterbox/Blanking".

Yep I read it, and I understand it. But it is not explaining anything about the bug.