HDR10+ General Discussion

[quietvoid]

Originally Posted by Nico8583

Thank you, I get only 2 FPS when I use it (I use copy to video codec, I don't recode it).
Could you give me the ffmpeg version you use and the command line ? Thank you.

Zeranoe FFmpeg builds should work, there's an example CLI:

ffmpeg -i "input.mkv" -c:v copy -vbsf hevc_mp4toannexb -f hevc - | hdr10plus_parser.exe -

[Nico8583]

Thank you, I just replaced an old ffmpeg with the newest version and now I get 500 fps instead of 50 fps...

[mini-moose]

double post by mistake. can be deleted.

[mini-moose]

hi quietvoid

I was playing around with a recent hdr10+ disc I got.

The tool dialog added something called "Mastering peak flag enabled":

Parsing HEVC file for dynamic metadata...
[00:02:25] 100%
Reading parsed dynamic metadata... Mastering peak flag enabled
Mastering peak flag enabled
Mastering peak flag enabled
Mastering peak flag enabled
Mastering peak flag enabled
Done.
Writing metadata to JSON file... Done.

the "Mastering peak flag enabled" repeats about 40-50 times.

Then I tried to serve the json to x265 and encode the vid.
Looked at the MediaInfo report and it shows just this:

HDR format : SMPTE ST 2094 App 4, Version 1

On previous encodes I did show up like this:

HDR format : SMPTE ST 2094 App 4, Version 1, HDR10+ Profile A compatible

or

HDR format : SMPTE ST 2094 App 4, Version 1, HDR10+ Profile B compatible

From the little I understand, it appears it can't recognize it as hdr10+

[quietvoid]

Originally Posted by mini-moose

hi quietvoid

I was playing around with a recent hdr10+ disc I got.

The tool dialog added something called "Mastering peak flag enabled":

Parsing HEVC file for dynamic metadata...
[00:02:25] 100%
Reading parsed dynamic metadata... Mastering peak flag enabled
Mastering peak flag enabled
Mastering peak flag enabled
Mastering peak flag enabled
Mastering peak flag enabled
Done.
Writing metadata to JSON file... Done.

the "Mastering peak flag enabled" repeats about 40-50 times.

Then I tried to serve the json to x265 and encode the vid.
Looked at the MediaInfo report and it shows just this:

HDR format : SMPTE ST 2094 App 4, Version 1

On previous encodes I did show up like this:

HDR format : SMPTE ST 2094 App 4, Version 1, HDR10+ Profile A compatible

or

HDR format : SMPTE ST 2094 App 4, Version 1, HDR10+ Profile B compatible

From the little I understand, it appears it can't recognize it as hdr10+

Hi, thanks for the feedback.

That part of the metadata was not implemented, I can't remember whether it was in the specs sheet I was using.

If it was, I did note that it was not supposed to be enabled in the version's specs.

So it is possible there is a new version of the spec, where some of these flags are actually being used now.

I had received some feedback regarding a disc failing to parse recently, but I have not had the time to look into it.
Can you share the title? If I remember correctly it was Shaun of the Dead.

Thanks.

Edit: I have done a bit of research and it seems the HDR10+ LLC alliance has published a press release on 2019/09/04.

Released this month is an updated Technical Specification to respond to industry’s request of new device categories and codecs as well as a comprehensive whitepaper to introduce the technology and it’s benefit to new adopters

So there's likely no access for now, even x265 will likely need to update their implementation.

[mini-moose]

Originally Posted by quietvoid

I had received some feedback regarding a disc failing to parse recently, but I have not had the time to look into it.
Can you share the title? If I remember correctly it was Shaun of the Dead.

Thanks.

Edit: I have done a bit of research and it seems the HDR10+ LLC alliance has published a press release on 2019/09/04.

I figured out the issue with x265, but like you said, the json made with this new hdr10+ profile may not have all that is required to work properly.

[quietvoid]

Originally Posted by mini-moose

I figured out the issue with x265, but like you said, the json made with this new hdr10+ profile may not have all that is required to work properly.

Turns out your issue was an edge case.

Updated the parser to 0.2.4

Fix edge case when AverageRGB, MaxScl are 0 but not targeted_system_display_maximum_luminance.
Add asserts to avoid invalid data.
Error when the dynamic metadata is from a (possibly) unsupported version.

https://github.com/quietvoid/hdr10plus_parser/releases

[mini-moose]

Originally Posted by quietvoid

Turns out your issue was an edge case.

Updated the parser to 0.2.4

https://github.com/quietvoid/hdr10plus_parser/releases

thanks for the update!

Edge case you mean why I couldn't get x265 to identify hd10+ without cutting out the first second?

[quietvoid]

Originally Posted by mini-moose

thanks for the update!

Edge case you mean why I couldn't get x265 to identify hd10+ without cutting out the first second?

The edge case was present in the first second so the JSON was most likely invalid for the first second.
So now if the parser doesn't error, the metadata should be valid.

[PapitaHD]

Hi all,
First of all let me thank quietvoid for the hdr10+ parser, works great and very easy to use!
My only issue is the same that mini-moose mentioned a few posts back. I'm trying to encode the quite recent American Gangster BD which has hdr10+. I used the parser as advised to get the json file with the metadata, but after encoding the movie, the resulted file doesn't have the hdr10+ flag in mediainfo. It shows: SMPTE ST 2094 App 4.
What can be the issue, maybe the parser needs another update?
Thanks in advance!

[quietvoid]

Sorry, I hadn't seen your post.
As far as I'm aware, HDR10+ is a format in conformance with SMPTE ST 2094 App 4.

MediaInfo sometimes includes a HDR10+ profile compatibility, but I have no idea what they mean.
The "HDR10+ profile A/B compatible" flag seems to be lost after reencode sometimes, however the presence of ST2094 App 4 metadata is still detected.
It mostly happens with titles released after the September 4 2019 specifications release.

If anyone has information regarding the HDR10+ profiles, it would be useful.
Otherwise I don't really know what is missing for HDR10+ profiles compatibility.

[mini-moose]

Originally Posted by quietvoid

MediaInfo sometimes includes a HDR10+ profile compatibility, but I have no idea what they mean.
The "HDR10+ profile A/B compatible" flag seems to be lost after reencode sometimes, however the presence of ST2094 App 4 metadata is still detected.
It mostly happens with titles released after the September 4 2019 specifications release.

Maybe another edge case like mine had?

[quietvoid]

Well, the metadata gets parsed correctly.
There might be some data missing (in the JSON or after x265's encode) but without the new specifications or an updated example JSON I don't know what the difference would be.

[quietvoid]

Sorry for the double post.
I've tested with the x265 example and it always has HDR10+ Profile B compatible.
Turns out it's not being flagged if there is no BezierCurveData object in the JSON.
So added it (it used to only be added when there is actual data for it), and refactored the code to be more readable. Also improved the JSON to be properly generated.

Fixed the cargo run not working, tested that all the input/output arguments work just like before. Hopefully nothing breaks.

Updated to 0.2.5
https://github.com/quietvoid/hdr10plus_parser/releases

[PapitaHD]

Thanks a lot, it's working perfeclty now!
I encoded The Shining and American Gangster with hdr10+ successfully and I'm gonna test it on Ad Astra as soon as it's out.

[quietvoid]

Further more, I just stumbled across this in the September 2019 White Paper from https://hdr10plus.org/wp-content/upl...WhitePaper.pdf

No matter the workflow used, HDR10+ technology can support the full range of HDR standards to 10,000 cd/m2, 8K and BT.2020 color gamut.
Being resolution agnostic, metadata needs to be created only once and can be applied to any target resolution.

So it seems there should be no problem with cropping and/or resizing.

[nevcairiel]

Originally Posted by quietvoid

So it seems there should be no problem with cropping and/or resizing.

Actually with cropping you would need to adjust the metadata, but resizing should be fine, as long as the relative dimensions and aspect ratio stay the same (but to change that you would probably crop anyway).

[quietvoid]

Originally Posted by nevcairiel

Actually with cropping you would need to adjust the metadata, but resizing should be fine, as long as the relative dimensions and aspect ratio stay the same (but to change that you would probably crop anyway).

I still think there's nothing relevant to the resolution/aspect ratio in the current version.
The Bezier curve data specifies info for the curve generation only.
The processing windows (which do mention pixel coordinates) are not in this version.
The only thing that might need adjustment is the average/max RGB values if they were calculated with the black bars not accounted for.

Unless I'm missing something..

[nevcairiel]

Oh, I didn't check what might be in there, I only remember from implementing HEVC metadata parsing that it contains the processing windows - on second thought they would probably also need adjustment for scaling, but thats easier since you just have to scale their coordinates.
But even then, resizing the processing windows is trivial, what the Whitepaper refers to would indicate actually having to determine full new metadata, which can be avoided that way.

[jlpsvk]

just a question... shoudn't be then all HDR video encoded without cropping? average CLL and MAX-CLL are calculated with the black bars, if I am right. so without it, it could cause playing somehow different, as with black bars... it's just my quess...