[DoVi_Scripts] Multi-Function Scripts for Dolby Vision processing and a lot more...

[TDS]

So just for dummies like me, it might be helpful if you mentioned where those 3 new .py files need to be put...

"new files needed"

[Kuler087]

in the tools folder

[TDS]

Originally Posted by Kuler087

in the tools folder

Thanks, updating everything now...

With this extra :-

::choose osd text color in 7-1  / 7-2  / 7-3 (default = dark grey: $808080)
set osdcolor=$808080

What are the "number codes" for other colours ??

https://html-color.codes/

[Kuler087]

What are the "number codes" for other colours ??

https://html-color.codes/

yes, any color code from that website should work... but instead of # you use $
EG: $3cb300

[TDS]

Hi Kuler,

I just want to run this by you, and get your thoughts, and possible suggestions -

I processed "The Fifth Element" yesterday, and ran it thru DS 8-1-1.

The movie is just over 2hrs:06minutes (181171 frames)

It started it's cm_analyze "countdown" @ approx 6hrs:30minutes, after several minutes, it was counting up, then it settled, then started counting down, (after a couple of hours) and it seemed that it was taking a little bit more time to process each frame.

So it ended up taking well over 4.5 hours to complete, and I was SO relieved that it did finish, and I had a workable file name, and plenty of free space

So a couple of questions, is this time frame is something I need to expect as the norm, with movie length encodes ??

I'd hate to think how long ultra long movies like the extended LoTR & The Hobbit, etc are going to take.

I'm glad I got a 4080, as I'm sure the poor little 3060 would take a LOT longer.

Is there anyway that this process can be sped up, or is it just the way it has to be ??

[Kuler087]

I have no control over the speed with cm_analyze unfortunately. I really wish Dolby fixed that slow performance issue on windows but they haven't updated the tools since February 26, 2024.

[TDS]

I did a really quick & dirty Resolve HDR10 & DV metadata edit today, and using the MP4 script (as I always do), after injecting the .xml, I got an .mp4, then I injected the .json into that file, and I got an .mkv

If I imported the folder containing both the .xml & .json it would state that the video file was only HDR10, and would quit.

Seems I have to inject the .json first, then inject the .xml into the .json video.

[Kuler087]

Yes, I think that if you inject HDR10plus, the script will force the MKV container since the only reason to use the MP4 container is for DV and the reason is that there's no code in my script to mux non-DV input to MP4

try this:
Rename hdr10plus json file the same as the HDR10 BL and input-1 the BL HDR10 video and input-2 the XML. This should inject both metadata in one job but you still might end up with a MKV, I'm not sure.

[ReciprocalUniverse]

Great Work Kuler!!

[Kuler087]

@ReciprocalUniverse thank's

[ReciprocalUniverse]

It is wonderful you are active and involved here and with Resolve. I hope you can help me with some answers. I can't afford the Dolby license for Resolve Studio but I can run the HDR10+ analysis and encode with x265 to get a compatible HDR10+ file. Running some of the DDVT scripts I made it into a playable DV Profile 8.1 (BL + RPU).

1.) With DV, there is a collection of available target RPU's. With HDR10+, you don't get that. So when a script converts HDR10+ json metadata to DV RPU, what exactly is the tone mapping target you end up with? 100 nit? 200 nit? Something else?
2.) If I upload an HDR10 to YouTube, they render a version in SDR. If the file is not streamed, it needs an HDR and SDR copy. A dual layer DV (BL + EL + RPU) would not be of help for a player not supporting DV. However, if it was supporting of DV what would be the point of an SDR layer if the player/display can already do the HDR layer? Part of my question then relates to, what is the special relevance of tone mapping to the 100 nit SDR level for a DV device?

As an aside, I don't think it's visually essential to encode a jpeg2000 sequence or open exr for the source to HEVC. DNxHR or Prores is fine.

**************************************************

Unrelated to the above questions, it came up in the conversation about doing trims on individual frames, and the amount of work involved. I would add that outside of the DV realm, grading a scene clip doesn't have to always involve every single frame or grading just one frame. Example, the bride is wearing a white dress in the bright sun, and when she is followed by the camera into a dark hallway, the scene darkens, the scene ends up with very different exposure than it started with. Another example is an unmanned camera on a tripod filming an outdoor wedding ceremony. Suddenly the sky darkens as clouds move overhead. The scene again ends up with very different exposure than it started with. In my opinion, these are not the intended situations for frame by frame dynamic metadata. The approach I take is to use Resolve keyframes to 'transition' between the changing conditions, averaging across a sequence of frames. Frame by frame would take too long. Carry on and thanks again!

[Kuler087]

I can't afford the Dolby license for Resolve Studio

DV is free for any user to use. Only the manual level 8 or 2 artistic trims require a license.

With HDR10+, you don't get that. So when a script converts HDR10+ json metadata to DV RPU, what exactly is the tone mapping target you end up with? 100 nit? 200 nit? Something else?

I don't know much about HDR10plus but in DV only the trim passes have targets which are used to tonemap the image more accurately according to the colorist's shot-by-shot adjustments.
HDR10plus doesn't have trim passes, so it most likely works like DV without trims and uses the TV internal target brightness to apply tone mapping according to the metadata.

A dual layer DV (BL + EL + RPU) would not be of help for a player not supporting DV.

The metadata in DV can be used on any display, HDR10 and SDR. This is done by forcing the player to decode the image + metadata (LLDV) and mapping the tone according to the EDID target brightness.

what would be the point of an SDR layer if the player/display can already do the HDR layer?

The 100nits SDR DV trim pass is extrapolated and can tone map to any target between 100-10 000nits. The point of doing the SDR trim pass(maximum compression an HDR grade can get) is that the colorist can control the tone mapping regardless of the display capabilities.

After grading the HDR master shot by shot, they do the automated DV metadata generation and then they monitor shot by shot(again) the mapping to different targets (100-600-1000-2000). They make adjustments to the trims target if they are not satisfied by Dolby default mapping.
They almost never do more than a 100nits trim since it's the lowest point of tone mapping possible but I've seen a couple of bluray that cancel the 100-600-1000nits interpolation(They do the 100nits trim and reset to 0 the 600nits-1000nits values which create a blank 600 trim). This is useful when they want to be more aggressive with the 100nits trim that will be used for the SDR bluray delivery without affecting the 600-1000 target interpolation mapping.

Unrelated to the above questions, it came up in the conversation about doing trims on individual frames, and the amount of work involved.

I've checked pretty much all the DV releases on this planet and only one time I've seen frame by frame metadata and it was for the Amazon show Jack Ryan. But they did not do any trim pass. I cant imagine how long it would take to grade a movie frame by frame.

[ReciprocalUniverse]

Originally Posted by Kuler087

DV is free for any user to use. Only the manual level 8 or 2 artistic trims require a license.

Yes, the DV trim panels are enabled with a license only. Without a license Resolve can still export an RPU.XML from an automated analysis of clips but not a DV compliant HEVC file. Perhaps not even with the license can you export compliant HEVC? Are you saying Resolve's licensed involvement stops with enabling the trim panel and generating an RPU.XML, but not generating a RPU.BIN or HEVC? I had assumed previously that with the license you could export compliant, playable DV HEVC files, but that may have been wrong.

I don't know much about HDR10plus but in DV only the trim passes have targets which are used to tonemap the image more accurately according to the colorist's shot-by-shot adjustments.
HDR10plus doesn't have trim passes, so it most likely works like DV without trims and uses the TV internal target brightness to apply tone mapping according to the metadata.

The Resolve DV display targets can be selected without the DV license but not the trim panel. For HDR10+ there are no selectable display targets, so my question to you was what then does it use which you didn't have the answer for, which is okay. I assume it really does not matter because the HDR10+ processing inside the display will use it appropriately.

The HDR10+ trim panel was enabled in Resolve until V.17 when it was disabled at the request of Samsung. Otherwise, the process workflow of running an automated analysis is identical. A json sidecar file is generated with dynamic metadata that can be interleaved in x265 with the HDR10 BL to produce a playable HDR10+ HEVC file. The x265 process for DV is similar except the RPU has to be in binary form, requiring the extra step from CMD tools (or yours) to convert the RPU.XML to binary.

Okay so the one, 100 nit target level is potentially sufficient to accommodate any DV display.

I've checked pretty much all the DV releases on this planet and only one time I've seen frame by frame metadata and it was for the Amazon show Jack Ryan. But they did not do any trim pass. I cant imagine how long it would take to grade a movie frame by frame.

Ignore my prior remarks about frame by frame, dynamic metadata as with Profile 8.4, it was a misunderstanding by me.

In summary, what this all boils down to for me is that HDR10+ only has intended application for Samsung tv's, and if you want to make playable HEVC files for DV tv's with Resolve, you will still need the DoVi tools whether you have a licensed Resolve copy or not. Have I got that right?

[ReciprocalUniverse]

Here's my final question, thanks for your patience!

Using the DoVi scripts and x265 is it possible to make a profile 8.4 playable HEVC from Resolve with no DV license?

I hope I'm clear about that. I'm not talking about manually doing a sequence of trim passes. I'm talking about automatic, frame by frame bitstream insertion of metadata for every single frame, like an i-Phone can do, with no trim passes.

[Kuler087]

Without a license Resolve can still export an RPU.XML from an automated analysis of clips but not a DV compliant HEVC file. Perhaps not even with the license can you export compliant HEVC?

Resolve hevc encoding sucks anyway so just export the XML and inject it with dovi_scripts workflow 1 but yes, a DV license will enable the P5/P8 hevc encoding in resolve delivery page. The best option is to deliver to a lossless codec and do the encoding with x265.exe and inject the xml/rpu.
FYI If you know how to change your edid, you can use this custom one that will allow P5/P8 encoding without a DV license.

Resolve P5/P8 encoding though, will ignore your existing shot by shot analysis and create new frame by frame metadata.

or HDR10+ there are no selectable display targets, so my question to you was what then does it use which you didn't have the answer for, which is okay. I assume it really does not matter because the HDR10+ processing inside the display will use it appropriately.

The HDR10+ trim panel was enabled in Resolve until V.17 when it was disabled at the request of Samsung. Otherwise, the process workflow of running an automated analysis is identical.

I'm no expert with HDR10plus but I think the process is 100% automated and is never monitored unlike DV trims creation. So there's no target AFAIK, it just creates shot by shot brightness metadata but I could be wrong.

Using the DoVi scripts and x265 is it possible to make a profile 8.4 playable HEVC from Resolve with no DV license?

My script cannot encode the HLG base layer but if you input any XML with an HLG video in workflow 1, the XML file will be converted to an 8.4 RPU and injected into your HLG BL. I've done it many times with BBC TV shows.

EG. You convert the HLG file to HDR10 prores with workflow 8-2-3, generate an XML as usual, and inject it with Workflow 1.

[ReciprocalUniverse]

Originally Posted by Kuler087

Resolve hevc encoding sucks anyway so just export the XML and inject it with dovi_scripts workflow 1 but yes, a DV license will enable the P5/P8 hevc encoding in resolve delivery page. The best option is to deliver to a lossless codec and do the encoding with x265.exe and inject the xml/rpu.
FYI If you know how to change your edid, you can use this custom one that will allow P5/P8 encoding without a DV license.

Resolve P5/P8 encoding though, will ignore your existing shot by shot analysis and create new frame by frame metadata.

I've tried lossless to x265 for P5.0 but without RPU got no DV. Would injecting shot by shot rpu.xml with DoVi tools give me compliant P5.0 DV?

[Kuler087]

P5 encoding from scratch requires the proprietary Dolby engine.

You can do that with Workflow 8-2-6 but you first have to configure your path to DEE.exe. Input can be prores mov, jpeg-2000 mxf, hevc DVP8/7, HDR10, HLG.
For mov, mxf, HDR10 and HLG inputs, you must provide an XML and Dolby default encoding scripts produce garbage 2-pass hevc encoding so you should modify it as I say in workflow 8-2-6 to allow custom settings and CRF encoding.

Also there's a bug in the latest version of dovi_scripts with mov and prores if you don't crop, so get the latest beta.

[TDS]

Originally Posted by Kuler087

P5 encoding from scratch requires the proprietary Dolby engine.

You can do that with Workflow 8-2-6 but you first have to configure your path to DEE.exe. Input can be prores mov, jpeg-2000 mxf, hevc DVP8/7, HDR10, HLG.
For mov, mxf, HDR10 and HLG inputs, you must provide an XML and Dolby default encoding scripts produce garbage 2-pass hevc encoding so you should modify it as I say in workflow 8-2-6 to allow custom settings and CRF encoding.

Also there's a bug in the latest version of dovi_scripts with mov and prores if you don't crop, so get the latest beta.

There are quite a few things you "need" DEE for with DS, BUT, where to get it, is the question.

Shame there's a bug in the latest release, I only just got it the other day, so I guess 3.0.2 might be out soon'ish...

[Kuler087]

There are quite a few things you "need" DEE for with DS

not really... only workflow 8-2-6 8-2-5 and 8-1-3

Shame there's a bug in the latest release, I only just got it the other day, so I guess 3.0.2 might be out soon'ish...

No, not soon. Nothing new and its a minor bug only in 8-2-6 and only when you don't crop and only with prores jpeg-2000 input.

[TDS]

Originally Posted by Kuler087

not really... only workflow 8-2-6 8-2-5 and 8-1-3

Maybe "need" wasn't the right word, it's a "nice to have", and adds extra functionality to the app

No, not soon. Nothing new and its a minor bug only in 8-2-6 and only when you don't crop and only with prores jpeg-2000 input.

OK, well that IS a fairly particular "bug".

All good,

Your 300th post