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

[TDS]

Originally Posted by Kuler087

I don't know, you said:

''8-2-5 SDR to HDR workflow, (on another PC using 2.0.7) but after running that thru Resolve, I couldn't produce a HDR .mp4 or .mkv...just more errors ''

You cant inject dv to prores with my script... the SDR to HDR process requires:
1- convert sdr to hdr bt2020 100nits prores
2- grade hdr in resolve
3- deliver to a lossless HDR codec (jpeg2000 or cineform for example)
4- encode to hevc HDR
5- export dv hdr10plus metadata
6- inject metadata

Aaah, that will make ALL the difference, to now actually know the workflow

I only missed 3 "important steps"...nice.

[TDS]

I am testing a small clip to see what "delivery" options Resolve has, and using JPEG2000 just creates 1,000's of files, what do you do with them ??

Also, I can't find a cineform option, the one that "sounds" close in cineon, and that creates 1,000's of files too.

UPDATE:-I think you probably mean using Quicktime as the format, and under that, the codecs are either GoPro Cineform, or Kakadu JPEG 2000 (remember I am a complete novice, and don't know the "shortcuts")

On to try my luck with blanking (not too hopeful)

=================================================================

I downloaded the latest beta (changed again), and the newest tools pack, and I got this when using 8-2-5, which used AVS, and created a .mov file, but the .mp4 muxing failed (by the looks)

See this:-

https://www.mediafire.com/file/9ap41...nings.txt/file

[Kuler087]

I downloaded the latest beta (changed again), and the newest tools pack, and I got this when using 8-2-5, which used AVS, and created a .mov file, but the .mp4 muxing failed (by the looks)

yea, this workflow is not supposed to mux, but you got the prores HDR file you needed.

[TDS]

Originally Posted by Kuler087

yea, this workflow is not supposed to mux, but you got the prores HDR file you needed.

I ran that .mov thru Resolve, did as much to it as I could understand, exported the .json & .xml, injected them, they don't play on the TV

[Kuler087]

again, you have to encode the Resolve delivery file to hevc

[TDS]

Originally Posted by Kuler087

again, you have to encode the Resolve delivery file to hevc

I had a 5 minute chunk file (.mkv), converted to prores .mov, (8-2-5) loaded it into Resolve, delivered it as a Quicktime, Kakadu JPEG 2000.

Loaded that into RipBot264 (took quite a while to process), so that produced an x265 .mkv.

Exported the .xml & .json files then tried to inject them, which seemed to work, played with PotPlayer, but the TV just wasn't interested, tried it in a little Android box, and they played, but very slow & jittery.

I again also tried blanking, but the clip I was using didn't seem to want to co-operate.

Is scaling the same as blanking ??? it's just that I watched a few other YT clips on blanking, and re-sizing clips.

[Kuler087]

Then you did not encode your HEVC file correctly. I don't know about ripbot and jpeg-2000 but here's a Staxrip template that will work with the cineform codec(the same delivery format I use in the video)
https://drive.google.com/file/d/12XP...usp=drive_link

I again also tried blanking, but the clip I was using didn't seem to want to co-operate.

the clip? the blanking is done in resolve and recorded in the DV XML metadata.

[TDS]

Originally Posted by Kuler087

Then you did not encode your HEVC file correctly. I don't know about ripbot and jpeg-2000 but here's a Staxrip template that will work with the cineform codec(the same delivery format I use in the video)
https://drive.google.com/file/d/12XP...usp=drive_link

I will have another attempt using the Cineform codec, but I did noticed that when encoding the JPEG2000 it wasn't running "smoothly" during the encode.

I'm not very familiar with Staxrip, but I will have a look at the script you sent.

the clip? the blanking is done in resolve and recorded in the DV XML metadata.

Sorry, by "clip" I was referring to the 5 minute chunk video that I was using in Resolve.

The "clip" had very small black bars top & bottom, but I couldn't remove them with blanking, but I could remove them with scaling.

PS:- should I be using the mkv.bat, instead of the mp4 one ??

[Kuler087]

I will have another attempt using the Cineform codec, but I did noticed that when encoding the JPEG2000 it wasn't running "smoothly" during the encode.

On my end, I could encode the jpeg2000 RGB 12bit master with staxrip/x265 but the RGB to YUV conversion with avsresize was slightly off so I'm using the Dolby Encoding Engine (DEE) and the encoded colors are perfect.

If you deliver to cineform and use the staxrip template I sent you, your hevc encode colors will be accurate, like my DEE workflow. I use an old staxrip version so that template might not work with the latest versions. Just copy it in your template folder and import a file, modify metadata/preset/crf if you want and encode. Staxrip is very easy to use.

The "clip" had very small black bars top & bottom, but I couldn't remove them with blanking, but I could remove them with scaling.

you're confusing blanking for DV/HDR10plus and cropping/removal of the letterbox. You will still see the black bars in Resolve when you set your blanking, it is normal. The blanking tells the DV and HDR10plus algorithm to ignore the black bars during the analysis (metadata creation).
The blanking will be recorded in the DV XML metadata which will be used by your TV during playback unless you crop your encode which will force the blanking to 0(because no more black bar to ignore)

PS:- should I be using the mkv.bat, instead of the mp4 one ??

All the script's versions are the same code. The only difference is the muxing when a workflow has been completed.
I'm almost done re-working the injector/editor menu selections. I really like how they work now, fewer user prompts and more straightforward. Direct drag and drop on the bat file is also much improved.

[TDS]

Originally Posted by Kuler087

On my end, I could encode the jpeg2000 RGB 12bit master with staxrip/x265 but the RGB to YUV conversion with avsresize was slightly off so I'm using the Dolby Encoding Engine (DEE) and the encoded colors are perfect.

If you deliver to cineform and use the staxrip template I sent you, your hevc encode colors will be accurate, like my DEE workflow. I use an old staxrip version so that template might not work with the latest versions. Just copy it in your template folder and import a file, modify metadata/preset/crf if you want and encode. Staxrip is very easy to use.

Like I said, I'm not familiar with Staxrip...what version do you use, (2.29) I might try and find it somewhere (videohelp, might have it)

you're confusing blanking for DV/HDR10plus and cropping/removal of the letterbox. You will still see the black bars in Resolve when you set your blanking, it is normal. The blanking tells the DV and HDR10plus algorithm to ignore the black bars during the analysis (metadata creation).
The blanking will be recorded in the DV XML metadata which will be used by your TV during playback unless you crop your encode which will force the blanking to 0(because no more black bar to ignore)

And also, I am not familiar with Resolve either, so it's VERY easy for me to get confused, trying to figure out something new.

I try to follow your tuto's, but because you know what you're doing, and I haven't got a clue, it's a little too fast for me to follow...it will take me some time

All the script's versions are the same code. The only difference is the muxing when a workflow has been completed.
I'm almost done re-working the injector/editor menu selections. I really like how they work now, fewer user prompts and more straightforward. Direct drag and drop on the bat file is also much improved.

More changes, sounds interesting....

Thanks

UPDATE:-

SUCCESS !!!!!!!!!!!!

It's an interesting process, lots of steps, but I have been successful in converting an SDR 1080p to HDR DV / HDR10+

I managed to get your "script" into StaxRip 2.93.3 (which is still oldish), processed that, I tried chunk encoding, but for some weird reason it didn't mux them into a single file, so I combined them with MKVToolNix.

Imported into Resolve, did some basic "fiddle", and did the Detect Scene Cuts, etc, etc (probably only really need the Dolby Vision enabled), exported the json & xml, and played them on the LG, and it worked (didn't have an audio track), but the process has enhanced any defects in the original, which has a lot of small black spotty artifacts jumping around the screen, in certain scenes, might need to find a filter to get rid of those.

Just out of interest, how would I output a HLG version ?? would that be simply a Color Management Setting ??

[Kuler087]

Just out of interest, how would I output a HLG version ??

I don't know

[wyup]

Hi, what do you think about adding a tonemapping mode to the scripts for upscaling brightness of a low-nit movie with libplacebo library, with the goal of getting more brightness on a regular tv?

I.e: A 600-nit graded movie has limited brightness when displayed on a oled tv, even with max brightness setting on the menu. By upscaling peak brightness to 1k, 4k or 10k we get more available brightness for our tvs, even without DV or HDR10+, just using the internal tv tonemapping with relevant HDR10 MaxCLL and MaxMDL metadata.

Alternatively, it would be nice to have a down tonemapping option to downscale a 4k, 6k, 10k nit movie to user input value, let's say 1k nit, using libplacebo's algorithms such as ITU-R Report BT.2446, method A, for which you only need peak brightness for source and destination.

[Kuler087]

I guess I could add the option to deliver the 600 or 1000nits dolby vision trim pass in 8-2-5 (it does 100nits only right now).
As for boosting brightness, I don't believe in any form of automatic/one-click HDR conversion. HDR must be graded shot by shot so this won't happen in DoVi_scripts sorry.

[DMD]

Originally Posted by Kuler087

......HDR must be graded shot by shot so this won't happen in DoVi_scripts sorry.

I think this is the most time-consuming work to do, and manual work is the one that gives the best results. But if there was an algorithm that could do this work automatically, it would be a dream.

[Kuler087]

Originally Posted by DMD

I think this is the most time-consuming work to do, and manual work is the one that gives the best results. .

Yes, indeed. It takes me about 2-3H for the HDR grading and another 2-3H for the Dolby Vision trim pass.

But if there was an algorithm that could do this work automatically, it would be a dream

It already exists but I won't say the name because he's asking money for it and he's sharing my scripts(without my consent) in the pack he sells.

[DMD]

Originally Posted by Kuler087

Yes, indeed. It takes me about 2-3H for the HDR grading and another 2-3H for the Dolby Vision trim pass.



It already exists but I won't say the name because he's asking money for it and he's sharing my scripts(without my consent) in the pack he sells.

I know who you are talking about, it has a disproportionate cost and a business policy that I don't like, with unfair behavior to you as well.
Software development with more reasonable cost in my opinion would have been more popular among enthusiasts.

[TDS]

Originally Posted by Kuler087

Yes, indeed. It takes me about 2-3H for the HDR grading and another 2-3H for the Dolby Vision trim pass.

It already exists but I won't say the name because he's asking money for it and he's sharing my scripts(without my consent) in the pack he sells.

Kuler,

Wow, I can't believe it takes so long to do all that, you must be VERY particular...

So the HDR grading is with Resolve, and the trim pass must be with your tools, correct ???

I haven't done any full length movies yet, only 1 hour +/- TV episodes, but I think the 4080 has made a nice difference to part of the process

Thing is, your grading could be a LOT different to someone else's "eye", for example, if you & DMD processed the same file, with basically the same processes, it could turn out a lot different.

With my communication with DMD (who actually introduced me to DS in the 1st place), he is a very particular encoder....

So is this something that you could work on, and figure out for yourself ??

Is there some way that you could arrange with some or your "friends" to buy this package, and then go from there ??

Or is there a principal in play here ?? I mean, he "stole" from you, you could "borrow" from him.

Just thinking out loud. (tell me if you don't want some of my comments to stay here, I will gladly edit it if you so desire)

[Kuler087]

So the HDR grading is with Resolve, and the trim pass must be with your tools, correct ???

No, everything is done in Resolve. The video you saw includes the DV trim pass creation too. The manual trim starts halfway through the video but remember that the video is sped up 50 (or 100%? I don't remember), I don't go that fast.

Once you're done with the shot-by-shot HDR grading(at 600-1000nits), I run the DV analysis which creates the DV L1 metadata and it also produces a preview of how the tone mapping will look at a 100, 600, 1000 and 2000nits target.
Then I go shot by shot again and do the 100nits SDR trim pass by referencing the original SDR grade(my source) side by side with my HDR grade tonemapped back to SDR and if it differs too much (too dark, too bright, crush, clip or saturation), I adjust the tone mapping with the Dolby Vision controls.
This trim metadata will then be used by the TV to tone map to its capabilities more accurately than if it had L1 only. It will reference what I did in the trims and extrapolate the adjustment to any TV brightness target.


FYI, DV trim pass is done at 100nits SDR only for about 80% of the Movies/TV because it's the maximum compression(lowest point) your HDR grade will get tone mapped. 600-1000nits trims are very rare, studios almost never require them.

Thing is, your grading could be a LOT different to someone else's "eye", for example, if you & DMD processed the same file, with basically the same processes, it could turn out a lot different.

of course

Is there some way that you could arrange with some or your "friends" to buy this package, and then go from there ??
Or is there a principal in play here ?? I mean, he "stole" from you, you could "borrow" from him.

Nah, I could easily do the same without having to buy their stuff. I'm just not interested in any form of one-click SDR to HDR conversion. You could just apply the same HDR settings to all the shots in Resolve and that would be the same.
They did not steal it, anyone can use my scripts. I'm just not a fan of their work and the fact that they sell stuff that should be free. I'm all about sharing...

[TDS]

Originally Posted by Kuler087

No, everything is done in Resolve. The video you saw includes the DV trim pass creation too. The manual trim starts halfway through the video but remember that the video is sped up 50 (or 100%? I don't remember), I don't go that fast.

Once you're done with the shot-by-shot HDR grading(at 600-1000nits), I run the DV analysis which creates the DV L1 metadata and it also produces a preview of how the tone mapping will look at a 100, 600, 1000 and 2000nits target.
Then I go shot by shot again and do the 100nits SDR trim pass by referencing the original SDR grade(my source) side by side with my HDR grade tonemapped back to SDR and if it differs too much (too dark, too bright, crush, clip or saturation), I adjust the tone mapping with the Dolby Vision controls.
This trim metadata will then be used by the TV to tone map to its capabilities more accurately than if it had L1 only. It will reference what I did in the trims and extrapolate the adjustment to any TV brightness target.

FYI, DV trim pass is done at 100nits SDR only for about 80% of the Movies/TV because it's the maximum compression(lowest point) your HDR grade will get tone mapped. 600-1000nits trims are very rare, studios almost never require them.

of course

Nah, I could easily do the same without having to buy their stuff. I'm just not interested in any form of one-click SDR to HDR conversion. You could just apply the same HDR settings to all the shots in Resolve and that would be the same.
They did not steal it, anyone can use my scripts. I'm just not a fan of their work and the fact that they sell stuff that should be free. I'm all about sharing...

I will have to watch that tuto again, slowed down

Like I said, I haven't done any full length movies yet, but I think my workflow would be something like this:-

If it was HDR (or SDR, but so much these days), I would run it thru RipBot264, and do any filtering, "grading", etc,

Then simply use 3-1 to create a DV, which looks so much nicer than HDR.

With SDR stuff (1080p or lesser),
I would still do my first run thru RipBot,
then convert to .mov with 8-2-5 (I think it is),
then run it thru Resolve, do the necessary steps, probably no "grading" or blanking, then deliver as CineForm, (can I get the HDR10 & DV metadata at this point ??)
then load that into StaxRip, using your script, (I'd really like to get a similar script set up for RipBot, (which would have to be for Avisynth) as it could be a LOT faster to process.)
then after that, back into Resolve to get the DV metadata, and inject that with 1-1.
And hopefully end up with "good enough for me" DV HDR .mp4, I can't justify spending so much time on one video

I will have to double check those steps, once I've finished GoT stuff.

[DMD]

Quickly summarizing the procedure to create REC709 SDR 100nits To REC2020 HDR10 Dolby Vision conversion:

1- Creating file with intermediate codec SDR 100 nit (ProRes 422 HQ) via DoVi script (8-2-5).
2- Via Resolve, scene-by-scene HDR detection and classification.
3- Via Resolve, Do.Vi analysis.
4- Via Resolve, Do.Vi xml metadata export.
5- Via Resolve, export file delivery format (Cineform).
6- Plotting HDR10 via DoVi Script (6-2).
7- x265 encoding via StaxRip.
8- Do.Vi. metadata injection via DoVi Script (1-1).

End.

Translated with DeepL.com (free version)