[DoVi_Scripts] Multi-Function Scripts for Dolby Vision processing and a lot more... [Archive] - Page 20

ac777103

16th January 2025, 17:42

Is anyone aware of a list of P7 disc-sourced titles that are CMv4 ? I just looked at the UK release of The Usual Suspects and doviscripts reports it as MEL CM 4. I can't imagine that my Panny UB820 can take advantage of that, but my Ugoos and ATV can. For curiosity I guess there is a plotting workflow that can show what the additional v4 levels are doing?

Kuler087

16th January 2025, 18:15

Is anyone aware of a list of P7 disc-sourced titles that are CMv4 ?

no, but there are a few cmv4.0 BD and all recent movies have cmv4.0 on streaming which can easily be transferred to the P7 rip. (https://www.youtube.com/watch?v=RI_e-MPIPlw)

For curiosity I guess there is a plotting workflow that can show what the additional v4 levels are doing?
You can use the new workflow I posted a few days ago (https://forum.doom9.org/showpost.php?p=2013053&postcount=947). Make a CMV4.0 and CMV2.9 versions and compare them.

ac777103

16th January 2025, 18:52

thanks. Among my modest collection of 81 P7 DV titles, The Usual Suspects is the only one that is CMv4, I'll look out for this in future.

Gatorman3385

19th January 2025, 19:26

I've noticed a similar CPU usage scenario when encoding from HDR to SDR ( x264) as with FEL to SDR (x264). Adding Prefetch(2) to ::tonemap, enabled ≈ 100% CPU usage from ≈ 70%.

::tonemap
if "%SDR%"=="y" echo ConvertBits(16) >> "%TEMP%script.avs"
if "%SDR%"=="y" echo libplacebo_Tonemap(src_csp=%inCS%, dst_csp=0, dst_max=%SDR_target_nits%, dynamic_peak_detection=%peak_detect%, smoothing_period=%smoothing_period%, percentile=%percentile%, tone_mapping_function="%tone_mapping_function%", gamut_mapping_mode="%gamut_mapping_mode%") >> "%TEMP%script.avs"
if "%SDR%"=="y" echo ConvertToYUV420(matrix="709") >> "%TEMP%script.avs"
if "%SDR%"=="y" echo ConvertBits(10) >> "%TEMP%script.avs"
if "%SDR%"=="y" echo Prefetch(2) >> "%TEMP%script.avs"

Kuler087

20th January 2025, 00:14

If anyone is interested in testing a new function? I did a new workflow(7-4) that can create DV metadata comparison videos with 10 frames in loop.
it takes 10 different frames (auto or manual) and makes a loop of 69 frames alternating No metadata/no tone mapping vs L1(L3) only vs L1+ artistic trims

The whole process is very fast, about 5 minutes on my PC and supports batch processing.
So comparing the DV tone mapping is now very easy and effortless.

latest beta: https://drive.google.com/file/d/128gq8aDUTKA_aT7SQsM9dkjA1EP1sosR/view?usp=drive_link
new files to put in the tools folder: https://drive.google.com/drive/folders/18403rSHIaVFkHIFVkO1ZYoIZmZrRKwag?usp=drive_link
I did a couple of samples: https://drive.google.com/drive/folders/1g5I-z_sJmVu-SAIPNiiSlcdMiy2ka0mf

(7-5) I'll also add another similar workflow that compares DV metadata with FEL baked and without FEL. This workflow is not completed yet.

Ok, (7-5) is now completed. You can now create BL vs FEL videos comparison that will show you exactly how FEL looks on device that can't decode it and on device that can.

latest beta: https://drive.google.com/file/d/128gq8aDUTKA_aT7SQsM9dkjA1EP1sosR/view?usp=drive_link
You must update these files again: https://drive.google.com/drive/folders/18403rSHIaVFkHIFVkO1ZYoIZmZrRKwag?usp=drive_link
I did a couple of samples: https://drive.google.com/drive/u/0/folders/1FS42T95TOSpoy4xtwUBIQmziCe_R_IKe

These updated files also fix a bug in 7-4 with input that had a 100 and 1000nits trim (rare).

these are SDR GIFs of HDR output, so ignore the clipping...
https://i.ibb.co/d6JMZQp/Application-Frame-Host-Pw-Xu-RJ91-LB.gif
https://i.ibb.co/6B7JKG2/Application-Frame-Host-PUQo-Dkjc2-Q.gif

Kuler087

20th January 2025, 00:14

I've noticed a similar CPU usage scenario when encoding from HDR to SDR ( x264) as with FEL to SDR (x264).

thanks

spasv

20th January 2025, 16:38

Some help please.
Running DoVi_Scripts_MKV.bat’s workflow(6-2) process Calculating per shot MaxFALL threshold...issues:
Python was not found; run without arguments to install from the Microsoft Store, or disable this shortcut from Settings.
Both python.exe and pythonw.exe are in a same folder. Python.exe opens its Dolby_Scripts_GUI V1.0.0_BETA Window.
And, how to understand run without arguments?
I’ve tried running DoVi_Scripts_MKV.bat without processing file but no result.
Thanks

Kuler087

20th January 2025, 17:41

install python:

https://apps.microsoft.com/detail/9pjpw5ldxlz5?hl=en-US&gl=US

spasv

20th January 2025, 19:45

Thanks.
Probably, the strange behavior I see is caused by dynamic RAM errors.

coopzr

22nd January 2025, 00:08

Hello,
My AVR is unable to play DTS or Dolby TrueHD. It supports up to E-AC3-JOC and Multichannel PCM connected via eARC. It has 5 channels.

If I wish to watch a video with DTS-HD MA or Dolby TrueHD, what is the best way to enjoy as much audio quality as possible?

I understand that Atmos will be lost no matter what I do.

As I see it, my options are:

1. Let Kodi transcode the audio.

2. Using Workflow 8-1-1 (Lossless audio --> E-AC3)
2.1. 7.1 channels @ 1024kbps
2.2. 5.1 channels @ 1536kbps

3. Do something else, perhaps convert to LCPM or FLAC?

Since DTS is open source, is there anything I can do with DTS-HD MA that I cannot do with Dolby TrueHD? Again, the goal is to retain as much quality as possible.

Thanks for your help!

Kuler087

22nd January 2025, 01:16

Kodi should be able to send the audio as lossless PCM so that would be the best option for your setup.

coopzr

24th January 2025, 06:32

Forgive me if this is off-topic. I don't know any other forum or community discussing HDR / Dolby Vision for home cinema. If there are communities out there, please link me to them so I can stop bothering you :p

I have come across two versions of the same movie.
The original source movie is in HDR10. It is my understanding that this HDR metadata is static and applies throughout the entire movie.
The modified version is DV/HDR10+ - "Generated in DaVinci Resolve Studio using the official Dolby Algorithm L5 - 280/280". It appears that an enthusiast has applied dynamic metadata across 2338 scenes.

On paper, the DV/HDR10+ version should be better, right? In the real world, does that mean all DV-generated releases are better than the original?

My questions are: For all movies, and all dynamic DV/HDR10+ versions that enthusiasts have generated, can these versions be worse than the original? Does this defeat the colorist's intent and style choices? Is this process completely automatic, and essentially flawless?

What should I look out for? What should I know about this process?

Thank you very much for any replies. I am very new to this field, as I just got a LG OLED for the first time a week ago.

Kuler087

24th January 2025, 13:03

The ultimate HDR intended look is the 12-bit or 10-bit Master. When the master is brighter than your TV and the bluray or stream encode doesn't have dynamic metadata, it relies on your TV's internal static tone mapping curve, which can defeat the director/colorist intent and style choices a lot more than dynamic HDR.
Unless you do manual trims which require a license, the process is non-destructive when you generate DV metadata. It just tells your TV about each shot's brightness, so the TV can tone map it more accurately and stay closer to the master.

coopzr

24th January 2025, 13:28

I would like to learn more about this stuff, to really understand all the intricacies. Can you please point me to where I can learn about this stuff?

Thank you for taking the time to reply ��

Kuler087

24th January 2025, 14:19

See the DV tone mapping in action with thes (https://drive.google.com/drive/folders/1g5I-z_sJmVu-SAIPNiiSlcdMiy2ka0mf)e and i did some videos on my YT channel (https://www.youtube.com/@RESET_-bg7kw)

Aldebaran_hun

31st January 2025, 14:09

Do you have a list which 4k films have the best picture? I've heard that the Dolby Vision CMv4.0 L8 is the best, like Dune part2 iTunes webdl version.

Kuler087

31st January 2025, 17:03

Many movies look great.
I don't have a list but find movies that are brighter than your TV capabilities in these HDR/DV plot. (https://drive.google.com/drive/u/1/folders/154fBNllwOHL4Lckc7wDV8QKFJwFxnDt-). This is when DV is really useful.

prudentavocado

1st February 2025, 08:40

Thanks Reset 9999, 3 questions for the Dolby_Vision_stuff:

On the "List of DV movies that can't be converted to profile 8 Dolby Vision" does this mean that as long as the DV injected is the one listed, it doesn't matter what HDR source is used?

So 8mile from itunes on that list shows as can't be converted - is that true whether I used the 4k HDR stream from itunes or elsewhere like the 4k HDR stream from UHD disk?

Second question - "PROFILE 7 DOLBY VISION SINGLE-TRACK DUAL-LAYER FEL SUPPORT" - do you recall how you got a file to play on the chromecast with GTV? Everytime I try it falls back to HDR regardless of audio/sub selections.

Third, the comment "Can only be played on a FEL device and can't be converted to P8 without baking FEL into BL" can the proper groups you listed actually bake the FEL into the BL and then it can be played on a p8 device?

p.s., any way to donate to you besides paypal?

Appreciate the assistance.

Kuler087

1st February 2025, 15:19

So 8mile from itunes on that list shows as can't be converted - is that true whether I used the 4k HDR stream from itunes or elsewhere like the 4k HDR stream from UHD disk?

Maybe other streaming services have a different encode but it is unlikely. This movie was mastered at 4000nits and they tone mapped a 1000nits version for the bluray release.

do you recall how you got a file to play on the chromecast with GTV? Everytime I try it falls back to HDR regardless of audio/sub selections.
With plex. The newer GTV chromecast probably dropped P7 support so you need to use the older device.
Anyway, don't use this device, the BL and EL are out of sync and it's doing ''fake'' TV-LED.
Get a Ugoos, you wont regret it, especially with the latest development that allows unaltered RPU processing and Level 5 + positive lift support.

"Can only be played on a FEL device and can't be converted to P8 without baking FEL into BL" can the proper groups you listed actually bake the FEL into the BL and then it can be played on a p8 device?
Yes, these groups bake FEL properly and can be played on any device. Their older encodes might not be though.

p.s., any way to donate to you besides paypal?
no

Vymrr

2nd February 2025, 11:23

Get a Ugoos, you wont regret it, especially with the latest development that allows unaltered RPU processing and Level 5 + positive lift support.

I seem to have problem viewing this setting on the demo files. Which build allows this? CoreElec latest nightly or which CPM build?

Korbus

2nd February 2025, 11:27

This (https://github.com/cpm-code/xbmc/releases/tag/CoreELEC-Amlogic-ng.arm-21.2-Omega_20250128030125.A13) build alone.

ac777103

2nd February 2025, 13:39

I installed the CPM A13 build on my Ugoos yesterday and at first didn't pay attention to the new parameters to force use of source DV levels but today have experimented with the level_5 option and just want to check I understand what's going on.

The first option, set to Y, is for use_source_meta_levels, which by its name sounds global but I guess isn't, and relates to peak brightness to avoid clippling?

the keep_source_meta_level_5 option when set to Y causes OSD masking on my Panny HZ980 when in tv-led mode, but this display does a pretty poor job on tv-led DV anyway as it has a tendency to drop frames (perhaps the processor isn't really up to the job). However, in player-led mode all is good and I've no complaints at all about the picture. Even though it's in player-led mode, am I benefitting from the L5? If so, are there any other devices that also do this, e.g. does Infuse on the ATV account correctly for L5 ?

Is the new option for keep_source_meta_level_6 going to improve HDR10?

Kuler087

2nd February 2025, 15:02

The first option, set to Y, is for use_source_meta_levels, which by its name sounds global but I guess isn't, and relates to peak brightness to avoid clippling?
This option is enabled by default so you don't have to send that cmd.
This setting enables true dv metadata passthrough. It was altering the RPU before that update. I explained everything in this video: https://www.youtube.com/watch?v=HyrA3KmcJBU

the keep_source_meta_level_5 option when set to Y causes OSD masking on my Panny HZ980 when in tv-led mode, but this display does a pretty poor job on tv-led DV anyway as it has a tendency to drop frames (perhaps the processor isn't really up to the job). However, in player-led mode all is good and I've no complaints at all about the picture. Even though it's in player-led mode, am I benefitting from the L5? If so, are there any other devices that also do this, e.g. does Infuse on the ATV account correctly for L5 ?

Yes, L5 is designed to mask the black bars. In the next version, there is an option to keep L5 to 0. This way it doesn't hide anything and it still allow the positive lift trim metadata to work... Without L5, the positive lift is ignored.
player-led ignores L5 but positive lift still works so it seems to be a tv-led bug. Player led mode on this player is only cmv2.9 though.

All the other HDMI devices ignore L5 and positive lift in TV-LED which is why this update is a very big deal.

L6 is useless, you can ignore it.

Vymrr

2nd February 2025, 16:11

This (https://github.com/cpm-code/xbmc/releases/tag/CoreELEC-Amlogic-ng.arm-21.2-Omega_20250128030125.A13) build alone.

Thanks, I installed A13 and also estuary A13 skin. I tried running the demo files listed in the 10 reasons video (https://www.youtube.com/watch?v=HyrA3KmcJBU&t=555s&ab_channel=RESET_9999) but I don't see any difference between L1/L2 and L1/L2 no offset. Only when L1 is on do I see a change, no changes too when running the real content. Also, when running the L5 aspect ratio demo file, nothing changes?? The red screen fills up my entire TV screen during the whole clip and no aspect cut off happens, same as the real content folder. Is there a setting somewhere I'm missing?

Just for reference, my TV is a Hisense U8H and I can run DV CM4.0 movies with no issue.

Kuler087

2nd February 2025, 16:35

Did you enable L5? In A13, it's disabled by default.

https://www.avsforum.com/posts/63803659/

ac777103

2nd February 2025, 17:23

Yes, L5 is designed to mask the black bars. In the next version, there is an option to keep L5 to 0. This way it doesn't hide anything and it still allow the positive lift trim metadata to work... Without L5, the positive lift is ignored.
player-led ignores L5 but positive lift still works so it seems to be a tv-led bug. Player led mode on this player is only cmv2.9 though.

All the other HDMI devices ignore L5 and positive lift in TV-LED which is why this update is a very big deal.

L6 is useless, you can ignore it.

thanks for the explanation. Perhaps my general impression of an improvement since installing A13 is that the RPU is being processed correctly for the first time on the Ugoos.

Will any of my HDR10 titles that have been processed via the 3-1 workflow have any L5 positive lift data, or is that a manual intervention?

Kuler087

2nd February 2025, 17:40

It depends... Does your TV support CMv4.0?

If yes, then there won't be any positive lift in the L8 trim because it's a manual intervention.
if no, then yes, the Dolby vision algorithm will create a small positive lift in the generated L2 600nits trim.

in player-led, the ugoos doesn't support cmv4.0, so yes, there will be a positive lift in the generated L2 600nits trim.

ac777103

2nd February 2025, 17:55

thanks, while my TV supports cm v4 I tend to avoid tv-led playback as it can be glitchy (regardless of DV playback device). So I tend to switch between the ATV and the Ugoos for player-led. Either way, looks like I should stick with the CPM build.

Vymrr

2nd February 2025, 18:00

Did you enable L5? In A13, it's disabled by default.

https://www.avsforum.com/posts/63803659/

Thanks! Got it enabled but it looks like my TV blacks out my OSD and subtitles on the black bars.. got it reverted and waiting for newer cpm version

I got another question not really related to this, but concerning DV movies in general with dark scenes, the dark is really dark and it's really hard to see sometimes, is this considered normal? For instance, for web streaming services (netflix, etc), this problem doesn't seem to exist, and the picture seems to be perfectly balanced in dark scenes where I can still see details.

Kuler087

2nd February 2025, 18:14

There's a black crush issue with cmv2.9 content especially if the RPU MDL is 4000nits. For 1000nits RPU, it's not really noticeable in real content but it is crushed in patterns.
I did some test files about this a while ago: https://drive.google.com/drive/u/1/folders/1OH-DT9jfodpGKX9LwN5FsewWGPBII1Ib

On my C2, HDR10 and DV dark scenes look pretty much identical as long as the RPU is not cmv2.9 4000nits.

ac777103

2nd February 2025, 18:15

thinking some more about player-led playback of HDR10 source files processed via 3-1. The ATV supports CM v4 in player-led, per the explanation in a previous post re positive lift, does this explain my impression that the devices have different brightness profiles in player-led mode? I find them to be more similar for MEL 2.9 content, and more different for my 3-1 conversions.

Vymrr

2nd February 2025, 18:24

Kuler087

2nd February 2025, 18:26

Kuler087

2nd February 2025, 18:28

How about CM4.0? The dark scenes also appear in movies converted to CM4.0 (using your tutorials).

As I said, as long as the RPU is not cmv2.9 and 4000nits MDL, DV dark scenes look identical to HDR10 on my C2.

ac777103

2nd February 2025, 19:30

Kuler087

2nd February 2025, 19:43

I have that disc and in my system the ATV does much better on that test. What we see is a combination of the player and the display and settings on both.

what?? the screenshot is an hdmi capture... No TV is involved...
Unless they fixed something recently, the ATV is broken in cmv4.0 at least with 4000-10000 content ( the blue channel specifically )

https://i.ibb.co/MX9kVz3/firefox-9yg-FEAm9-GF.png

ac777103

2nd February 2025, 20:06

Since they are two different algorithm, there will be a difference between cmv2.9 or cmv4.0 playback but not much. In a cmv4.0 workflow, the v2.9 L1+generated_L2 should roughly equal v4.0 L1+L3

The ATV seems broken with cmv4.0 in the tests I did a while ago. with the pattern specifically: https://drive.google.com/file/d/1s-3GHjwqvuAy7lxehBoBukvz18NqKYia/view?usp=drive_link

I tried this test file and I think it's probably demonstrating how bad my Panny HZ980 is at TV-led DV.

With TV-led DV, from the Ugoos the first 3 or 4 red and green steps are a single block, no grading at all, but the blue channel has all the steps. On the ATV/Infuse all the steps are present in the red channel but there is terrible noise in the 10000 red block. Like the Ugoos, the green channel is one block until it drops to 2500 nits or so. The blue is cleaner on the Ugoos than from the ATV.

In player-led, the ATV loses the step grading in the red as well as the green at the brightest levels, it's not very good. The Ugoos renders a clean pattern with all the steps but of course it's not CMv4.

This test might confirm my impression that picture quality on my TV is degraded in TV-led mode, since neither the ATV nor the Ugoos were able to produce that good a result in TV-led mode. But it might also be showing that player-led CM v4 from the ATV isn't that great either, suggesting that I may as well stick with player-led from the Ugoos until I change my display?

ac777103

2nd February 2025, 20:08

apologies for the confusion, my earlier reply related to a different screenshot you'd posted.

prudentavocado

3rd February 2025, 20:02

Maybe other streaming services have a different encode but it is unlikely. This movie was mastered at 4000nits and they tone mapped a 1000nits version for the bluray release.

With plex. The newer GTV chromecast probably dropped P7 support so you need to use the older device.
Anyway, don't use this device, the BL and EL are out of sync and it's doing ''fake'' TV-LED.
Get a Ugoos, you wont regret it, especially with the latest development that allows unaltered RPU processing and Level 5 + positive lift support.

Yes, these groups bake FEL properly and can be played on any device. Their older encodes might not be though.

no

Thanks

Oh I see.....So, "List of DV movies that can't be converted to profile 8 Dolby Vision" doesn't necessarily mean there is a technical hurdle to where nobody can convert it. If it's one of the proper groups then they could actually do it. (as long as it's not an older encode)

That's a big difference I was missing; I was avoiding all those releases in that red section because I thought it was an impossibility. Why include the BD-FEL ones in the red section then? Seems like you could move those to the green list if we just stick to the proper encoder groups.

Kuler087

3rd February 2025, 20:12

The list is for REMUX rip and I say:

Can only be played on a FEL device and can't be converted to P8 without baking FEL into BL

The list about the proper encoder groups is just a bonus. I really don't care about encodes, I only watch/keep REMUX.

Kuler087

4th February 2025, 16:12

3.0.6 Released: https://github.com/R3S3t9999/DoVi_Scripts/releases

** LATEST Tools pack + bat files + GUI (bat files only link below)**
[GoogleDrive](https://drive.google.com/file/d/1S4dqemaD8snI7QW29InG_XjI6VP0PNhk/view?usp=drive_link)
[MEGA](https://mega.nz/file/BO0TFK5Y#m3jW75vgavjO1kAbSNoQuWsZrIPLqmvntq_OFpUJWZM)
(case sensitive) PASSWORD FOR THE TOOLS PACK IS THE ZIP FILENAME= DoVi.Scripts

**CHANGES:**

- (7-4) NEW workflow that can quickly and automatically create 2min (10 frames) DV L1 vs L2 metadata comparison [videos like these](https://drive.google.com/drive/folders/1g5I-z_sJmVu-SAIPNiiSlcdMiy2ka0mf)
- (7-5) NEW workflow that can quickly and automatically create 2min (10 frames) FEL vs MEL comparison[ videos like these.](https://drive.google.com/drive/u/0/folders/1FS42T95TOSpoy4xtwUBIQmziCe_R_IKe)
- (8-2) NEW: Can now create DV/HDR10 samples of a single frame in loop (line 180, default = NO)
you must [add these new files](https://drive.google.com/drive/folders/18403rSHIaVFkHIFVkO1ZYoIZmZrRKwag?usp=drive_link) to your tools folder. (or re-download the tools pack)
- (6-2) Sped up the P5 input measurement 2.2X (if your cpu can handle it)
- (6-2) Sped up when you opt to crop and use prores
- (3-1) Added a workaround for mkv/ts/m2ts files that fail with madmeasureHDR.exe
- (6-3)(6-4) Added more stats to the L1 plot
- (1) RPU input + ''e'' selection can now loop the RPU frame(s) to check/info.
- (7-1) OSD will now display the frame(s) Level 1 rpu values in nits
- (7-2) OSD can now display the frame(s) Level 1 rpu values in nits if DV is present (line 123 , default = NO)
- (7-2) Can now accept up to 20 inputs
- (7-2) Manual mode will now ask for the frame to export before indexing.

ac777103

16th February 2025, 22:40

I've been looking at the custom skin for the Ugoos that provides in-play DV data, and noticed a few titles that seem to be static: e.g. Aliens and The Abyss (disc sourced MEL). I just ran a plot of Aliens and indeed it's flat. For such titles, would I get a better result via 3-1 and discard the original RPU?

Kuler087

17th February 2025, 00:02

Well, those movies are just 200nits grade. It won't make much difference unless you have a very low nits display...

ac777103

17th February 2025, 09:11

thanks, so for a release like Aliens, the provision of DV on the disc is just a marketing gimmick - the peak nits level per shot is really low, and the RPU is static.

to confirm presence of a static RPU, do I just look at all the plots from 6-4, and check that they are all flat? (at least for MEL discs). It's not sufficient that L1 on its own is flat? I guess it's harder to check for FEL titles?

Korbus

17th February 2025, 10:32

thanks, so for a release like Aliens, the provision of DV on the disc is just a marketing gimmick - the peak nits level per shot is really low, and the RPU is static.

The same is true for the Abyss and Titanic and, I suspect, for most of Cameron's latest batch of UHD releases (Avatars). Seems to be another of his oddities after DNR'ring Terminator 2 UHD to death ...

Kuler087

17th February 2025, 12:43

to confirm presence of a static RPU, do I just look at all the plots from 6-4, and check that they are all flat? (at least for MEL discs). It's not sufficient that L1 on its own is flat? I guess it's harder to check for FEL titles?
yes 6-4. Sometimes L1 is static but the trims are not.

I guess it's harder to check for FEL titles?
FEL can be checked with 7-1 or 7-5

ac777103

17th February 2025, 21:33

The same is true for the Abyss and Titanic and, I suspect, for most of Cameron's latest batch of UHD releases (Avatars). Seems to be another of his oddities after DNR'ring Terminator 2 UHD to death ...

Yes, I just ran through my Cameron mkv's. Aliens and The Abyss (MEL) are completely static. The Terminator (MEL) is static apart from some negligible dynamic data early on. Titanic and True Lies (FEL) have completely static RPUs, the FEL is between 5-6gb for each title. I don't have an nVidia GPU so the FEL workflows are a challenge for me to run. I'll look at new P7 titles more carefully from now on.