Dolby Vision now possible through MP4 Mux.

[kazuma]

reset_9999

the muxing time of shooter and the godfather untouched bluray is very short, and the end TS file is only between 13GB and 17GB. what is the reason for this? (4-1-1)

i've selected the longest mpls file. i'm sure of that.

[RESET_9999]

i've selected the longest mpls file. i'm sure of that.

I doubt that because I just tried Shooter BD and there are two 20min small playlists and the main movie one.
I took the right one and got a normal 44gb TS.

So either you're out of HDD space on your output path or you selected the wrong playlist.

[Gatorman]

Does this comparison look typical? HDR10 Resolve generated DV.

https://slow.pics/c/fVBUu802

[RESET_9999]

yes, as you can see madvr min_pq is always 0 which makes avg_PQ inaccurate.
Resolve's scene cut detection is also substantially better than madvr. With madVR, there will be (sometimes) one shot very bright and the next one(within the same scene) darker. I never had this issue with Resolve.

[kazuma]

reset_9999

i did it again with shooter. now with screenshots:

https://ibb.co/q1ZL44M

https://ibb.co/6HKkGnk

https://ibb.co/gv1pB4k

[Manixx2020beyound]

yes, as you can see madvr min_pq is always 0 which makes avg_PQ inaccurate.
Resolve's scene cut detection is also substantially better than madvr. With madVR, there will be (sometimes) one shot very bright and the next one(within the same scene) darker. I never had this issue with Resolve.

I’ve come to the conclusion that if we re analyze a movie that’s already graded scene by scene technically we are producing a diff graded result as well.(bc color decisions are being decided by present graded elements)
& that’s why they are diff types of analysis modes some of us have only experienced 1type of analysis mode.

[RESET_9999]

100% non-sense.
L1 metadata generation is just pixels brightness measurement(this is what the dovi L1 plot is). When you set your resolve project correctly you don't touch the original grade. it's like working with the master.
You could export it to 422 and then re-encode it to x265 and it would still be the same grade(unless you start playing with the colors and brightness).

[saba30]

Little bug in DoVi_Scripts_1.6.4
7-2-4
Audio file not included, because audio file name set incorrect
14587- Line

:skip.dts
"%eac3to_path%" "%TEMP%THD.mkv" 1: "%TEMP%audio.ac3"
if NOT ["%errorlevel%"]==["0"] pause

:no.dts.encodingg

set audio=%TEMP%audio.ac3

::--------------------------------------------------------------------------------------------MP4.muxing----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
echo muxing into a MP4 profile 8 DoVi file. Be patient, it can take a while. 40min^+ on slow HDD!
"%mp4muxer_path%" --dv-profile %dv.profile% --input-file %BL% --input-file %audio% %comp% %ID% --output-file "%TEMP%DoVi.au.mp4"
::--------------------------------------------------------------------------------------------SUBS----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
if exist "%TEMP%en.forced.srt" if not exist "%TEMP%en.srt" set subtitles_path="%TEMP%en.forced.srt"& goto :er
if not exist "%TEMP%en.srt" if not exist "%TEMP%en.forced.srt" goto :skip.sub
if exist "%TEMP%en.forced.srt" if exist "%TEMP%en.srt" goto :2.sub

[RESET_9999]

Little bug in DoVi_Scripts_1.6.4
7-2-4

yes I see my mistake...
In next version, I wont use eac3to.exe to demux anymore so that should fix it.

[Manixx2020beyound]

100% non-sense.
L1 metadata generation is just pixels brightness measurement(this is what the dovi L1 plot is). When you set your resolve project correctly you don't touch the original grade. it's like working with the master.
You could export it to 422 and then re-encode it to x265 and it would still be the same grade(unless you start playing with the colors and brightness).

The meta data results are close but never identical facts.
To get identical, a 12bit true master is necessary,
Also that’s like saying the 8.1 dv/hdr resolve produces is NOT an official 8.1?
Our experience with 8.1 only came about with xml/rpu (quietvoid)

[RESET_9999]

.

[RESET_9999]

You can't expect madvr pixels measurements to be 100% identical to resolve or the original. The different algorithms will produce slightly different results which is acceptable. Just like generating static maxcll/fall in Resolve will produce a slightly different result than in Transkoder, all normal stuff. Even Resolve CmV2.9 vs Cmv4.0 produce different L1 as shown here: https://slow.pics/c/ipOD7eqm
Then you have the scene cut that can be different as well, so the plot graph will be different in that regard too.
Anyway, I compared brightness of tons of movie with madvr vs original. Some L1 peaks are almost identical to original, see: https://docs.google.com/spreadsheets/d/ ... edit#gid=0

Rougly the HDR/DV workflow is:

1- Grade the 12 bit HDR master ( not in DV not in HDR10 not in HDR10+) shot by shot , scene by scene. This is where you do the color, brightness, highlights contrast, shadow, black, specular etc....
This will be encoded to 10 bit and will result in the Base layer.

2- Do the Dolby Vision Analysis, this an automated process done by an algorithm. This will generate L1 metadata that you cannot edit unless you change the master. This is the actual HDR grade brightness mapping.
They can "blend" shot when they feel its needed though. This just mean, you copy the metadata from one shot to another similiar shot.
Also, in resolve if you set your project to 1000nits but the content itself goes brighter than that, L1 will be max out at 1000nits.
I believe this is incorrect but some movies are done that way. Is it on purpose or an error , i dont know.

3- Do the L2 trim passes. This is the manual part of the Dolby vision creation.
The point of this process is not to change the final look. the point of th trims is to preserve the original look on different TV with different brightness capabilities instead of relying on TV internal tone mapping.

4- Then they decide whether to keep 12-bit data in FEL or not. The data(diff between 12bit and 10bits encode) is encoded in the 2nd layer (the 1080p stream).

Anyway, thanks to @Woodstock's and @manix's now-deleted comments, this is the last post from me in this forum!!! I'm out!

[Gatorman]

Seriously, why be toxic here? It does nothing to move the discussion forward.

[Woodstock]

For what it is worth, my comment was a reply to one that was deleted, NOT RESET_9999's messages. It was a mistake because I mistook who wrote a message.

[RESET_9999]

For what it is worth, my comment was a reply to one that was deleted, NOT RESET_9999's messages. It was a mistake because I mistook who wrote a message.

yes, your post really confused me
But it's all good, needed a break anyway. Anyone can still PM me if you have suggestions, bug or questions about my script.
thank you to all for the help and great discussions over the past years.