BD3D2MK3D v1.17: Convert 3D BDs or MKV to 3D SBS, T&B or Frame-sequential MKV

[tebasuna51]

No problem, it was only to show the full info.

[tebasuna51]

Delay MakeMKV test.

I created with tsMuxeR a Bluray folder adding manualy a delay of 200 ms to a DTS-HD track.
The m2ts output have the delay included:

M2TS, 1 video track, 2 audio tracks, 3 subtitle tracks, 0:02:01, 24p /1.001
1: h264/AVC, 1080p24 /1.001 (16:9)
2: DTS Master Audio, Spanish, 5.1 channels, 16 bits, 48kHz, 200ms
(core: DTS, 5.1 channels, 1509kbps, 48kHz)
3: DTS Master Audio, English, 5.1 channels, 16 bits, 48kHz
(core: DTS, 5.1 channels, 1509kbps, 48kHz)
4: Subtitle (PGS), Spanish
...

I used MakeMKV with defaults and the mkv generated preseve the delay:

MKV, 1 video track, 2 audio tracks, 3 subtitle tracks, 0:02:01, 24p /1.001
1: h264/MVC (right eye), English, 1080p24 /1.001 (16:9)
2: DTS Master Audio, Spanish, 5.1 channels, 16 bits, 48kHz, 200ms
(core: DTS, 5.1 channels, 1509kbps, 48kHz)
"Surround 5.1"
3: DTS Master Audio, English, 5.1 channels, 16 bits, 48kHz
(core: DTS, 5.1 channels, 1509kbps, 48kHz)
"Surround 5.1"
4: Subtitle (PGS), Spanish
...

Then can exist "custom" BD's with positive delay.

When I try to add a negative delay (-200) in tsMuxeR, seems than work like mkvmerge, cut the first DTS frames (verified decoding the track) and add a positive remaining delay:

M2TS, 1 video track, 2 audio tracks, 3 subtitle tracks, 0:02:01, 24p /1.001
1: h264/AVC, 1080p24 /1.001 (16:9)
2: DTS Master Audio, Spanish, 5.1 channels, 16 bits, 48kHz, 2ms
(core: DTS, 5.1 channels, 1509kbps, 48kHz)
...

I don't know if can exist "custom" BD's with negative delay, at least not using tsMuxeR.

[sneaker_ger]

Why do you need to demux the audio tracks in the first place?

[r0lZ]

To include them in the final MKV. ;-)

It is theoretically possible to use the original MKV to grab the streams, but that's not at all the current method used by BD3D2MK3D. It has to demux the audio streams from the BD (in BD mode) because it is too difficult to get them from multiple M2TS files (with seamless branching MPLS), and also to be able to convert them to AC3 if the user wants so. It is way too difficult for me to handle the BD and the MKV input modes totally differently. There are already enough differences to take into account. I can't add new difficulties, just to speed things up.

@tebasuna51:
Thanks for the test!

Can you show me the output of MkvMerge -F verbose-text -i INPUT.mkv ? Does it show the delay ?
And same question for the output shown by tsMuxeR INPUT.mkv

[EDIT] Never mind. I'll do the test myself.

[r0lZ]

Originally Posted by tebasuna51

AFAIK the only method is extract the timecodes of each audio track:
mkvextract timecodes_v2 INPUT.mkv 1:timecodes_1.txt
and read the second line of the txt:

# timecode format v2
200
210.666667
...

Here the delay of track 1 is 200 ms.

Indeed, the delay is not easy to retrieve. And I would like to avoid mkvextract, because it requires to save a big temp file on disc for each audio stream, and it takes ages to process the whole MKV file. I need only the first occurrence of the audio timecode, but I have no way to stop it to process the whole file. Also, currently, mkvextract is not included in the toolset, and I don't want to add it just for the delay problem.

Luckily, there is another solution. I can use mkvinfo --track-info (or -v) to output a lot of lines to stdout. And when I find a line with the following structure (for all audio tracks), I have found the delay, and I can kill mkvinfo:

| + SimpleBlock (key, track number 2, 8 frame(s), timecode 1.024s = 00:00:01.024)

That will certainly be much more rapid. And mkvinfo is already distributed with the toolset of BD3D2MK3D. :-)

Another solution would be to use the command-line version of MediaInfo, but it is not included in the toolset. However, that solution is so simple that I may be tempted to use it. ;-)

[r0lZ]

Here is a first beta with support for the MKV 3D input.

The source MKV file must have been created by MakeMKV (with the "Mpeg4-MVC-3D" video stream selected). Currently, the TrueHD audio tracks are ignored. I may add later a second demux pass to demux the THD track(s) with eac3to, but it's not yet implemented. Everything else should work as expected (including the audio delays), but remember that it's a beta! (Or should I say alpha?) It may have (and probably has) some bugs.

The program tries to find the info about the stereoscopy and 3D-Planes in the source 3D MKV, and should be ready when a new version of MakeMKV that include these fields/tags will be released, cut currently you will have to fill these values manually (in the first and second tab).

Please note that MkvMerge, MkvInfo and MkvPropEdit have been updated to the latest MkvToolnix version (v9.0.1), but that new version is NOT COMPATIBLE with the previous versions of BD3D2MK3D, because MkvInfo has a new --no-gui command line option that must be specified to use MkvInfo to grab the info from a MKV file. (Other tools, such as gMkvExtractGUI have the same problem.) So, after having tested the new beta, if you want to come back to v0.86, you will have to change the exes in the toolset folder, or restore the whole archive. It is not sufficient to simply overwrite BD3D2MK3D.exe ! (You can still download the latest stable version here: BD3D2MK3D.7z (stable))

This version fixes also the MKV compatibility option --engage no_simpleblocks. It was misspelled no_simpleblock, without the "s". (Thanks odyssey! And sorry. I forgot to fix this issue in v0.86.)

v0.87b1 (March 30, 2016)
- New possibility to open an interleaved 3D MKV file created by MakeMKV as input.
- Fix: The MKV compatibility option --engage no_simpleblocks was misspelled no_simpleblock
- Updated the MkvToolnix exes to the latest version (v9.0.1)

Download the beta: BD3D2MK3D_v0.87b1.7z [Link removed. Download the latest beta below.]

Please test the beta as far as possible. I have tested it with a few shorts, but I haven't much time to generate big MKV files with MakeMKV and then demux them, so my tests have been very limited. Please comment and report bugs and discrepancies here. Thanks!

[tebasuna51]

Thanks r0lZ.

I have now 2 BD's in my DD, one of them a custom JMBD with a extra audio track with a negative delay.
I'll make test with them.

[r0lZ]

Originally Posted by tebasuna51

... with a negative delay.

How is it possible? My understanding of the delay is that the audio can start after the video, but not the opposite.

Anyway, the fact that the delay is negative should not be a problem for BD3D2MK3D, but I think that MkvMerge will cut the audio file when it will mux it.

[tebasuna51]

Originally Posted by r0lZ

How is it possible?

I asked the authoring tool used to the author.

Anyway, the fact that the delay is negative should not be a problem for BD3D2MK3D, but I think that MkvMerge will cut the audio file when it will mux it.

Yep, is not a problem, I'm still checking some rounding differences but work fine.

--------------

A bug?

I'm pretty sure than I marked 'Left view in AVC stream', but in .avs:

# Current base view: right eye!
# The views are inverted: AVC stream = right view, MVC stream = left view.
right = SelectEven(interleaved)
left = SelectOdd(interleaved)

I attach the log's using the iso and the mkv from MakeMkv:

[r0lZ]

Oops, yes, there is a bug. The data are not correctly updated when the user ticks the option in tab 1, and the value -1 (meaning "undefined so far") is used instead of 0. I'll fix the bug asap...

[r0lZ]

OK, views order bug fixed.

I have also fixed another little bug. When a file from a BD was dropped over the main window (with tab 1 active), the BD was loaded, but without going to the BD input mode.
It is now also possible to drop a 3D MKV file over tab 1 to load it.

Download: [link removed]

[r0lZ]

Another bug fixed. tsMuxeR doesn't support meta files encoded in UTF-8, and crashed when there is a foreign or accented character in the filename of the input 3D MKV. Now, the meta file is saved in the encoding of the system. I hope that works in all cases, but at least that works with Latin European languages. If that doesn't work for you, you will have to rename the input 3D MKV file to remove all foreign characters and retry.

Download: [link removed]

[tebasuna51]

Thanks, downloaded b2.

How do you extract chapters from mkv?

With eac3to from mpls: 25 chapters
With mkvextract from mkv: 25 chapters
With BD3D2MK3D from iso: 26 chapters (one more at end like expected)
With BD3D2MK3D_beta from mkv: 21 chapters
Seems than lose 1 chapter each 5 (the 6, 11, 16, 21 and 26 from iso disappears)

EDIT: I don't see beta 3 but is not a problem for my test.

[tebasuna51]

With beta 2 avs problem solved, log is the same than beta 1.
The chapters problem remain.

[r0lZ]

I retrieve the chapters with tsMuxeR. And you're right. Some chapters were missing. That bug is fixed in beta 4.

Note that the additional chapter at the end of the movie can be removed just like with MakeMKV (if the last chapter is close enough to the end of the movie). See Settings -> Chapters.

Download: BD3D2MK3D_v0.87b4.7z

[tebasuna51]

Thanks for the quick changes.

One last question about subtitles.

The .sup's extracted from mkv have little differences in PTS than the extracted from the iso, and always the DTS value to 0.
Is not a problem of tsMuxeR extraction because extracted with mkvextract have the same values.

Also there are differences between calculations converting sup to 3D-sup (like can be see in log's).
I not know if is a problem or not, because is out of my knowledge.
If is ok for you also for me.

I'll make now some test with my second BD.

[r0lZ]

Sorry to reply only now. This morning, I was blind, without internet!

I haven't noticed the PTS differences yet, but anyway, I use the same program (tsMuxeR) to demux the subtitles from the M2TS of the BD and from the 3D MKV, so I suppose it's not tsMuxeR's fault, and BD3D2MK3D is not responsible. I suppose the little differences are introduced by MakeMKV, perhaps because at some point, the PTS are converted to frame number, and then back to PTS? Anyway, when BDSup2Sub converts the subtitles, similar rounding problems happen. As long as the difference is less than one video frame, I don't care.

I know also that BDSup2Sub is totally wrong when it converts the SUP to XML/PNG. IIRC, the XML contains always the timings for 24 fps, regardless of the options you use during the conversion and the frame rate written in the XML. When the XML/PNG stream is converted back to SUP (or DVD SUB), it is necessary to specify that the frame rate must be converted from 23.976 to 23.976, or the timings are completely wrong. That conversions can introduce rounding problems, and finally a difference of one video frame, but IMO, for the subtitles, that doesn't matter much.

I will verify this when I'll have some time to do new tests, with long movies. But that may take some times. I'm currently working on the second demux pass for the TrueHD tracks...

[r0lZ]

Here is the latest beta. It can demux the TrueHD audio with eax3to (during a second demux pass). Of course, it's slow, but it's better than nothing. Again, there are probably some bugs...

Also, the current input mode (BD or MKV 3D) is now saved with the preferences, so that it will be used by default the next time you open BD3D2MK3D.

Download: BD3D2MK3D_v0.87b5.7z

(I have not verified the PTS of the subtitles yet. I have to go now, but I should have some time tomorrow...)

[tebasuna51]

Ok, I know than little PTS differences in subs aren't important. We can forget that.

My test with my second BD work fine, all is the same using iso or mkv, with the know differences in sup's and chapters.

I'm check the chapters differences, and qpfile generated, and I see:

               Time         Time       Time   I-fra.  I-fra    I
Chapters   Bad from mkv  Ok from mpls  Dif.   mkv QP  mpl QP  Dif
---------  ------------  ------------  ----   ------  ------  ---
CHAPTER01  00:00:00.000  00:00:00.000     0
CHAPTER02  00:06:08.618  00:06:08.785  -167     8838    8842  - 4
CHAPTER03  00:12:20.406  00:12:20.490  - 84    17752   17754  - 2
CHAPTER04  00:17:30.591  00:17:30.591     0    25189   25189    0
CHAPTER05  00:23:51.054  00:23:51.180  -126    34311   34314  - 3
CHAPTER06  00:31:26.509  00:31:26.676  -167    45231   45235  - 4
CHAPTER07  00:35:54.485  00:35:54.486  -  1    51656   51656    0
CHAPTER08  00:42:29.421  00:42:29.547  -126    61125   61128  - 3
CHAPTER09  00:47:03.362  00:47:03.612  -250    67693   67699  - 6
CHAPTER10  00:53:16.318  00:53:16.318     0    76635   76635    0
CHAPTER11  00:59:12.632  00:59:12.716  - 84    85178   85180  - 2
CHAPTER12  01:04:38.040  01:04:38.041  -  1    92980   92980    0
CHAPTER13  01:08:32.108  01:08:32.108     0    98592   98592    0
CHAPTER14  01:14:22.416  01:14:22.416     0   106991  106991    0
CHAPTER15  01:21:03.483  01:21:03.609  -126   116607  116610  - 3
CHAPTER16  01:25:42.011  01:25:42.012  -  1   123285  123285    0
CHAPTER17  01:36:08.679  01:36:08.680  -  1   138310  138310    0
CHAPTER18  01:41:29.333  01:41:29.625  -292   145998  146005  - 7

Like I can't understand for what the differences I checked the I frames in video (must be the sames in mkv and mpls) and I discover than the I frames calculated with the mpls aren't always I frames in the video, but all calculated with the mkv chapters are I frames in video.

Then seems than MakeMKV correct chapters times to point real I frames in video, and the qpfile is better because point to real scene changes.

[r0lZ]

OK, thanks for the tests.

If I understand correctly, the chapters in the MKV are different than in the original BD, but they are better because they are perfectly aligned with the I-frames. Right?

For me, it's not necessarily a good point, because BD3D2MK3D creates an I-frame at the chapter point anyway (with the qpfile), but indeed, it MakeMKV sets it to a real scene change (which coincides with an I-frame almost always), it's good. It's less good if it chooses the I-frame immediately before or after the scene change.

Anyway, I have no way to modify the chapters myself. I can only trust the chapters in the MKV. I will just verify if the chapter timings are identical with all programs (tsMuxeR, eac3to, MkvMerge, MkvInfo and MediaInfo).

I may release the official v0.87 in a couple of days, if no new bugs are reported...