Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
eac3to Exploiting Seamless Branching
#31
(2020-09-26, 06:46 PM)Chewtobacca Wrote: ^ That's essentially correct, although there's an audio "frame" of overlap on each side of the join. People often report difficulty creating branching discs, outside of using high-end solutions such as Scenarist.  It might be theoretically possible to encode the files in such a way as to ensure seamless playback via a playlist, but you'll have to experiment.  I've thought about doing this but have never had cause to try it, so please post your findings.  It will be fascinating to see if a poor man's seamless branching can be achieved, and if so, how programs such as eac3to process the results.

I don't use Scenarist but Blu Disc Studio is a similar pro level piece of software and it does offer seamless playlists so pretty sure it will work, just need to do some experimenting. I've used it before to offer end credits in a different language, but the jump wouldn't be as obvious then, better to try a mid scene jump. I'll let you know how it goes. 

The biggest problem I can foresee is the splitting of video, audio and subs into different files. The video for example, I usually make the disc, add the menus, extras, audio etc and see what space I have left, then use ripbot as it has a target file size option, so my 264 encode is exactly the right size for the disc. I suppose I could encode the movie as one file, then split it afterwards, but I don't think that will work as I'll be limited to cutting at key frames. 

Maybe I'll split the files at AVI stage then proportion the size. So for example if the overall video is 150,000 frames and my first segment is 15,000 frames, and my disc space is 18 gb, I'll encode the first segment with a target size of 1.8 gb and so on. The subs can be cut into segments in something like subtitle edit, and maybe the authoring program is clever enough to duplicate an audio frame either side of the jump. 

I think I'll try it with something pretty basic first that only needs a couple of segments!
Reply
Thanks given by:
#32
Yeah, encoding the segments separately would probably be best. AviSynth's Trim() command enables one to send different sections of the video to the encoder, avoiding the need to make multiple AVIs.
Reply
Thanks given by: alleycat
#33
(2018-12-01, 10:32 PM)Chewtobacca Wrote: The Aluminum Falcon raised this topic a while back on ot.com.  See his informative thread here.  While he uses tsMuxeR, eac3to is a better tool for the job because it runs a second pass when fixing the gaps that result from seamless branching.  EDIT:  See schorman's posts below.

I had fun trying this to combine the Blade Runner Director's Cut with the extra seconds of violence from the International Cut, an idea mentioned in the first reply in the OT.com thread. (This makes the DC a bit closer to the Final Cut, but of course without the FC's CGI enhancements and other additions and tweaks.)

This takes some prep work because the branched Blu-ray has two different sets of audio and subtitle tracks. One set is activated when you play the Theatrical/International Cut, and the other set only for the Director's Cut. Fortunately, you can use tsMuxeR to remux each individual .m2ts file to contain only the right audio/subs for each piece.

You'll end up with an eac3to command like this:

Code:
"filepath\eac3to334\eac3to.exe" "X:\00022.M2TS"+"X:\00023.M2TS"+"X:\00024.M2TS"+"X:\00033.M2TS"+"X:\00026.M2TS"+"X:\00034.M2TS"+"X:\00028.M2TS"+"X:\00035.M2TS"+"X:\00030.M2TS"+"X:\00031.M2TS" -demux -no2ndpass

It worked wonderfully!

I assume the warning messages in the log about "audio overlaps" are expected and can be ignored. Is that correct? I don't notice any obvious audio glitches at the joins:

Code:
eac3to v3.34
command line: eac3to  "X:\00022.M2TS"+"X:\00023.M2TS"+"X:\00024.M2TS"+"X:\00033.M2TS"+"X:\00026.M2TS"+"X:\00034.M2TS"+"X:\00028.M2TS"+"X:\00035.M2TS"+"X:\00030.M2TS"+"X:\00031.M2TS" -demux -no2ndpass
------------------------------------------------------------------------------
M2TS, 1 video track, 1 audio track, 2 subtitle tracks, 1:56:50, 24p /1.001
1: VC-1, 1080p24 /1.001 (16:9)
2: AC3 Surround, English, 2.0 channels, 192kbps, 48kHz, dialnorm: -27dB
3: Subtitle (PGS), English
4: Subtitle (PGS), Spanish
[s03] Extracting subtitle track number 3...
[v01] Extracting video track number 1...
[s04] Extracting subtitle track number 4...
[a02] Extracting audio track number 2...
[a02] Removing AC3 dialog normalization...
[v01] Creating file "00022 - 1 - VC-1, 1080p24.vc1"...
[a02] Creating file "00022 - 2 - AC3 Surround, English, 2.0 channels, 192kbps, 48kHz.ac3"...
[s04] Creating file "00022 - 4 - Subtitle (PGS), Spanish.sup"...
[s03] Creating file "00022 - 3 - Subtitle (PGS), English.sup"...
[a02] Audio overlaps for 15ms at playtime 0:42:10.  <WARNING>
[a02] Audio overlaps for 6ms at playtime 0:42:59.  <WARNING>
[a02] Audio overlaps for 30ms at playtime 1:26:06.  <WARNING>
[a02] Audio overlaps for 33ms at playtime 1:33:04.  <WARNING>
[a02] Audio overlaps for 14ms at playtime 1:40:14.  <WARNING>
[a02] Audio overlaps for 16ms at playtime 1:40:43.  <WARNING>
[a02] Audio overlaps for 16ms at playtime 1:52:34.  <WARNING>
The 2nd pass is skipped by request.  <WARNING>
Video track 1 contains 168083 frames.
Subtitle track 3 contains 881 captions.
Subtitle track 4 contains 745 captions.
eac3to processing took 10 minutes, 4 seconds.
Done.
Reply
Thanks given by:
#34
(2021-03-30, 07:41 AM)resolution Wrote: This takes some prep work because the branched Blu-ray has two different sets of audio and subtitle tracks. One set is activated when you play the Theatrical/International Cut, and the other set only for the Director's Cut. Fortunately, you can use tsMuxeR to remux each individual .m2ts file to contain only the right audio/subs for each piece.

That's exactly the solution upon which I hit when coming across this problem. 

(2021-03-30, 07:41 AM)resolution Wrote: I assume the warning messages in the log about "audio overlaps" are expected and can be ignored. Is that correct? I don't notice any obvious audio glitches at the joins:

That's correct.  If you're interested, you can read the discussion from schorman's first post in the thread onwards, but the upshot is that we skip the second pass because it causes eac3to to mishandle (rather than fix) the gaps.  Unless you observe a sync problem, everything should be fine
Reply
Thanks given by: resolution


Forum Jump:


Users browsing this thread: 1 Guest(s)