About "Skipping Identical Frames" on Eac3to?

[theliver9x]

Sometimes I get the message "Skipping identical DTS or AC3 frames (seamless branching).
That means it removed some frames from the original disc. So does it cause loss of original movie data and I am very worried that the software will cause the movie to lose some images making it not full content like the original movie that the producer made.
Can anyone explain this to me. Thanks.

[filler56789]

Audio frames are groups of digital-audio samples.

(Yes, the tech terminology sucks, the "pros" love to be unimaginative and prefer adding new meanings to old words -_-)

Therefore, no, eac3to doesn't remove images from the video streams.
BUT the removal of audio frames MAY cause loss of synchronization,
generally speaking,
but I admit that I have no experience with seamless branching :-/

[Columbo]

Originally Posted by filler56789

BUT the removal of audio frames MAY cause loss of synchronization, generally speaking :-/

No, the point of removing the audio frames is to maintain sync for seamless branching.

[filler56789]

Originally Posted by Columbo

No, the point of removing the audio frames is to maintain sync for seamless branching.

Thanks for erasing the part that says that I might be wrong.

[tebasuna51]

Originally Posted by Columbo

No, the point of removing the audio frames is to maintain sync for seamless branching.

This is how multiple M2TS files are generated: the cut is exact in video frames, but it cannot be exact in audio frames at the same time.

Therefore, audio frames are repeated at the end of one M2TS file and the beginning of the next, compensated for by timestamps, and these must be removed to generate complete audio while maintaining synchronization.

[Columbo]

Yes, exactly what I said.

[tebasuna51]

Originally Posted by Columbo

Yes, exactly what I said.

Of course, I just wanted to explain the reason.

[Columbo]

Got it, thank you.

[hubblec4]

Everything written so far is not entirely correct.

There are no duplicate audio frames on Blu-rays!

The problem is MkVToolnix when muxing; there's a serious bug.

The workaround that eac3to and DGDemux use is OK, but it removes original audio frames, which ALWAYS leads to a slight synchronization issue.

[Columbo]

Strongs claims need strong evidence. May I ask for your evidence?

[hubblec4]

Hi

First the bug
https://codeberg.org/mbunkus/mkvtoolnix/issues/3728

My chapterEditor has a fix for that MKVToolNix bug
https://forum.videohelp.com/threads/...e6#post2754681

Unfortunately, I can't quickly find the relevant section in the Blu-ray specifications I have here.

But to summarize: Each play item must begin with a video keyframe, and the audio frame must not have an offset.

Finally, each play item may start an audio frame, as long as it's within the last video frame.

Inevitably, the last audio frame will "run" a tiny bit longer than the last video frame.

But no new audio frames may start after the last video frame.

And the problem is this: MKVToolNix appends the first video frame of the second play item directly after the timestamp of the last audio frame of the first play item. This creates a video gap of a few milliseconds. This quickly adds up when there are many play items.

[Columbo]

Don't fully follow you, but are you claiming no gaps correction is needed, or just that the issue is not audio frame duplication? domy has pretty clearly shown frame duplication without MKV being involved.

DG's gaps correction doesn't rely on duplication. It just accumulates elapsed time for audio and video and compares them. Kind of like Bresenham's algorithm.

Sounds like your MKV issue is a different thing than M2TS gaps correction not involving MKV.

[hubblec4]

Eac3to and DGDemux work like this:
Audio and video times are continuously added together, and when the audio becomes approximately one frame longer than the video, an audio frame is simply discarded.

Yes, my chapterEditor (cE) doesn't perform any gap correction because it's not necessary.

All cE does is create a timestamp file for audio frames, which then tells MKVToolNix how to handle them. Each new play item is then appended directly to the previous one, seamlessly, for both video and audio.

[Columbo]

Yes, I know how eac3to and DG do correction. I am a DG Tools team member.

So this is exclusively an MKV issue? That's all I'm asking because your post claimed that there is never any duplicated audio frames but that is demonstraby false. It arises when demuxing ES. Read the spec about connection conditions.

I think preparing things for MKV muxing is different from demuxing ES from multiple seamless branched M2TS's.

[hubblec4]

It's definitely a problem for MKV, yes. But I'm not familiar with demuxing ES from m2ts files.

However, I've always searched the MKV files for "duplicate" audio frames, and by duplicate, I mean bit-identical.

In any case, MKVToolNix shows an Adler checksum for all frames, and this should be identical if there are duplicate audio frames. But I haven't found any identical Adler checksums at the transitions between one m2ts file and the next.

[Columbo]

Gotcha, thanks for the clarification.

[tebasuna51]

Originally Posted by hubblec4

It's definitely a problem for MKV, yes. But I'm not familiar with demuxing ES from m2ts files.

However, I've always searched the MKV files for "duplicate" audio frames, and by duplicate, I mean bit-identical.

In any case, MKVToolNix shows an Adler checksum for all frames, and this should be identical if there are duplicate audio frames. But I haven't found any identical Adler checksums at the transitions between one m2ts file and the next.

That may be, but searching in mkv's cannot prove that m2ts files (Blu-Ray) do not have duplicate audio frames:

Originally Posted by hubblec4

There are no duplicate audio frames on Blu-rays!

I extracted audio tracks from two adjacent m2ts (seamless branching) and they are bit-identical frames (last and first).

[hubblec4]

Originally Posted by tebasuna51

That may be, but searching in mkv's cannot prove that m2ts files (Blu-Ray) do not have duplicate audio frames:

Let's put it this way: for me, what's in the final MKV file is sufficient, since I completely trust MKVToolNix(MTX).

Mosu never programmed in gap correction or the removal of audio frames.

Every audio frame is used, and if no bit-identical frames are found, then there were no identical frames for this Blu-ray.

Originally Posted by tebasuna51

I extracted audio tracks from two adjacent m2ts (seamless branching) and they are bit-identical frames (last and first).

Could you please provide the name of the Blu-ray.
I'm very sure it is only one of 10000 BDs which have identical frames.

[Columbo]

MONSTERS_UNIVERSITY

See the FAQ here for some pretty pictures.

https://github.com/domyd/mlp

[tebasuna51]

The mlp demux is a exact method to extract even thd tracks (with frames 1/1200 ms), but with AC3 frames (32 ms) was more necesary like I show in 2012. All seamless branching BD's can be a example.