Demuxing a TS file

[carlmart]

Usually I deal with mkv files, which I understand are really "capsules" for other files inside them, like H264, AC3, DTS, subs, etc. Right?

This time I have a TS file that apparently contains a 1920 x 1080 video file and an AC3 file. How do I demux it?

DGIndex won't work and the other ways I tried, ProjectX and Xport, I don't know how to handle. Is there any other way to open these files?

[poisondeathray]

tsmuxer

[carlmart]

tsmuxer

Yes, of course! Thank you, Poison .

I even had that installed on my video programs directory, but I had forgotten about it.

Demuxing as we speak.

Great!

[carlmart]

An epilogue to this thread.

In spite of all the time I lost on this conversion, I couldn't use the files. There were lots of artifacts in the video when it was running. Apparently badly converted or compressed. It had happened when I played the original TS file, but I thought it was a player's problem. It was not.

It was useless to bring that problem here, as I don't think it could be solved. Perhaps I was wrong.

[poisondeathray]

Does the .ts file have corruption when you play it back in MPC?

Did you use DGAVCIndex? If so, there is a known issue with h.264 transport streams (I'm assuming it's h.264 and not MPEG2)and mbaff interlacing causing blocky corruption. (it's based on a very old libavcodec build)

The work around is DirectShowSource() or DSS2() + a recent ffdshow build , coreavc, or divx h.264 as the decoder. I think the new ffmpegsource2 (beta10) should work as well

Also it's very common for .ts to have corruption, drop outs etc.. during recording. If the original .ts has corruption (not through DGAVCindex) , You can try fixing with TS-Doctor

Why are you demuxing? Usually there is no need...

[carlmart]

Does the .ts file have corruption when you play it back in MPC?

No, I didn't play it in MPC. Only in WMP and VLC. Would it have made any difference?

Did you use DGAVCIndex? If so, there is a known issue with h.264 transport streams (I'm assuming it's h.264 and not MPEG2)and mbaff interlacing causing blocky corruption. (it's based on a very old libavcodec build)

The work around is DirectShowSource() or DSS2() + a recent ffdshow build , coreavc, or divx h.264 as the decoder. I think the new ffmpegsource2 (beta10) should work as well

Yes, I did use DGAVCIndex in order to have a dga file and convert it with avisynth. DSS I can implement through avisynth, but what about the others you suggest?

In any case, that would be for the next time I have a similar problem.

Also it's very common for .ts to have corruption, drop outs etc.. during recording. If the original .ts has corruption (not through DGAVCindex) , You can try fixing with TS-Doctor

Pity I didn't know about this TS doctor. I don't have the files anymore. I went a different way for that project, not using ts files. Asked for an mkv and got it.

Why are you demuxing? Usually there is no need...

I wanted to burn a DVD from the files, so I need conversion.

[poisondeathray]

I'm willing to bet it was the interlaced h.264/mbaff issue with DGAVCIndex. Just keep it in mind the next time with any interlaced h.264 content

DSS2() and FFMpegSource2() are other source filter alternatives to DSS. Both are frame accurate, so usually better choices than DSS(), which is not necessarily frame accurate (sometimes gives problems and sync issues). You can read the documentation and corresopnding thread on Doom9 if you want more info

Another alternative if you have a nvidia card is using DGNVTools ($15)

Cheers

[carlmart]

I'm willing to bet it was the interlaced h.264/mbaff issue with DGAVCIndex. Just keep it in mind the next time with any interlaced h.264 content

DSS2() and FFMpegSource2() are other source filter alternatives to DSS. Both are frame accurate, so usually better choices than DSS(), which is not necessarily frame accurate (sometimes gives problems and sync issues). You can read the documentation and corresopnding thread on Doom9 if you want more info

Another alternative if you have a nvidia card is using DGNVTools ($15)

I was having strange problems with DSS, and I never got DSS2 to work. That's why I went the DGAVCIndex way. I was also advised that that way was frame-accurate. But I do have sync problems sometimes using h264 files. Very few, fortunately, but I would like things to be more predictable.

[poisondeathray]

DGAVCIndex is great for all h.264 that is progressive. Very stable & predictable. It's just that it shouldn't be used for transport streams (due to the libav issue) which the majority are interlaced

DSS & DSS2 will be reliant on you system installed splitters & decoders which maybe why some people have issues

ffmpegsouce2 works independent of your system installed decoders and is probably more consistent

When possible, try not to demux, because the container can have timing information (e.g. transport streams, even mkv) that is lost when you demux. This is especially important for transport streams which frequently have drop outs and desynchronizations points which are "resynched" by information in the container

And yes, the choice of MPC vs. VLC is important. MPC (with internal filters & source filters disabled) is a good diagnostic utility, because it can help tell you if your directshow system is working. So if everything works fine in MPC, chances are DirectShowSource or DSS2 will work. VLC won't work for this test, because it relies on internal splitters/decoders. MPC is a directshow player, VLC is not

[carlmart]

DGAVCIndex is great for all h.264 that is progressive. Very stable & predictable. It's just that it shouldn't be used for transport streams (due to the libav issue) which the majority are interlaced

So if it's 23.976 it can be used? How can I find out if the file is interlaced or progressive? Mediainfo does not tell you that.

DSS & DSS2 will be reliant on you system installed splitters & decoders which maybe why some people have issues

ffmpegsouce2 works independent of your system installed decoders and is probably more consistent

Which script syntax should I use to load the file with ffmpegsouce2?

When possible, try not to demux, because the container can have timing information (e.g. transport streams, even mkv) that is lost when you demux. This is especially important for transport streams which frequently have drop outs and desynchronizations points which are "resynched" by information in the container

How can I convert if I do not demux?

And yes, the choice of MPC vs. VLC is important. MPC (with internal filters & source filters disabled) is a good diagnostic utility, because it can help tell you if your directshow system is working. So if everything works fine in MPC, chances are DirectShowSource or DSS2 will work. VLC won't work for this test, because it relies on internal splitters/decoders. MPC is a directshow player, VLC is not

OK. Got that. Apparently WMP is not an option. Not even 11.0 version? That's what I use, and it works better than VLC on some video files.

[poisondeathray]

So if it's 23.976 it can be used? How can I find out if the file is interlaced or progressive? Mediainfo does not tell you that.

You just look at it . Examine it in avsp or vdub

Which script syntax should I use to load the file with ffmpegsouce2?

It tells you how to use it in the documentation. There are examples

How can I convert if I do not demux?

What do you mean? It's not necessary to demux most of the time. If you are making a DVD, just feed your .avs script into HCenc. If there is no reason to demux, why waste the time/disk space and risk desynchronizing it? DGAVCIndex doesn't work with some containers, but the other source filters do.

OK. Got that. Apparently WMP is not an option. Not even 11.0 version? That's what I use, and it works better than VLC on some video files.

Yep WMP11 is a directshow player as well and can be used for testing

[carlmart]

It tells you how to use it in the documentation. There are examples

OK. I could load an mkv file using dss2. The subtitles do not seem to burn in, as when I use straight dss.

I couldn't find out, looking at the documentation, if or how I can use FFMpegSource to load the mkv file.

I wrote it like this: FFMpegSource ("d:\file.mkv)". Nothing happens. Is something missing?

[carlmart]

It's not necessary to demux most of the time. If you are making a DVD, just feed your .avs script into HCenc. If there is no reason to demux, why waste the time/disk space and risk desynchronizing it? DGAVCIndex doesn't work with some containers, but the other source filters do.

Correct me if I am wrong. Even if I do not demux the video, I will have to demux the audio anyway in order to author the DVD. I need that to add the subs and all that.

[poisondeathray]

Download ffms2 from here, check the thread for updates, there are frequent changes

http://ivtc.org/new/FFmpegSource-2.00b10.rar
http://forum.doom9.org/showthread.php?t=127037

Unzip the contents into the plugins folder, including the ffms2.avsi

You probably downloaded the wrong version, but EVERYTHING is included in the documentation. PLEASE read it.

The generic syntax is
FFMpegSource2("file.mkv")

If you want audio as well (this frameserves both audio & video)
FFMpegSource2("file.mkv", atrack=-1)

If your audio is already DVD compliant (e.g. AC3), then it would be better to demux it and use it as the original because no quality loss. If it something else e.g. AAC, and you are re-encoding it anyway, then use the "atrack=-1" or demux it then re-encode. But if you demux it, don't forget the delay

If you don't want subs burned in using DirectShowSource(), just repackage the mkv with mkvmerge beforehand, without the subs; or disable VSfilter or subs in ffdshow (whatever you are using)

Finally, if you're happy with DGAVCIndex, and your current method with progressive h.264 content, just stick with it. "Why fix it if it a'int broke"

[carlmart]

Download ffms2 from here, check the thread for updates, there are frequent changes

http://ivtc.org/new/FFmpegSource-2.00b10.rar
http://forum.doom9.org/showthread.php?t=127037

Unzip the contents into the plugins folder, including the ffms2.avsi

You probably downloaded the wrong version, but EVERYTHING is included in the documentation. PLEASE read it.

The generic syntax is
FFMpegSource2("file.mkv")

OK. Downloaded it and installed as you said. Things take a little longer to test out in AvsP, but it seems to work. Also loads fine in HCEnc.

But if you demux it, don't forget the delay

When will I have a delay and what should I do to cure it? I only get delay on some files, usually about 100ms or a bit more.

If you don't want subs burned in using DirectShowSource(), just repackage the mkv with mkvmerge beforehand, without the subs; or disable VSfilter or subs in ffdshow (whatever you are using)

I want the conversion to be as transparent as possible, and that's why I want to find out which is the best way. Quality image, movement, etc. might be different.

Repackaging the mkv seems like a better soluion. But they do not show if I use DSS2 or FFMpegSource2.

Finally, if you're happy with DGAVCIndex, and your current method with progressive h.264 content, just stick with it. "Why fix it if it a'int broke"

Indeed, but if I can lower the conversion times it's god too. As I said, I am trying to improve my conversions, quality wise, as much as possible.

[carlmart]

Just curious.

Why does it take so much to load a script where I am using FFMpegSource2?

That happens whether I am loading the script in AvsP or in HCEnc.

[poisondeathray]

How long are you talking about?

It should take a short time the first time around , proportional to the length of your video, because it generates an index (much like DGAVCIndex or DGMPGIndex does)

[carlmart]

How long are you talking about?

It should take a short time the first time around , proportional to the length of your video, because it generates an index (much like DGAVCIndex or DGMPGIndex does)

It takes about 4-5 minutes for a 3 hour long video. Now that you told me what happens, which I guessed might be something like it, it's explained.

For now FFmpegSource2 seems to be slightly better than the "h264 + DGAVCindex" path I was taking.

I wrote the script with FFmpegSource2 and with AvsP splitted the video accurately in two. Now how can I split the audio?

What I used with DGA was this:

v=AVCSource("d:\file.dga")
a=NicAC3Source("d:\fileA.ac3")
audiodub(v,a)
trim(0, 116558)
soundout

Can I use the same script just replacing the video source?