[MEncoder-users] A/V sync problems when digitizing old VHS tapes with mencoder

[Josef Wolf]

On Fri, Apr 30, 2010 at 03:12:18AM +0200, Miroslav Rovis wrote:

> > > > But did you get the sync OK?
> > I did only a short check, but at first glance, it looks much, much
> > better!
> Needs to be just right... Not just better.

Yeah, it seems to be "just right" AFAICS! I have not written "just right"
in my previous post because I had checked only a couple of sequences on
just one tape. So I have no idea what is on other tapes or even on other
recordings on the same tape. But the sequences I have seen up to now were
"just right".

> > But I still don't really get it. In your other post you write:
> I'll try make my point again.
> > > >> while it is halted. Still bad sync.
> > > > You seem to be issuing the reindexing command while the mencoder
> > > > is halted. No. That is completely useless. mencoder needs to be
> > > > running in it's own other konsole. You need to start the
> > > > "mencoder -forceidx ..." command *before* you halt it.
> Yes! Before.
> > I don't understand this. I send SIGSTOP, do the fixup, and finally kill
> > the capturing process. Should the capturing process be running all the
> > time and eating CPU and I/O instead?
> It needs to be at least *still running*  when the copy video/audio
> reindexing command is started, yes.

For me, it seems that SIGSTOP is just enough. But I think it should be
possible to insert a "sleep(5);" in my script to achieve your suggestion.

> > I haven't seen any noticeable difference between sending SIGSTOP and
> > leaving it running while fixing up. Have you seen a difference?
> Yes. Noticeable difference.

Well, I haven't noticed. Both variants look perfectly synchronized to me.
Might be dependant on the hardware or the quality of the recording?

AFAICS, the important point is to keep the capturing process from rewriting
the index. For this, SIGSTOP should be enough, IMHO.

Of course, it would be interesting to find the real cause of the problem
and fix the real bug instead of introducing such workarounds. Since I
already jumped through this compile hoop, I'm planning to dive a little
deeper into this issue and try to find out what is the real cause of the
problem. But for now, digitizing my family recordings has the higher
priority ;-)

> > BTW: in reality I use a little perl wrapper.
> Interesting, and next time I digitize tapes I might be back to try and
> use it.

Oh, I'm still fiddling with it. What I posted was far from being usable.
If there's any interest, I'll post the newest version once I get it into
a proper state. I wrote in my previous mail which improvements I am
thinking of (e.g. lossless capture, cut)

> Also, I am certain debian/ubuntu is fine breed of Linux, and it'd
> certainly be interesting to use, but I am all over gentoo to use it
> properly in my 50s and my mind is not so fresh anymore :-)

Ough! I tried several times to get into gentoo, but found it too
complicated for me (at least, I had not enough time to learn all the new
concepts). But I agree that gentoo has a very interesting concept. And I'd
like to learn it, but I just can't get around my time-shortcomings.

Regarding debian, I agree with you. But with ubuntu, you can have a full
desktop install up and running within 15 minutes. IMHO, getting into ubuntu
is not as hard as getting into gentoo. But I am biased, so better strip
those last two paragraphs.

[ ... snip ... ]

> I hope you'll be digitizing fine your tapes soon!

Well, sync was only the first part of the story. Next part is to find the
right balance between quality and file size (as I wrote in my previous mail)
But this is a different topic and probably deserves its own thread.

Miro, _Very_, _Very_ BIG thanks to you! I was almost about to give up with
all this sync problems. Without you, I would probably have given up and
thrown those tapes into the oven...

Hey, are there any developers out there? Miro has shown several times how
to work around this bug. There are lots of reports which acknowledge that
this work-around actually works. So the next step would be to inject some
debugging into this last fixup passage. Are there any developers out there
to give some advice? I'd be happy to help trying to track that problem down.
Any suggestions where to start debugging?