[MEncoder-users] [BUG] a-v sync problems with pullup, softskip (only on some DVDs)
D Richard Felker III
dalias at aerifal.cx
Wed Mar 16 04:46:48 CET 2005
On Tue, Mar 15, 2005 at 03:48:56PM -0800, Corey Hickey wrote:
> D Richard Felker III wrote:
> >On Tue, Mar 15, 2005 at 12:41:48PM -0800, Corey Hickey wrote:
> >
> >>I've been seeing an elusive a-v sync problem recently with a few DVDs
> >>and have finally tracked it down to a reproducible test case.
> >>
> >>Using pullup,softskip on the (recently released) Star Wars DVD results
> >>in audio that plays about 0.3 seconds early.
> >
> >
> >Interesting. Can you tell me if strange frame skips or duplicates
> >happen at the beginning (with or without pullup loaded). Unfortunately
> >mencoder doesn't like filters that buffer a few frames ahead, so
> >pullup has to do a nasty hack to hide that it's doing this.. If the
> >first few frames are supposed to be skipped it could mess things up
> >rather badly, perhaps...?
> >
>
> I don't know exactly what you mean by strange, but there certainly are a
I meant extra/fewer skips or dups at the very start of the encode.
> I also just noticed something I hadn't before: when using pullup
> mencoder reports "Could not get buffer from pullup!" four times. I just
This is very bad and should never happen.. It means something is going
radically wrong with the buffer management. Can you help me isolate
the problem so I can see it happen on a small clip? If not, I'll try
to come up with a patch that would help make a more useful debug log
for me to read when it happens, or I might even download the big file
if this problem proves to be very elusive..
> The messages appear after a long series of duplicate frames, between
> about 84 and 88 MB into the file. I tried encoding just that portion,
> and pullup doesn't give me an error - just lots of duplicate frames. If
> you're interested, you can download just that portion:
I looked at it and this clip is very strange. probably the motion is
so tiny and so aliased that pullup is getting confused about which
fields match better.. I don't know a good way to fix it, but I should
find something since pullup should do nothing when acting on input
that came from soft-telecine. I'm committing something that will help
improve this situation but it still won't be perfect.
Rich
More information about the MEncoder-users
mailing list