[MEncoder-users] differences in divx2pass.log between -oac copy and mp3lame

[Guillaume POIRIER]

Hi,

On 9/4/05, John Suzuki <jsuzuki at free.fr> wrote:
> Rich Felker wrote:
> 
> >On Sat, Sep 03, 2005 at 03:32:08PM +0200, John Suzuki wrote:
> >
> >
> >>Between:
> >>  $ mencoder -ovc lavc -lavcopts vpass=1 -oac copy ...
> >>and
> >>  $ mencoder -ovc lavc -lavcopts vpass=1 -oac mp3lame ...
> >>the generated divx2pass.log files are different.
> >>I am curious as to why, since the log file depends on the video part,
> >>not the audio part, of the input, right?
> >>
> >>
> >
> >This is due to a bug in mencoder's a/v sync, especially with lame. You
> >must use the same audio options for both passes. I would recommend
> >using -oac copy for both passes, then using -ovc copy -mc 0 -noskip
> >-oac mp3lame to reencode audio separately afterwards. Or demux the
> >audio and encode it separately with external lame, then remux. Or..
> >there are many other ways. The dumb, slow, but easy way is just to
> >encode audio twice with -oac mp3lame on both passes.
> >
> >
> Thanks everybody for your replies.
> I tested with -mc 0 added and it seems to remove the problem.
> But are there any disadvantages using -mc 0 in all passes?

I'm working on a doc patch about this. Part of it is there:
http://mplayerhq.hu/DOCS/HTML/en/menc-feat-dvd-mpeg4.html#menc-feat-dvd-mpeg4-av-sync.
Basically, -mc 0 may work just all right if your source is "clear and
reliable". However, it may harm more than do good in case your video
filters chain discards/copy some frames.

Experiment with a without, and see what works best for you. It's not
really possible to foresee how it'll affect your encode before
actually trying out.

Guillaume
-- 
Reading doesn't hurt, really!
  -- Dominik 'Rathann' Mierzejewski