[MPlayer-users] Support of ogmtools inside mencoder?[was: Ther is a way to encode directly in ogm format]

Michael Waters stonin at rogers.com
Tue May 6 19:47:53 CEST 2003


Hi,

On Tue, May 06, 2003 at 13:23 -0400, D Richard Felker III wrote:
> > Because the method of ogmtools readme take too long. I mean that
> > when I encode divx with mencoder first I cacule the lenght of the cd
> > with frameno then I calcule the scale parameters, and then I can
> > leave the pc doing the 2 video passes, and I can go to sleep!
> 
> WTF?? Why are you even using frameno? You're just going to throw away
> the audio encoded at this pass (mp3) and replace it with ogg later,
> which will probably average a different bitrate, so the size info it
> tells you won't even be useful!!

In the ogmtools README, it does advise to use frameno:

   a) Extract the audio to PCM audio and let mencoder calculate the
   video frame numbers:

   mencoder -dvd 1 -ovc frameno -oac pcm -o frameno.avi

   b) Extract the audio again, this time to a plain WAV file:

   mplayer -dvd 1 -vc dummy -vo null -hardframedrop -ao pcm -aofile
   audio.wav


I did try encoding one movie to ogg media without the first `frameno`
dump and IIRC, the a/v sync was not as good as with. It was pretty much
synced but was noticeably sloppy in spots.  So I think it is useful.

> To other developers: IMO we should remove the frameno/3pass nonsense
> at some point. It's buggy and lots of users misunderstand it and use
> it when there's absolutely no point. Lame's bitrates are predictable
> enough that you don't need to actually encode audio to decide what
> video bitrate to use. But if this bitrate prediction nonsense is
> really important to some people, we should incorporate it into the
> 2pass encoding, so that it's done along with video pass #1. That would
> fix the bugs (A/V sync issues) caused by the current 3pass mode, since
> video is actually decoded.

I just encoded an ntsc dvd with the 3pass method and vbr mp3 and the
a/v sync is perfect.  But I've seen the bug reports about this.  A
while back, I did a different 3pass encoding from ntsc dvd and the a/v
snyc slowly goes bad from about halfway thru until the end.  IIRC, I
got about 50 `duplicate frame` or `skipped frame` messages (i can't now
remember which).  I think possibly some ntsc dvds are mastered in a
weird way with screwy framerates and such. Can anyone shed more light
on this?  Anyways, I'd miss the frameno pass if it were removed; I like
using lame vbr instead of abr or cbr and it's useful for making ogm
files ...but for sure it would be great if a/v sync was always accurate
with it.

Michael





More information about the MPlayer-users mailing list