[MPlayer-users] insane compulsive maniacal mencoder settings

[Robert R. Wal]

On 02.11.22 D Richard Felker III pressed the following keys:

> [Automatic answer: RTFM (read DOCS, FAQ), also read DOCS/bugreports.html]
> On Fri, Nov 22, 2002 at 06:58:21PM +0100, Robert R. Wal wrote:
> > [Automatic answer: RTFM (read DOCS, FAQ), also read DOCS/bugreports.html]
> > On 02.11.22 D Richard Felker III pressed the following keys:
> > 
> > > high-motion scenes) on one cd at 640x304, with 66 kbit/sec abr ogg
> > > audio, using the options I described above. So, IMHO it's definitely
> > 
> > Did you really use ABR audio? Don't. Use -q presets in oggenc (like -q0
> > for ~64kbps) since it produces much better quality than ABR. Also use
> > --resample 44100 if the source is 48kHz, since libvorbisenc is much
> > better optimised for 44.1 than 48.
> 
> Yes, I used abr. I'm sorry but I like to be able to specify the
> desired final file size. Especially when I've already encoded the
> video and there's only so much space left for audio!

But the problem is, that the ABR gives much lower quality than -q modes
of the same nominal bitrate (ABR encoding engine is younger and less
tuned than the -q one). Last time I checked the difficult samples (high
bells, broken glass, percussion ``plates''(?), rain) encoded with
1.0 release ABR 64kbps sounded like -q -1 (~48kbps). There was similar
analogy between ABR 80kbps and -q 0 (~64kbps), but at higher bitrate
the difference disapeared.

> IMO, the ideal situation would be to encode the audio track at max
> quality with vorbis, then use vorbis's scalability features to crop to
> the desired average bitrate in a second pass, without having to
> re-encode.

There is one more possibility. Encode sound first and let the script do
the math about desired bitrate for the video. Most of the time
mencoder sticks to the bitrate you tell him to use with 0.2kbps acuracy.

Robert

-- 
Bastard Operator From 149.156.96.35