[MPlayer-users] mencoder/lameopts aq=<value> ?

[Haas Wernfried]

as no one else seems to be interested in this for me quite interesting issue 
i just did some more research for myself. attention: i am not a programmer, so
don't flame me for anything wrong. you could have done this yourself and
better ;)

what i might have found out:
   {"aq", &lame_param_algqual, CONF_TYPE_INT, CONF_RANGE, 0, 9, NULL},
aq is passed to lame_param_algqual which is passed to lame_set_quality in
mencoder. according to lame sources/include/lame.h:
// internal algorithm selection.  True quality is determined by the bitrate
// but this variable will effect quality by selecting expensive or cheap
   algori\t// quality=0..9.  0=best (very slow).  9=worst.
// recommended:  2     near-best quality, not too slow
//               5     good quality, fast
//               7     ok quality, really fast
int CDECL lame_set_quality(lame_global_flags *, int);
   ----------------

   {"q", &lame_param_quality, CONF_TYPE_INT, CONF_RANGE, 0, 9, NULL},
q is passed to lame_param_quality which ends up with 
// VBR quality level.  0=highest  9=lowest
int CDECL lame_set_VBR_q(lame_global_flags *, int);
in lame.h

so aq actually is quality by algorithm (slow/fast) and
q is actually higher/lower vbr bitrate.

please check this for yourself if you think i am wrong and update the docs

regards
	wernfried




On Tue, Jul 09, 2002 at 12:49:45PM +0200, Haas Wernfried wrote:
> [Automatic answer: RTFM (read DOCS, FAQ), also read DOCS/bugreports.html]
> hello,
> i think i might found out what this value is being used for:
> if you lower the value (aq=0), encoding time rises, with high values (aq=9)
> encoding time decreases. i guess, that this is the switch for the "-q" in
> lame:
> (from lame manpage):
>        -q qual
>               0 <= qual <= 9
> 
>               Bitrate is of course the main influence on quality.
>               The  higher  the  bitrate,  the higher the quality.
>               But for a  given  bitrate,  we  have  a  choice  of
>               algorithms  to  determine the best scalefactors and
>               huffman encoding (noise shaping).
> 
> q=2 is recommended and always enabled when using vbr. 
> 
> q=value in lame seems to refer to the bitrate in vbr which is -V in lame
> 
> a few examples to illustrate this (only audio was encoded in 3-pass style):
> (1)
> cbr:br=128:aq=2
> Pos: 485.3s  12131f (100%)  65fps Trem:   0min   7mb  A-V:-0.045 [0:128]
> filesize: 8.2 megs
> ordinary cbr encoding...
> 
> (2)
> vbr=2:q=0:aq=2
> Pos: 485.2s  12131f (100%)  31fps Trem:   0min   9mb  A-V:-0.016 [0:157]
> 9.9 megs
> vbr with highest quality results in higher bitrate/bigger file...
> time: user    6m4.300s
> 
> (3)
> vbr=2:q=9:aq=2
> Pos: 485.2s  12131f (100%)  29fps Trem:   0min   8mb  A-V:0.017 [0:136]
> 8.7 megs
> no big surprise, filesize/bitrate are lower now, encoding was a bit faster.
> time: user    5m57.660s
> 
> (4)
> vbr=2:q=9:aq=9
> Pos: 485.2s  12131f (100%)  46fps Trem:   0min   8mb  A-V:-0.016 [0:137]
> 8.8 megs
> time: user    3m27.720s
> filesize/bitrate are almost the same, but encoding was much faster (and
> the quality should be worse than in (3)
> 
> 
> if aq(mencoder) is -q (lame) (which i am quite sure) and q(mencoder) is 
> -V (lame) the whole thing gets a bit confusing with these different q's,
> maybe it would be wise to name them as in lame. maybe not, as this just
> messes up existing encoding scripts and people wonder, why the q-switch
> does no longe affect bitrate. however i hope you can now take add the
> description for the aq-switch to the manpage.
> 
> regards
> 	wernfried
> 
> -- 
> -> Wernfried Haas - w.haas at xover.mud.at / amnenion at xover.mud.at 
> -> Homepage: http://xover.mud.at/~amnenion
> -> Beutelland: http://bl.mud.at
> 
> _______________________________________________
> RTFM!!!  http://www.MPlayerHQ.hu/DOCS
> Search:  http://www.MPlayerHQ.hu/cgi-bin/htsearch
> http://mplayerhq.hu/mailman/listinfo/mplayer-users

-- 
-> Wernfried Haas - w.haas at xover.mud.at / amnenion at xover.mud.at 
-> Homepage: http://xover.mud.at/~amnenion
-> Beutelland: http://bl.mud.at