[FFmpeg-devel] [PATCH] libmp3lame: set noise shaping & psycho acoustic algorithms quality

[Michael Niedermayer]

On Thu, Jul 31, 2008 at 06:07:19PM +0200, Nicolas George wrote:
> Hi. Thanks for your answer.
> 
> Le quartidi 14 thermidor, an CCXVI, Robert Swain a ?crit?:
> > The AAC profiles (as all MPEG profiles) affect the bitstream features
> > used.
> 
> Yes, I had thought of that. I do not think that is actually a problem. That
> would have probably been a problem if it were the other way around, bit in
> this case I think it is ok.
> 
> >	It seems to me (at first glance) -q is more analogous to -*cmp
> > or -dia_size for video or subq/subme in x264. I'm not sure if we have
> > an appropriate option for audio speed/compression efficiency tradeoff.
> 
> I browsed all existing options, keeping only the ones tagged "E.VA." and
> "E..A.", and -profile was the best I found. I saw -compression_level you
> suggest in your other mail, but it is always used for lossless compression,
> so I thought it would be misleading.

i agree with robert that profile is not the correct choice and that 
compression_level would be an option.


> 
> In fact, I am deeply unsatisfied by the option system. The way things work
> right now, you can put "-level 42" in your command line and have it silently
> ignored, just because the selected codec do not use the level field.
> Silently ignoring an option is bad.
> 
> I am not sure what could be done instead. Maybe a codec-specific option
> system would be better. Something like:
> 
> ffmpeg ... -acodec libmp3lame:q=3:aq=2 ...

no

if you want an error message for unsupported options, it should be very
trivial to do once you have a list of supported options for each codec.
And before you have such a list no system will be able to tell you that
the option has no effect.

To clarify how the actual code could work:
Each AVCodec exports the list of options that do have some effect.
ffmpeg checks that list before setting an option and prints a warning if
its not on the list ...

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Many things microsoft did are stupid, but not doing something just because
microsoft did it is even more stupid. If everything ms did were stupid they
would be bankrupt already.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20080731/44341867/attachment.pgp>