[MPlayer-cvslog] CVS: main/libaf af.c,1.35,1.36

[D Richard Felker III]

On Mon, Dec 27, 2004 at 02:22:42PM +0100, Reimar Döffinger wrote:
> Hi,
> > > > > > > CVS change done by Reimar Döffinger CVS
> > > > > > >
> > > > > > > Update of /cvsroot/mplayer/main/libaf
> > > > > > > In directory mail:/var2/tmp/cvs-serv8321/libaf
> > > > > > >
> > > > > > > Modified Files:
> > > > > > >  af.c
> > > > > > > Log Message:
> > > > > > > Use lavcresample when accuracy-optimized audio filter chain is
> > > > > > > requested.
> > > > > >
> > > > > > shouldn't it be dependant on wether lavc is compiled in?
> > > > >
> > > > > Yes, you are right... Actually on thinking once again it was quite a
> > > > > stupid patch. What do you think about the attached one?
> > > >
> > > > Hmm, I'm not sure, are there any advantages for the lavcresampler, IIRC
> > > > the mplayer one was pretty good, but is lavc one faster?
> > >
> > > the mplayer one is just about the most broken resampler code i've ever
> > > seen. it has errors that mangle the audio, sometimes dropping samples
> > > and perhaps even desyncing which channel is which. it's also slow,
> > > except in the ultra-low-quality "linint" mode (which is misnamed, it's
> > > really nearest-neighbor sampling).
> > >
> > > the only (re)sampler i've ever seen that's as bad or worse was written
> > > by yours truly about 5 years ago.. :)
> > eh? you mean the SDL audio built in? =) If it's so bad, than why not remove it 
> > and only allow resampling if mplayer is compiled with lavc (which it should 
> > be in 99% of cases anyways)?
> 
> I though resample was a bit faster... Well, do you want me to change it
> to always use lavcresample when libavcodec is compiled in? I actually

with force= set for speed, please don't use lavc!!
i use the shitty 'linint' resampler because anything else is too
slow..

rich