[MPlayer-users] Re: PSNR tests: mbd=2, mv0, trell, and cbp (and...

[D Richard Felker III]

On Fri, Oct 17, 2003 at 08:52:06AM +0200, Stefan Seyfried wrote:
> [Automatic answer: RTFM (read DOCS, FAQ), also read DOCS/bugreports.html]
> rcooley <rcooley at spamcop.net> writes:
>  
> > As for artifacts, I happen to like MPEG1 because it has much more
> > plesant artifacts.  I never see blocking, stair-stepping
> > lines, etc, when I've used it.
> 
> that's my experience, too. MPEG4 has its special artefacts which
> -once you have "locked onto them"- are a bit annoying. It is the
> same with mp3 artefacts for me, once i noticed what to listen for,
> i hear them everywhere ;-)
> This is one reason i do my videolan live streaming here with mpeg1
> instead of mpeg4 - it is easier on the decoders and it looks better.
> And bitrate doesn't really matter.
> It is no question, if size matters most, mpeg4 is the way to go, but
> probably doing it in 2 steps (capturing high-resolution, high-bitrate
> mpeg1, post-processing and transcoding to mpeg4) may be something
> worth trying. IF it is better, we can try to do it in one step after
> creating a "de-detailer-like-mpeg1"-videofilter :-)

Try mpeg_quant. You'll get the same type of artifacts as mpeg1/2
instead of h263/mpeg4, but with all the compression benefits of mpeg4.

Rich