[MPlayer-DOCS] [PATCH] Choosing the video codec

Rich Felker dalias at aerifal.cx
Mon Dec 26 20:21:14 CET 2005


On Mon, Dec 26, 2005 at 05:21:21PM +0100, Guillaume Poirier wrote:
> Diego Biurrun wrote:
> > On Mon, Dec 26, 2005 at 03:30:22PM +0100, Guillaume Poirier wrote:
> 
> [..]
> 
> >>  <systemitem class="library">x264</systemitem> instead of MPEG-4 codecs
> >>  such as <systemitem class="library">libavcodec</systemitem> or
> >>  <systemitem class="library">XviD</systemitem>
> > 
> > 
> > missing period
> > 
> > Ouch, you HAVE TO mention here that newer codecs trade quality for
> > speed, H.264 is no fun with less than a 1GHz (2GHz?) CPU..
> 
> Just to prevent ppl from keeping thinking that H.264 is god awfully
> slow, my Celeron M 1.3Ghz decodes (close to) DVD-size using only around
> 50% of CPU time.
> I don't know what is the minimum CPU for most CPU flavours thought.

Celeron M is insanely fast. Minimal target cpu is 300-400 mhz celeron
(p2 generation) or so, IMO. I regularly play movies on a 366 and it
can play anything properly encoded without slowass codecs (h264 and
aac).

> Well, I do see that you're jocking here... I've used the 3 above
> mentioned codecs, and I like them all for different reasons, and what
> thing is clear to me: it's darn useless to try to find which of those
> codecs are better when there are so many options to play with that it's
> as easy to screw things up...

Well I partly agree. I do think it's good to promote lavc especially
though since it's highly under-appreciated. The whole doom9 fiasco
hasn't helped that...

> Not to mention that the picture quality you'll get with ASP codecs also
> (sadly) depends on the decoder (which lead to the flamewar on
> mencoder-users), just as much as the kind of monitor/video projector/TV
> you're gonna use, each of which make some encoding artifacts more visible.

This is only the case if you're really anal. I've never noticed a
difference between idcts during normal watching, but then again I
don't use an artifact-amplifying lcd either..

Rich




More information about the MPlayer-DOCS mailing list