[MPlayer-dev-eng] [RFC] preliminary x264 encoding help text

[Guillaume POIRIER]

Hi,

On 5/2/05, Ken Johanson <ken at kensystem.com> wrote:
> Guillaume POIRIER wrote:
> 
> >I'm not sure what profile x264 conforms to, but apparently, Google
> >being my friend:
> >http://forum.doom9.org/showthread.php?s=&threadid=89979
> >
> >x264 limitations: "x264 doesn't do any of extended profile or high
> >profile. (Though "extended" is misleading, since it's all about error
> >resilience, and is worse in compression ratio than main profile.)
> >Properly dealing with adaptive interlacing in all cases takes lots of
> >code. x264 doesn't support interlacing.
> >There are a few other features not supported..."
> >
> >
> >
> >
> So presumable it does support Main profile then, since that wasn't
> mentioned in the list of unsupported profiles..
> 
> I'm trying to figure out if the weighted prediction is pretty much as
> good as it gets; I've seen several clips that don't encode gracefully
> when there is luminance fluctuation caused by camera auto-iris.. Having
> tried multiple other AVC encoders and getting the same results, I'm
> wondering if this is a limitation of the AVC design/spec (the same clips
> encodes with much fewer artifacts when using another 'popular' codec).

You should probably ask this kind of question on x264's ML or on doom9
forum, but AFAIK, x264 still need some tuning, in the sense that all
the features that it supports lacks good heuristics so that the
decisions the encoder makes yields the best visual quality.

And that's possible that you faced that kind of problem, which doesn't
mean that x264 is bad nor weighted prediction useless.

Regards,
Guillaume