[Ffmpeg-devel] [PATCH] fix compilation with cavs decoder

[Stefan Gehrer]

Von: Lauri Hahne

> Guillaume Poirier <gpoirier at mplayerhq.hu> kirjoitti Tue, 04 Jul 2006  
> 11:17:01 +0300:
> 
> > Hi,
> >
> > M?ns Rullg?rd wrote:
> >> "Lauri Hahne" <lauri.hahne at tut.fi> writes:
> >>
> >>
> >>> M?ns Rullg?rd <mru at inprovide.com> kirjoitti Tue, 04 Jul 2006
> 00:41:08
> >
> > [..]
> >
> >>> Ijust pulled the latest revision from svn and now it works. Though I'm
> >>> not  quite confident that this decoder will ever be usefull for nobody
> >>> is  interested in encoding these files.

Disabling it at compile time should work after Reimar's patch is applied.
Feel free to do so.

> >> It's an official standard in China.  Presumably someone over there is
> >> interested in encoding.
> >>
> >
> > Is there any side-by side comparison of cavs VS h.264? I think it
> > would be quite interesting... all the more if cavs guys hasn't make
> > the choice of cabac but chose an entropy encoder that can be made fast.
> >
> > Guillaume
> 
> There is http://www.ee.cuhk.edu.hk/ispacs2005/tutorial/Yu%20Lu.pdf which  
> was mentioned here earlier. It left me with the impression that AVS is a  
> light-weight version of AVC 

agree

> much like VC-1.

disagree. VC-1 is a mess. Ask kostya :)

> Thus I can't see much interest 
> outside China as there already is VC-1 for those who need something like  
> AVC but faster.

I am outside China and am interested in anything BUT VC-1.
And of course in order to achieve world dominance of ffmpeg we should
attract the Chinese to it too :)

On another note:

The way it usually works in standardisation bodies is more or less like
this: Companies and universities having expertise in that area form
a discussion group in which everybody tries to push in his own ideas,
trying to maximise the share they get out of the later to be formed
patent pool and therefore license revenue. The final standard is a 
compromise of these discussions, and these compromises are often not
found on technical arguments alone. Worst example in my experience
is MPEG4 part 2, the amount of unused nonsense in it is ridiculous.
H.264 spec is much better in the aspect that the scope of the spec is
quite exactly what implementations need, but implementing a codec is
IMHO more complex than would strictly be necessary.

Now as far as I understand the AVS way it is roughly like this:
Some academic body with experience in the field of video coding pulls
together a standard, largely consisting of existing technology, with
the goal of finding a good trade-off between coding efficiency
and implementation complexity. Then they seek out to those who might
claim patent rights in the used technology and try to get a good deal
for the licensing. The decisions of what technology to use in the
standard are purely (or at least much more) on technical merit and 
IMHO the result of this is worth a look.

Regards
Stefan Gehrer
-- 


Echte DSL-Flatrate dauerhaft f?r 0,- Euro*!
"Feel free" mit GMX DSL! http://www.gmx.net/de/go/dsl