[Ffmpeg-devel] Re: [PATCH] dv pixel format encoding fix

Roman Shaposhnick rvs
Sun Mar 19 22:22:21 CET 2006


On Fri, Mar 10, 2006 at 10:14:43AM +0100, Baptiste COUDURIER wrote:
> >   All this is fine, but I thinks this is really getting to a point where
> >   this has to be per profile. For example, your patch doesn't take into
> >   account the fact that encoding 411P for NTSC could mean either IEC 61834
> >   or SMPTE 314. The VAUX STYPE is also not quite clear for IEC 61834.
> >   I'll have to either try digging up something resembling an IEC spec
> >   or look at the IEC 61834 raw output. Stay tuned ;-) 
> > 
> I assume that 411P is SMPTE 314M, revision is from 2005. Let's forget
> about IEC 61834 for 411P. The only case where VAUX STYPE is set is for
> 422P and that is obviously SMPTE 314M.

  Ok. I almost give up. Reverse engineering of how other codecs tell SMPTE 
  and IEC apart didn't really paint a coherent picture. Which means that
  at the moment we really have just two choice:
     1. Somebody interested in ffmpeg's DV codec supporting IEC61834-2 on the
     encoding side contacts me and either provides an access to the spec
     or at least answers my questions.  

     2. We just give up and go SMPTE all the way. This, of course, will
     make supporting consumer camcorders a big nuisance.


P.S. Speaking of IEC -- I've just noticed that these guys have quite a
few "extensions" to the original IEC61834-2: HD DV, PALplus, etc. So, 
if anybody cares about these things -- giving me access to the specs (~$400)
might not be a bad idea at all.

More information about the ffmpeg-devel mailing list