[MPlayer-users] Re: slightly confused by mencoder output

[Rich Felker]

On Thu, Sep 15, 2005 at 11:43:58AM +0200, Matthias Wieser wrote:
> Am Donnerstag, 15. September 2005 00:18 schrieb Rich Felker:
> > On Wed, Sep 14, 2005 at 07:15:31PM +0200, Matthias Wieser wrote:
> > > Am Mittwoch, 14. September 2005 18:22 schrieb Rich Felker:
> > > > > > or 2:2:2:2:2:2:2:2:2:2:2:3. :) Also there's no guarantee
> > > > > > that 2:2 PAL is in phase with the field pairing of the capture
> > > > > > device.
> > > > >
> > > > > There is no formal guarantee but it works for 99,...% of all
> > > > > cases. To be precise: I have never had any captured video which
> > > > > was out of phase. Oh,
> > > >
> > > > It's been reported on the list fairly often, and was important
> > > > enough that someone even made a filter for it..
> > >
> > > I think, those were caused by buggy kernel modules or video cameras.
> >
> > A video camera does not do pulldown. It generates _VIDEO_, which is
> > interlaced...
> 
> I thought that both interlaced video and 2:2 can get out of phase. At 
> least I have seen normal interlaced video where bob made motions appear 
> like forth-back-forth-back-... (kernel bug...).

That's not out of phase, but reversed field dominance. And it's not a
kernel bug (unless the field dominance is mislabelled), just
misinterpretation by the app which is assuming top-field-first.

Out of phase capture of 2:2 pulldown:
 A A B B C C D D E E F F G G
  |   |   |   |   |   |   |
(lines indicate how the fields are paired in captured pseudo-frames)

Reversed field dominance (video or post-pulldown film):
 B A D C F E H G J I (video)
 B A C B D C E D F E (film)

Rich