[MPlayer-users] Question on converting to DivX4...

[Rich Felker]

On Fri, Jul 08, 2005 at 06:54:14PM -0700, RC wrote:
> On Fri, 8 Jul 2005 10:01:11 -0400
> Rich Felker <dalias at aerifal.cx> wrote:
> 
> > > I was refering to the issue mplayer has when detelecining/
> > > deinterlacing.  Certain frames are displayed for different lengths
> > > of time, so panning usually looks (unnecessarily) jumpy.  However,
> > > after deinterlacing and re-encoding, mplayer can play back the
> > > progressive (30fps) video perfectly smooth.
> > 
> > This is true for inverse telecine from 60i to 24p. 
> 
> Yes.  Can the problem be fixed?  (la la la...  not mentioning
> threads...)

yes, with a total rewrite of the mplayer video core to actually use
timestamps.

> > > A matter of preference.  I prefer 30fps 720x480 material, rather
> > > than 60fps 720x240 material...  (ghosts and all)
> > 
> > If you really insist on throwing away half the content in your video
> > and making it choppy, I won't argue with you. But properly
> > deinterlaced 60p video looks REALLY good. The difference in smoothness
> > is amazing.
> 
> tfields halves the height rather than halving the FPS, so it's
> throwing away half the picture as well.

no it is not. the original content is 'half height'. interlaced video
works like this: every 1/60 second (or 1/50 for pal) there is an
independent picture consisting of 240 scanlines (or 288 for pal). they
are alternately displayed at 1/2 scanline offset from each other so
that during complete stills, the viewer will have the illusion of
double resolution, but this leads to all sorts of problems whenever
the video is not still. THE ORIGINAL 'FRAMES' are 240-pixels tall!!
combining 2 fields into one pseudo-frame is an artifact of computer
video processing and is totally unnatural.

> What tfields does may be more
> easily reversible than what pp=lb does, but either way, you lose about
> half.

no. tfields loses no information, pp=lb loses up to half the
information, same for pp=md. pp=li or ci ALWAYS loses half the
information. (first 2 are not entirely true due to rounding errors and
such, but close enough..)

> Of course, if mplayer could properly playback interlaced video on an
> interlaced display, I wouldn't have to throw away anything...

tfields throws nothing away. playing back video encoded with tfields
to an interlaced display should look identical to watching the
original interlaced video...

rich