[MPlayer-users] MENCODER: AuDIO OF SYNC when recording from TV

[D Richard Felker III]

On Mon, Aug 11, 2003 at 10:43:09AM +0200, Bokor Norbert wrote:
> [Automatic answer: RTFM (read DOCS, FAQ), also read DOCS/bugreports.html]
> > I dunno about Voyager, but ST:TNG is NOT interlaced but rather
> > telecined. If the same is true for voyager, you need to inverse
> > telecine instead of using a deinterlacer of you'll kill quality and/or
> > make ghosts (depending on which deinterlacer).
> 
> what deinterlacer do you recommend? I use -vf lavcdeint. Is there any better method? I'm thinking of making some
> tests, but if you have any suggestions, I can avoid this :)

lavcdeint will definitely create ghost outlines from the other field
when two fields differ by too much, so like pp=lb, it's very bad for
telecined material. (But for this you should inverse telecine anyway!)
On the other hand, with true interlaced content, the two fields
usually don't differ by too much, so pp=lb and lavcdeint would be ok
choices.

Actually, my preferred filter for true interlaced content would
probably be pp=md. You can also use pp=ci or pp=li (cubic or linear
interpolation) -- these are correct deinterlacers, meaning that they
only use data from one field at a time, rather than trying to mix the
two (which don't go together) -- but that also means you're throwing
out half the information, so you might as well just capture one field
to begin with.

Ultimately I don't have any good recommendation for you. Interlacing
sucks, and most of the ways to deal with or remove it also suck...

Rich