[MPlayer-users] filters order: pullup vs postprocessing

[D Richard Felker III]

On Thu, Mar 24, 2005 at 03:49:29AM +0300, Vladimir Mosgalin wrote:
> Hi D Richard Felker III!
> 
>  On 2005.03.23 at 19:09:55 -0500, D Richard Felker III wrote next:
> 
> > > However, pullup seems to perform much worse than detc and filmdint on
> > > this material, it prints a lot of "skip frame" messages - there are
> > > low-motion sequences up to 15 seconds where it can't drop anything at
> > > all. Is there some solution to get both postprocessing and nicer ivtc
> > > process than with pullup?
> > 
> > Explain better what's happening. It sounds like you have a long still
> > sequence during which ALL the fields are from the same source frame,
> > in which case it doesn't make any difference which ones you drop.
> 
> No, it isn't still. There are several scenes in that unfavored sequence,
> with low-motion panning and rotation. When watching them frame by frame,
> telecine patterns are unnoticeable on some parts and hardly noticeable
> on another. -vf spp=6,detc=dr=2 produces very good results (even if it
> is wrong, looks like detc is smart enough to ignore blurring ;) ).
> Anyway, since detc works this way, I'm happy enough. Thanks for writing
> such solid filters! :)
> 
> Btw, does "duplicate frame" message after detc or pullup always mean
> that it was in original material (it's anime and most likely there is
> frame duplication on dvd) or it can be something else?

Duplicate frame means there were too few frames output, and thus
mencoder has to duplicate some of them to fit the fixed-fps. If you
just see one every now and then, it's just a matter of telecined
content not being perfectly mastered, but having bad splices/edits,
most likely. If you see them often, then the content is too messy for
pullup to find sufficiently many clean progressive frames out of it.

Rich