[MPlayer-users] SkipFrame/Duplicate Frame

[Charles Henrich]

> > Skip frame messeges mean that mencoder has dropped a frame, either to keep
> > the he audio syncronized or because of a 3:2 pull down removal.  If this
> > is true is
> it shouldn't happen for pulldown if user specified the exact -ofps

No matter what fps I specify for this DVD it always comes up with some skip
frame and duplicate frame messages.

> > there a way to tell mencoder to encode every frame regardless of the audio
> > sync?  
> yes. -mc 0

Even with -mc 0 I see "skip frame" messages?

> > Duplicate frame messages on the other hand are information meaning
> > mencoder found a duplicate frame, but is happily encoding it (whhich would
> > show up as jerky pans in the video).
> nope. it means mencoder duplicated a frame to keep a-v sync or requested
> -ofps

Is there anyway to force mencoder to not duplicate frames?  Aka "look encoder,
I want an exact frame for frame copy of this source material?"

> > How does mencoder actually do the removal, is it smart enough about the
> > field d encoding to remove the appropriate fields, or does it just whack a
> > whole frame?  (How does this work with deinterlace post processing?)
> whole frame. it works for pulldown

Perhaps this is the trouble Im having with these streams that no matter what
-ofps I select I get a mixture of skip/duplicate frames.

The only real way to do proper 3:2 pulldown removal is to compare fields
against one another, and remove the duplicate field.  I.e. the stream will
contain something like:

Original Speed      F1 F2 F3 F4   (Frame 1-4)
    
Field 1             F1 F2 F2 F3 F4
Field 2             F1 F2 F3 F4 F4

The pattern is not guranteed, and varies by different encoder unfortunatly!
The simple solution is to keep a running count of fields, and whenever a field
occurs three times, skip the third field.  Which would give you the correct
output rate (calculated) as well as a perfect progressive playback.

-Crh

       Charles Henrich         Eon Entertainment         henrich at msu.edu

                       http://www.sigbus.com:81/~henrich