[MPlayer-users] new version of detc filter! (inverse telecine)

[D Richard Felker III]

On Sun, Mar 16, 2003 at 01:20:08AM -0600, John Evans wrote:
> [Automatic answer: RTFM (read DOCS, FAQ), also read DOCS/bugreports.html]
> I have finished some initial testing of -vop detc on the Star Trek : TNG
> series DVDs.
> 
> Your filter is by far the best solution for mencoder. It produced results
> similar to using
> decomb on the Win side. There is just a little more areas of interlacing
> slipping through
> than on decomb.
> 
> Are there any thresholds to control the amount of combing that is detected?
> 
> I saw the tr values but did not know whether they applied to the filter or
> filter options.
> 
> The telecide filter within decomb has a threshold option that can be set to
> detect
> differences at the field level and uses the field that best matches the one
> that pre/procedes
> it.
> 
> Of course, there will still be choppy motion in some SFx scenes since I
> would rather see
> 8 to 15 % of the video with a dropped frame rather than 85 % 92% of the
> live-action with
> a double frame every 5 frames.
> 
> I still have to try -ofps 29.97 & -vop ~,detc=dr=1. My guess is that the
> results would be the
> same as -ofps 23.976 & -vop ~,detc. Doing -ofps 23.976 & -vop ~,detc=dr=1
> produced
> 100 % choppy video ( dropped frame every 5 ). ( ~ = my crop, resize
> params ).

Hmm. mencoder -ofps 23.976 -vop detc (without dropping) should not
work right at all. The problem is the the filter won't drop frames
when it can't detect any interlacing pattern, so mencoder will drop
the frames instead before passing them to the filter layer. This will
totally ruin the interlace detection, since detc might get an
interlaced frame, but not have the correct previous frame to compare
it to.

Basically, if mencoder says "skipping frame" more than a couple times
at the beginning, you can be sure that detc will NOT work as expected.
:(

Rich