Peter:
    Have a look at the yait filter (Yet Another Inverse Telecine)
filter.  It's pretty kludgey but will let you force an inverse telecine
and hence enforce the frame rate.  It requires a pre-pass to build a log
file.  You then run an analysis tool (tcyait), which then writes out an
operations file to tell transcode exactly what to do, ie. blend, drop,
combine rows, etc.

Here is a quick summary:

    $ transcode ... -J yait=log
        - generate the log file

    $ tcyait ...
        - analyze the log (detect telecine patterns), and write an
operations file

    $ transcode ... -J yait=ops
        - generate the video using the pre-computed ops file, (one or
two pass)

    If this doesn't daunt you, I can give you specific examples.
 
    The filter is VERY good at detecting interlaced frames, and attempts
to reconstruct the original progressive frames rather than blending to
eliminate them.  It's just hard to use.  I was never able to automate
the tcyait component.  It needs a human to tweak threshold values if you
want to achieve 100% progressive frame reconstruction.
                                                                      
                    Allan


On 06/24/11 01:59, Peter Federighi wrote:
> Hello.
>
> I am trying to convert some sci-fi VHS recordings to DVD.  The happen to be 
> films and are telecined.  So, ivtc and decimate to the rescue, right?  Well, 
> not entirely.  Since the movies have many scenes in outer space, the video is 
> mostly black, which I've found doesn't work so well with ivtc.  Thus, it 
> doesn't always convert and get rid of the frames it's supposed to which makes 
> the video longer than it should.  It is almost always visually unnoticeable, 
> but every time it happens the audio gets a little more out of sync,  By the 
> end of the movie it's fairly obvious that the audio doesn't match the video 
> (which I hate).  Is there a way to stretch the audio in order to make it 
> match the video?
>
> As far as I can tell, the telecining pattern never changes.  Is it possible 
> to specify at what frame the inverse telecine should start and to never 
> deviate from the pattern even if it thinks that there are no interlaced 
> frames?
>
> Thank you,
> - Peter Federighi
>
>

Reply via email to