[FFmpeg-devel] [PATCH] avfilter/vf_tinterlace: Favor using standard timebases for the output
Michael Niedermayer
michaelni at gmx.at
Mon Dec 1 22:55:06 CET 2014
On Mon, Dec 01, 2014 at 08:54:16PM +0000, Vittorio Giovara wrote:
> On Mon, Dec 1, 2014 at 8:04 PM, Michael Niedermayer <michaelni at gmx.at> wrote:
> > Inspired by discussion with kierank
> > Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
>
> Missing a 'Reported by' tag I believe.
added
>
> > ---
> > libavfilter/tinterlace.h | 1 +
> > libavfilter/vf_tinterlace.c | 24 ++++++++++++++++++++++--
> > 2 files changed, 23 insertions(+), 2 deletions(-)
>
> This leaves vf_interlace broken, and having working timestamps was
> among the main reasons of having a spin-off filter. Whatever the fix
> ends up being, please apply it there too.
I dont think the patch will apply cleanly there, also
If correct timestamps are the main reason for vf_interlace, then why
was tinterlace not fixed instead of taking a subset of its features
and creating a new filter out of that ?
also are any of these broken timestamps remaining in tinterlace ?
if so please elaborate on when they occur and how to reproduce
iam interrested in fixing these issues
>
> > diff --git a/libavfilter/vf_tinterlace.c b/libavfilter/vf_tinterlace.c
> > index 133f073..b368e7c 100644
> > --- a/libavfilter/vf_tinterlace.c
> > +++ b/libavfilter/vf_tinterlace.c
> > @@ -35,6 +35,7 @@
> > +static const AVRational standard_tbs[] = {
> > + {1, 25},
> > + {1001, 30000},
> > +};
>
> This effectively limits interlacing to two framerates. What about pure
> 30i?
added
> What about some future (or past) framerate we didn't think of?
> Listing all possible framerate combinations is simply not
> maintainable.
i very much hope that the number of interlaced frame rates will be
finite, small and not growing.
In case that turns out not to materialize that way some other
heuristic can be used instead of a table.
>
> If you reeeeeeally want interlaced vfr why don't just add a filter
> option like "keep_timebase" or something? Imho, it would be enough to
> revert the broken patches.
i can revert it for vf_interlace if you like?
[...]
--
Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
In a rich man's house there is no place to spit but his face.
-- Diogenes of Sinope
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 181 bytes
Desc: Digital signature
URL: <https://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20141201/62aba0b5/attachment.asc>
More information about the ffmpeg-devel
mailing list