[FFmpeg-devel] [PATCH 3/4] Add setpts filter by Victor Paesa.

[Stefano Sabatini]

On date Wednesday 2010-06-09 03:39:04 +0200, Michael Niedermayer encoded:
> On Wed, Jun 09, 2010 at 12:17:35AM +0200, Stefano Sabatini wrote:
> > ---
> >  doc/libavfilter.texi     |   46 +++++++++++++++++
> >  libavfilter/Makefile     |    1 +
> >  libavfilter/allfilters.c |    1 +
> >  libavfilter/vf_setpts.c  |  127 ++++++++++++++++++++++++++++++++++++++++++++++
> >  4 files changed, 175 insertions(+), 0 deletions(-)
> >  create mode 100644 libavfilter/vf_setpts.c
> > 
> > diff --git a/doc/libavfilter.texi b/doc/libavfilter.texi
> > index 1de27d9..cf53a93 100644
> > --- a/doc/libavfilter.texi
> > +++ b/doc/libavfilter.texi
> > @@ -237,6 +237,52 @@ ratio of the input image.
> >  
> >  The default value of @var{width} and @var{height} is 0.
> >  
> > + at section setpts
> > +
> > +Change the PTS (presentation timestamp) of the input video frames.
> > +
> > +Accepts in input an expression evaluated through the eval API, which
> > +can contain the following constants:
> > +
> > + at table
> > + at item PTS
> > +the presentation timestamp in input
> > +
> > + at item PI
> > +Greek PI
> > +
> > + at item E
> > +Euler number
> > +
> > + at item AVTB
> > +the FFmpeg Time Base
> > +
> > + at item N
> > +the count of the input frame, starting from 0.
> > +
> > + at item STARTPTS
> > +the PTS of the first video frame
> > + at end table
> 
> missing:

> average fps

computed in which interval?

> interlace flag
> file position

> timebase

which timebase? As far as I know the timebase in the filterchain is
fixed, and is AV_TIME_BASE.

> guessed real minimal timebase

please suggest an algorithm for computing it if you have already an
idea

> previous input pts
> previous output pts

I added them in a previous version, but then I didn't know how to deal
with them for the first frame. I suppose they may be set to 0 or
AV_NOPTS_VALUE, but then I need to understand how can I deal with
them, for example in the expression: outpts=prev_outpts+XXX.
 
> obviously some of these could be usefull

Also I suggest to add the more complex features in successive patches,
this is a port afterall.

Regards.
-- 
FFmpeg = Fascinating and Fierce Merciless Pitiful Efficient Gadget