[mythtv] Ticket #4872: Detect if video is letterboxed and switch to corect fill mode

[Daniel Kristjansson]

On Wed, 2008-03-19 at 13:32 +0100, Mårten Bohlin wrote:
> On Wed, Mar 19, 2008 at 12:20 PM, Doug Larrick <doug at parkercat.org> wrote:
> > Mårten Bohlin wrote:
> >  > Yes "2" means "XVMC_IDCT_MPEG2". Unfortunately I do not know how all
> >  > that stuff works.
> >  >
> >  > Could anybody tell me if it is possible to analyze a decoded frame
> >  > when you run XvMC? And if that is possible where should I look for the
> >  > decoded frame?
> >
> >  No, I'm afraid MythTV has no access to the decoded frame when using
> >  XvMC.  The decoding takes place on the video card into its own memory.
> >  For the same reason, the deinterlacers available when using XvMC are
> >  severely limited.
> >
> >  -Doug
> >
> 
> OK. So this is another argument that this detection should be done
> together with commercial detection when possible.
> 
> But I do not have the time to do that now.

Marten, yes detection during commercial detection would work with
more video renders. DVDV and XvMC frames in particular can not be
pulled from video RAM efficiently (when it is possible at all).

But I think if the kinks are worked out detection at the playback
stage is valuable in itself. Many MythTV features are only available
with some of the Video Renderers, and sometimes only with specific
hardware.

Also, if the code is clean and well compartmentalized it makes it
easier to add support for it in the commercial detector later on.

-- Daniel