[FFmpeg-devel] [RFC] DV encoding interface issues

[Roman V Shaposhnik]

On Mon, 2009-07-13 at 10:15 -0700, Jason Garrett-Glaser wrote:
> On Mon, Jul 13, 2009 at 10:11 AM, Roman V Shaposhnik<rvs at sun.com> wrote:
> > On Mon, 2009-07-13 at 10:08 -0700, Jason Garrett-Glaser wrote:
> >> >> 2.  Furthermore, if the user's pix_fmt is, for example, yuv420p, but
> >> >> the height and width match an allowed profile (e.g. 1440x1080), it
> >> >> will *still* terminate rather than perform the colorspace conversion.
> >> >>
> >> >> What's the proper way to deal with these issues?
> >> >
> >> > I'm not sure where this is going but even if it gets to be perceived as
> >> > an issue worth addressing it is clearly an ffmpeg.c issue.
> >>
> >> Perhaps, though it seems ugly to special-case individual codecs in ffmpeg.c
> >
> > Are you suggesting that codec *itself* performs colorspace conversion?
> 
> No, I'm suggesting that the codec better signal ffmpeg in terms of
> what colorspace it needs, so that ffmpeg can do the conversion.

That's a fair request.

Thanks,
Roman.