[FFmpeg-devel] [PATCH] imgconvert: remove PixFmtInfo.pixel_type field

Stefano Sabatini stefano.sabatini-lala at poste.it
Mon Jul 25 11:31:20 CEST 2011


On date Sunday 2011-07-24 16:31:09 +0200, Stefano Sabatini encoded:
> On date Sunday 2011-07-24 16:16:35 +0200, Michael Niedermayer encoded:
> > On Sun, Jul 24, 2011 at 01:45:06AM +0200, Stefano Sabatini wrote:
> > > On date Saturday 2011-07-23 17:12:01 +0200, Michael Niedermayer encoded:
> > > > On Sat, Jul 23, 2011 at 03:58:56PM +0200, Stefano Sabatini wrote:
> > > > [...]
> > > > >      [PIX_FMT_NV12] = {
> > > > >          .color_type = FF_COLOR_YUV,
> > > > > -        .pixel_type = FF_PIXEL_PLANAR,
> > > > >          .depth = 8,
> > > > >      },
> > > > >      [PIX_FMT_NV21] = {
> > > > >          .color_type = FF_COLOR_YUV,
> > > > > -        .pixel_type = FF_PIXEL_PLANAR,
> > > > >          .depth = 8,
> > > > >      },
> > > > 
> > > > does the output match for these ?
> > > 
> > > No, and I don't think the av_picture_crop/pad code for NV12/21 ever
> > > worked.
> > 
> > no disagreement here
> > 
> 
> > but to me NV12 is a planar format, theres a luma plane and a chroma
> > plane
> 
> The problem is the definition of planar:
> 
> 1) planar is when every component is on a separate plane, so you can
>    tell the component from the plane (one-plane-per-component)
> 
> 2) planar is when there are at least two distinct components which
>    reside on different planes
> 
> I tend to prefer definition 1), and the code seems to agree with that
> either.
> 
> > not that i mind it much when the code doesnt work either way.
> > so commit what you prefer

And... pushed.
-- 
FFmpeg = Foolish and Fast Maxi Powered Everlasting Gem


More information about the ffmpeg-devel mailing list