[FFmpeg-devel] [RFC] 16-bit component library

[Øyvind Kolås]

On 5/26/07, erik98 at sunflower.com <erik98 at sunflower.com> wrote:
> >?yvind Kol?s wrote:
> >One thing you do not seem to take into account is the positioning of
> >chroma samples for planar pixelformats using chroma sub-sampling,
>
> There seem to be two options here.
>   1) Explicitly enumerate the extra "components" (ie. Y,U,Y',V = 4 components)
>   2) Add a component sub-index. (ie. Y[0],U,Y[1],V)

The issue I tried to raise is not that a you end up with multiple Y
components, but the offset of the chroma samples in relation to the
luma samples. Some googling brought me to a page with some diagram
showing some different ways the chroma samples can be positioned.

http://www.mir.com/DMG/chroma.html

Different positions for chroma samples in relation to luma samples
means that the interpolation of chroma samples is needed when
converting to a pixel representation without chroma subsampling.

/?yvind K.

-- 
?The future is already here. It's just not very evenly distributed?
                                                 -- William Gibson
http://pippin.gimp.org/                            http://ffii.org/