[MPlayer-dev-eng] Re: ccache

Billy Biggs vektor at dumbterm.net
Mon Jan 6 05:29:31 CET 2003


Tobias Diedrich (td at sim.uni-hannover.de):

> Billy Biggs wrote:
> 
> > I don't really think this is valid, since depending on your standard
> > for 4:2:0 the chroma samples are located in different places.
> > Sometimes the chroma samples are from the middle of the four Y'
> > samples, sometimes it's between the two left samples.
> 
> Huh? Why should the be in different places according to the above ?

  In MPEG2:

  Y1    Y2
  C
  Y3    Y4


  In MPEG1, H.261, H.263, and most JPEG images:

  Y1    Y2
     C
  Y3    Y4

  And you want to output this:

  [Y1,C1]    [Y2,C2]

  [Y3,C3]    [Y4,C4]

  And so you interpolate differently to get these values depending on
where the chroma sample is positioned.  These are both 4:2:0 but the
chroma value is in a different location for interpolation.

  This article shows using Catmull-Rom interpolation for the MPEG2 case
of 4:2:0 to 4:2:2.

  http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnwmt/html/YUVFormats.asp

  Note that you have to convert interlaced MPEG2 images differently than
progressive MPEG2 images, as the chroma is also interlaced.

-- 
Billy Biggs
vektor at dumbterm.net


More information about the MPlayer-dev-eng mailing list