Re: xine's osd

On Wed, 2003-08-13 at 04:46, Reinhard Nissl wrote:
> > it's hard to say from the image, but i guess you are probably
> > experiencing some sort of "color leaking"?
> 
> What can I do about that issue?

not much, except fixing our blending code to do something better....
(src/video_out/alphablend.c)

> > the cause is the current blending code for YV12, it doesn't account the
> > subsampling correctly. it is possible that U and V planes are
> > incorrectly blended on edges (the behaviour may depend on even/odd
> > pixels).
> 
> Please explain that a little more, if you think, that I can do anything about 
> that issue in the supplied OSD data, because the same data looks quite ok 
> with 
> xshm.

you are probably sending correct OSD data, it is a bug inside overlay
blender.

xshm doesn't show issues because blending is done at RGB colorspace,
where every pixel has these 3 components. however, the YV12 format
subsamples the U and V planes, that is, the same U value is shared by 4
pixels.

see this M$ page about the formats:
http://msdn.microsoft.com/library/en-us/dnwmt/html/YUVFormats.asp

we would need to be smarter about blending at YV12, doing some kind of
filtering i guess.

regards,

Miguel




-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01