On Tue, Jun 15, 2004 at 11:46:53AM +0200, Stefan Lucke wrote:
> Quoting Denis Oliver Kropp <[EMAIL PROTECTED]>:
> > Quoting Ville Syrj�l�:
> > > On Tue, Jun 15, 2004 at 12:51:03AM +0200, Stefan Lucke wrote:
> > > > On Dienstag, 15. Juni 2004 00:15, Ville Syrj�l� wrote:
> > > DirectFB doesn't have any method of specifying source cropping for layers.
> > 
> > > It would be a useful thing to add though.
> > 
> > It's already prepared to being added easily, at least the driver API covers
> > source AND destination rectangles.
> 
> As easy that you could give a time frame ?

It needs a public function (trivial) and some work in the drivers to make 
it work. I can implement it for matrox and mach64. Now, what should we 
call the public function? Pan() maybe?

> > > But you since you're looking to fill the entire screen you don't actually 
> > > need source cropping. SetScreenLocation() will suffice provided that the 
> > > driver crops things properly. You just have to calculate the coordinates 
> > > and size accordingly.
> > 
> > Or you configure the layer to have the pixel size of the active area and
> > copy only those pixels to the surface.
> 
> Thats the way I did it for softdevice vidix out. I'll try that.
> 
> But with I420 and YV12 output formats I get sig11s with my radeon. Pitch
> value returned from videoSurface->Lock seems to be bogous, see my mail on
> directfb-users list.

I lost the original mail but I looked at the archives...

The pitch 752 does seem strange. It's a multiple of 16 but so is 720 
so I don't know why it would choose 752.

As for the segfault it looks like a bug in the softdevice code since it's 
clearly trying to copy too much data.

I just had a look at vdr-softdevice-0.0.5 and it doesn't even handle 
planar YUV. What code are you using?

-- 
Ville Syrj�l�
[EMAIL PROTECTED]
http://www.sci.fi/~syrjala/


Reply via email to