On Mittwoch 11 Oktober 2006 01:53, Herbert Poetzl wrote:
> On Tue, Oct 10, 2006 at 11:57:57PM +0200, Stefan Lucke wrote:

> > 
> > Yes changing pixelformat is the other way to circumvent this segfault.
> > Please get an updated version from cvs, as the bug was reproducable
> > for me and I commited a fix.

There was still an issue with color jumping, when cut lines from left
was set to an odd value. Fixed now.

> 
> seems to work fine here, thanks!
> 
> btw, a few things I noticed regarding softdevice
> but I didn't bother to mention (until now), but 
> maybe that input is appreciated:
> 
>  - depending on the pixel format, the entire
>    picture and overlay shifts slightly left and
>    right, where:
> 
>    - YV12 is one pixel right compared to YUY2

One pixel ? That _was_ much more as for some reason edge_with was 16,
which must be multiplied by 2 for YUV2 offset calculations.

>    - YV12 and I420 are perfectly aligned
> 
>  - the pseudo alpha blending never worked for me
>    neither with different nvidia cards nor with
>    matrox cards

Ups, both modes work perfectly fine for my G550 ;-) .

> 
>  - I would expect the overlay to be pixel aligned
>    (as it is on the full featured card) i.e.
>    without strange scaling/alignment effects
>    while in fact it changes slightly with each
>    channel (duplicating pixels and sometimes
>    even antialiasing pixels) see

Could be an edge_width problem too (<= vs <) .

>   
>    http://www.13thfloor.at/~herbert/chan{05,14,26}.png

That link does not work.

> 
> that aside, I'm happy that softdevice was not
> abandoned and seems to be well maintained now
> 


-- 
Stefan Lucke
_______________________________________________
Softdevice-devel mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/softdevice-devel

Reply via email to