On Mon, May 31, 2010 at 08:46:25PM +0300, Risto Pajula wrote:
> Any help would be appreciated. I'm also willing to debug the problem
> further if you could give any pointers where to look.
Well, (part of) the file would be most useful for fixing.
Otherwise I expect that e.g. -vo gl would avoid the issue.
Something like this (completely untested) patch might help,
though I expect it will still display wrong:
Index: vo_xv.c
===================================================================
--- vo_xv.c (revision 31291)
+++ vo_xv.c (working copy)
@@ -119,6 +119,10 @@
int stride)
{
x0 += image_width * (vo_panscan_x >> 1) / (vo_dwidth + vo_panscan_x);
+ x0 = av_clip(x0, 0, image_width);
+ y0 = av_clip(y0, 0, image_height);
+ w = FFMIN(w, image_width - x0);
+ h = FFMIN(h, image_height - h0);
vo_draw_alpha_yv12(w, h, src, srca, stride,
xvimage[current_buf]->data +
xvimage[current_buf]->offsets[0] +
