Success! I checked out a new copy of DirectFB form CVS, patched it with your patch, built and installed it. Rebuilt and installed DirectFB-extras against the new version of DirectFB.
New kernel source, patched with patcher2k patch, and then your via_fbobj.diff and via_vt1622a.h.diff (the latter applied by hand because applying it with patch failed). Installed. And it now works properly, i.e. that interlace_test file plays smoothly how it is supposed to. Going back to basics definitely paid off! What's more, vdr works fine with softdevice (apart from some A-V sync problems but that is down to the plugin), even though I thought it only did such field-synchronisation with Matrox cards. BBC News24 is a good channel for testing for interlace artifacts: there is a scrolling newsticker thing across the screen which looks perfect with the patched kernel module and DirectFB. CPU usage seems to be ca. 50% when watching live TV, depending on the bit-rate. The OSD is much sharper and more responsive than with the xine output plugin because it is using its own layer. Hopefully the A-V sync problems will be fixed soon so it outperforms the xine plugin even more! Are there plans for these patches to be incorporated up-stream, or are they still a bit too 'hacked together'? Cheers, Laz _______________________________________________ directfb-users mailing list [email protected] http://mail.directfb.org/cgi-bin/mailman/listinfo/directfb-users
