alanwww1 Wrote:Thanks for the testing. Was this test done with YOUR patch or with MINE. Are there any differences between the patches ? Could you please try my samples as well ?
Thx, Alan
The patches I used are identical to the ones you had pulled into xbmc recently. I tried the second source code vc-1 sample and see the same thing you posted in the image. Lots of artifacts - I should add I played back the sample under windows (media player and MPC-HD + Lavfilters + LavCUVID) and initially there are a lot of artifacts there too (I guess from missing key frames or whatever?).
On a different but related subject. If you talk to someone who has influence please please get them to push forward on the implementation of limited/full range colour support in the driver. I looked back over submitted patches and it seems
some work was done for pre-SandyBridge+ GPU's but I can't find anything since.
Additionally I tried Gnif's AE branch to see if it would now work given the updated ELD information available. Unfortunately it does not, not for DTS HD or Dolby TrueHD. It throws a stream of buffer underrun errors (similar to those reported by someone else in that thread using different hardware). When the 3.2 kernel goes final (and hopefully Gnif returns) I'll hope to contribute to testing for AE (enjoyed it for a long time with my Nvidia G210!).
EDIT: Played around a little more today and one thing I noticed was xrandr is reporting just two refresh rates for my TV's default resolution (1280x720); 50 and 60Hz but it can also do 59.94Hz. In Windows 7 the
Intel driver reports three refresh rates; 50, 59.9 and 60. Equally when I use a G210 the Nvidia driver reports the 59.94Hz rate - it labels this as 1280x720_60_0. So not sure what is going on here. A sidenote - and I see you raised this in the mailing list alanwww - the modes reported by the
intel driver are of the form "0000x0000"x0.0 - I'd expect them to be "0000x0000x50.0" or whatever. Or better the same as Nvidia - "0000x0000_00_0" (if that makes sense!).