2013-01-21, 20:46
@rogerjames99:
OpenELEC uses 12.10 of Catalyst. There is no plan to change this (yet) - as 13.1 has some issues, especially with 50i/50p with Vertical Blank Sync enabled.
Btw. as you are watching 1080i50 - could you check with Adjust Refreshrate to match Video enabled (it should switch to 50hz) and Vertical Blank Sync always enabled - if there are any drops with your hardware?
I tried to debug this artefacts last weekend. My testing revealed that there are some Reframes missing, but the "stream still references them" - so one has to check the ReFrames referenced by the current picture and remove all of them that did not pass yet :-) A similar thing is done within vdpau, as the decoder has the option to change the actual references (pointers) to the non existing Reframes. I did not find a solution for AMD yet - as I don't know where to store this updated table - but in theory it should work - if we can make sure, that the first Reframe is the first in the stream an no upcoming frames references Frames before this - it is not easy. I hope AMD implements it in their driver - as it should be quite doable, as all the frames + reframes must be used within the decoder.
I will ask them at Februar 1st :-)
OpenELEC uses 12.10 of Catalyst. There is no plan to change this (yet) - as 13.1 has some issues, especially with 50i/50p with Vertical Blank Sync enabled.
Btw. as you are watching 1080i50 - could you check with Adjust Refreshrate to match Video enabled (it should switch to 50hz) and Vertical Blank Sync always enabled - if there are any drops with your hardware?
I tried to debug this artefacts last weekend. My testing revealed that there are some Reframes missing, but the "stream still references them" - so one has to check the ReFrames referenced by the current picture and remove all of them that did not pass yet :-) A similar thing is done within vdpau, as the decoder has the option to change the actual references (pointers) to the non existing Reframes. I did not find a solution for AMD yet - as I don't know where to store this updated table - but in theory it should work - if we can make sure, that the first Reframe is the first in the stream an no upcoming frames references Frames before this - it is not easy. I hope AMD implements it in their driver - as it should be quite doable, as all the frames + reframes must be used within the decoder.
I will ask them at Februar 1st :-)