Jeroen Brosens wrote:
>
>>I think the combination of the low CPU usage of XvMC and the Vsync
>>provided by OpenGL provides a very good looking picture.  However, if
>>the Bob would actually take place, I think it might indeed be "optimal"
>>as I had originally stated.
>>
> So which version of MythTV are you using, CVS perhaps? I know there were
> some changes to the bobdeint code in Xv and XvMC video output.


I'm running 0.16 right now. With 0.17 I had too many problems with mythbackend crashing.


You are correct. 0.16 has broken Bob for XvMC in that it only displays the first field, so it's not actually displaying double the frame rate. In a previous message to this thread I have posted a link to the patch that fixes that. I have manually applied it to my source code and it does indeed fix Bob. The only issue with it now is that the OSD only displays every other frame, creating a flashing effect. I tried to hack it myself, comparing the videoout_xvmc.cpp file to the latest CVS version to determine the possible change, but I've only managed to get it to either hang on first displaying the OSD, or display the OSD OK but crash mythfrontend when I hit pause. Maybe if I check the commits list I could find the specific fix for the every-other-frame OSD bug with XvMV and Bob. Or if somebody here can point me in the right direction, that'd be nice! :)

But if you're running a post-0.16 CVS version or 0.17 you might not have to much with any of this.

Please let me know how it works out. I thought the onboard SIS graphics on my Pundit was pretty good, but now that I've got the nVidia card working properly, the difference is absolutely stunning! I absolutely cannot tell the difference between a recording and the incoming broadcast signal.

If I hear from at least one other person with similar results, maybe I'll submit my instructions to Jarod or a Wiki or something.


-WD _______________________________________________ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

Reply via email to