<div dir="ltr">2014-05-12 14:50 GMT+02:00 David Edwards <span dir="ltr"><<a href="mailto:david@more.fool.me.uk" target="_blank">david@more.fool.me.uk</a>></span>:<br><div class="gmail_extra"><div class="gmail_quote">
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">I can't remember having had any segfaults yet, but it's great to hear<br>
VAAPI is getting some love.<br></blockquote><div><br></div><div>The bug was only triggered if the decoded frame height was not exactly the same as the display height (eg. 1088 instead of 1080), so if you only used true HD material (exact 1080p or 1080i) in your tests you couldn't be hit.<br>
</div><div> <br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
It sounds like they're working on supporting software deinterlacing so<br>
as not to be constrained by the VAAPI library. I would still be<br>
interested to know why the Motion Adaptive deinterlacer is not<br>
available in Mythfrontend even though the Intel release notes claim it<br>
is now supported on Haswell.<br></blockquote><div><br>In order to benefit from the hw deinterlacers available on the different Intel platforms it is necessary to use an extension of the VAAPI (the video processing interface, aka VPP, which was added in libva-1.2.0 in June 2013) to setup a postprocessing pipeline. It is also necessay that the libva-intel-driver implements the corresponding deinterlacer for your device. As of libva-intel-driver-1.3.0 you'll get :<br>
<br>- Broadwell: Bob, MotionAdaptive, MotionCompensated<br>- Haswell: Bob, MotionAdaptive, MotionCompensated<br></div><div>- Ivy Bridge: Bob, MotionAdaptive<br></div><div>- Sandy Bridge: Bob<br><br></div><div>For now, the pipeline setup has not been implemented in MythTV, that's why it's not available in the profiles. I may look at this in a not so distant future, but don't hold your breath...<br>
</div></div></div></div>