>> Hey, I had e-mailed the dev list a month or so ago with a patch that<br>
>was<br>
>> never committed to CVS, I figured since I see mention of a 0.18<br>
>version<br>
>> being released sometime soon it would be good to get this minor<br>
>patch in.<br>
>><br>
>> You can find it by searching the dev group for "patch missing<br>
>keyframes".<br>
>><br>
>> Jay<br>
><br>
>Is this the one about missing keyframes when transcoding using<br>
>mythtranscode? Are you running current CVS and still see this as an<br>
>issue?<br>
>I have transcoded several recordings from original MPEG4 nuppel files<br>
>and<br>
>am seeing the keyframes in the file. Are these new or old .nuv files<br>
>you're transcoding?<br>
><br>
>--<br>
>Chris<br>
<br>
I don't know about THE latest CVS. I know that this first startedfor me
back in November, occurs with the 0.17 release, and still occurs as of
CVS back in mid March when I went from 0.17 to CVS and forgot to apply
my patch. This always happens when I have a RTJpeg recording and then
transcode it into an mpeg4 recording using mythtranscode. When I
debugged the problem it was all occuring in the nuppeldecoder.
Apparently the <font color="black" face="Verdana,Arial,Helvetica" size="2"><font color="black" face="Verdana,Arial,Helvetica" size="2">isLastFrameKey() was never returning true. I tracked the problem down to a line of code that was commented out in November, (</font></font><font color="black" face="Verdana,Arial,Helvetica" size="2"><font color="black" face="Verdana,Arial,Helvetica" size="2">framesPlayed
= frameheader.timecode - 1;). Apparently whenever isLastFrameKey() was
being called, the numbers were always off by 1. Once I uncommented that
line, everything went back to normal and life was good.<br>
<br>
Jay<br>
</font></font>