On 07/30/2010 10:36 AM, Josh Allmann wrote: > On 30 July 2010 00:22, Martin Storsjö <[email protected]> wrote: >> On Thu, 29 Jul 2010, Josh Allmann wrote: >> >>> Here is a packetizer and a depacketizer for VP8 RTP. >>> >>> Playback is only semi-smooth; it drops way too many frames. The >>> depacketizer likes to claim it gets an extra packet after the >>> frame-end (which is typically one byte smaller than the actual >>> frame-end packet). Still a work in progress. >> >> Didn't feel any significant frame dropping here (btw, if playing with >> ffplay, make sure you use -noframedrop, sometimes ffplay constantly feels >> it's behind and needs to drop some frames to catch up, but since the >> source is realtime, it still has to block for more frames, so it >> constantly drops frames), but the playback console is spammed with >> messages like this: >> >> [vp8 @ 0x10188fe00] Unknown profile 6 >> [vp8 @ 0x10188fe00] Header size larger than data provided >> [rtsp @ 0x10202dc00] Received no start marker; dropping frame >> > > -noframedrop helped, but the last message is what I am concerned with; > it really should not happen. Those seem like phantom packets. >
Assert on it and see what's the content of buf. You end up there if you miss the start packet. btw do we cleanup when the seqno jumps? 1 2 3 4 5 6 7 8 9 S _ _ _ E S _ _ E What happens when we miss both 5 and 6 ? lu -- Luca Barbato Gentoo/linux http://dev.gentoo.org/~lu_zero _______________________________________________ FFmpeg-soc mailing list [email protected] https://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-soc
