On Saturday 03 May 2008, Malcolm Caldwell wrote: > On Fri, 2008-05-02 at 14:30 +0100, Laz wrote: > > I've just built a new vdr box with vdr-1.7.0 (with liemikuutio path > > but I'm pretty sure this stuff isn't touched), softdevice cvs from a > > few days back, ffmpeg svn from a few days back (SVN-r12942 it > > claims), DirectFB-1.1.1, a Matrox G450 for output. > > > > All works fine apart from when I try to scan a few seconds backwards > > through a recording with trickspeed 2x. When I begin replay, it has > > gone back much further than it should have (and has done so in the > > past with a similar setup!). I haven't tested this in a forwards > > direction but think it will probably be the same. Am I missing > > something obvious here or are other people seeing this? > > I have a similar setup (g450) and can confirm this problem. > > > Looking at the HISTORY file for vdr, vdr-1.5.15 has: "Improved > > sending all frames to devices that can handle them in fast forward > > trick speeds, including subtitles". This is through > > the virtual function bool cDevice::HasIBPTrickSpeed(void) { return > > false; } which doesn't seem to be implemented by softdevice and so > > the return value should be false. > > > > What I'm seeing is almost as if softdevice is only expecting I-frames > > during trickspeed but vdr is sending it all frames and so when > > playback returns to normal speed, it is many frames away from where > > it should be. > > > > Another thing I've noticed is that if I press "0" during replay to > > add a cut-mark, it adds it about a second or so after when I pressed > > it. I know it will put the cut mark at the nearest I-frame but this > > is a much larger jump.
A little bit more information after a few more tests: I have keys '1' and '3' set in cReplayControl::ProcessKey in menu.c as follows: case k1|k_Repeat: case k1: SkipSeconds(-10); break; case k3|k_Repeat: case k3: SkipSeconds( 10); break; I.e. skip 10 s backwards or forwards. I'm pretty sure that these used to give jumps which were pretty close to 10 s in each direction. I now get 7 s backwards and 14 s forwards instead of 10 s! Jumping a minute forward or backwards gave me something like 53 s backwards and 1 min 14 s forwards! Not sure how close they used to be to a minute, though. If I pause and then press a forward or backwards key for slow motion, forwards works as expected. Backwards runs back rapidly for about a minute (of time in the recording) in about a second, then slows for a further 10 s or so (of recording time) and then finally starts stepping slowly backwards! Cheers, Laz _______________________________________________ Softdevice-devel mailing list Softdevice-devel@lists.berlios.de https://lists.berlios.de/mailman/listinfo/softdevice-devel