Es schrieb Erik Slagter: > > -----Forwarded Message----- > > From: Erik Slagter <erik@xxxxxxxxxxxxx> > To: video4linux-list@xxxxxxxxxx > Subject: Re: Recommended kernel patches for video capture > Date: 02 Dec 2003 14:37:15 +0100 > > > > BTW you can never trust the synchronisation of TV cards. Therefore > > > capturing programs should always explicitely keep/fix a/v > > > synchronisation. Mencoder does something like that, transcode (cvs) v4l2 > > > import module and nvrec do it too. The video capturing feature of other > > > programs (like ffmpeg, transcode using v4l, mjpegtools) are actually not > > > suitable for capturing video directly. > > > > How does mencoder do this? I have never captured properly synchronized video > > with mencoder (in fact, that's why I switched to transcode :-) > > I used mencoder on my bt848 card and that worked fine. On the SAA7134 > the sync compensation seems to be insufficient. > > -- Hi, nvrec (divx4rec mostly) I'm using now since about March close to daily on my saa7134 md9717 and now a cvs version also with a FV3K on test kernels. I do timeshift playback with xine to Nvidia GF4 tvout, also over nfs mounts, and _never_ noticed user visible a/v sync problems. For mencoder just repeated the maximum load test with the 1 GHz Athlon I did with 2.4.20-v4l2, when Jindrich brought the v4l2-patch, using the same settings on test11 with test10-kraxelpatch-2. (new patch with MT2050 tuner is released) Same result. A/V is completly in sync for 720x576 PAL mpeg4 and audio stereo with lame for what I can see :) Crond was disabled and no user activities, mplayer-cvs-030915. Don't know if it is currently broken. Full log is here (179K): http://homepages.compuserve.de/doxa1121/misc/linuxstuff/menc-test11.out.tar.bz2 Greetings Hermann
