> >JS> "M. Osten" wrote: >>> >>> I use vcr to capture television programs with a brooktree tuner/capture >>> card. It seems that the channels at the lower end of the spectrum (ie >>> 2,3,4) have poor reception due to bad in wall cables. When I do >>> captures on these channels drop lots of frames in an hour. Channels at >>> the higher end of the spectrum do fine. >>> >>> Machine is a dual P3 933 with 1 gig of ram, and 170 gig of ATA 100 >>> 7200rpm drives. >>> >>> Capture thread exiting >>> Captured 111504 frames. >>> Frames dropped in capture: 652 (0.0%) >>> Frames dropped in encoder: 1191 (0.0%) >>> Exiting.. >>> Waiting for write thread >>> MP3EncodeR::Close() >>> Write thread exiting >>> Waiting for audio thread >>> Waiting for video thread >>> All threads exited >>> v4l1: VIDIOCCAPTURE in ::setCapture >>> v4l1: VIDIOCSYNC in ::~v4l1if >>> v4l1: ioctl VIDIOCCAPTURE in v4l1if::~v4l1if >>> Total Unfree 3596428 bytes cnt 49 [0,0] > >JS> If the signal is weak, the video capture card will often miss the >JS> vertical synchronisation pulse (it gets lost in the background noise). >JS> When this happens, it detects that too many lines have been received in >JS> a frame, and does a resynchronisation, effectively dropping a frame. >JS> The only way around this is to get a better signal. > >If I understand the problem correctly, there is a workaround. I had >similar problem when capturing from bad-quality VCR tapes. Frame drop >was much worse than in this case ( 652 out of 110k frames dropped is >nothing; I was experiencing up to 30-50% drop ). At last I wrote a >small patch for bttv (http://null.ru/people/hamster/bttv-drop.diff.gz) >that completely fixed the problem. It was taken against 0.7.39, >shouldn't be hard to port to newer version. If after applying this you still have the problem with the low frequencies (VHF low), then I think that the tuner programming is the problem. The tuner PLL is programmed differently for cable vs. antenna. These values will vary between tuner types and analog video standards. I suspect that some bad values were used in a lookup table for these low frequencies and these values were propagated from windows code to different projects. I have the same problem with my Bt878 board under windows. To fix it, you could trial-and-error some different PLL values for the low channels. Hope it helps... -- Peter
