On Fri, 2004-10-29 at 20:34, Martin Samuelsson wrote: > On Thursday 28 October 2004 13:29, Richard Ellis wrote: > > > Although I notice today that the whole picture jumps up and down a small > > > amount when played back VCR->DC10->xawtv and this vertical shaking is > > > visible in the lavrec recorded avi and the encoded mpeg (on the computer > > > and tv). The video played back VCR->TV is steady. Has anyone else had > > > this? > > > > This might be the source of your troubles. It sounds like either > > your VCR has some sync issues (or the tape itself is causing the VCR > > to have sync issues) or else the timing signals coming from the VCR > > during playback are outside the range that the DC10 can reliably lock > > onto. The digitizer chip on the DC10's is a bit particular about > > having video signals that are quite close to "proper" spec and most > > VCR's tend to be designed to play fast and loose with the sync > > because most TV's are far more tolerant of out of bounds video sync. > > Another possibility is that scott is experiencing the dreaded field reversal > effect, or a compounded effect caused by fields being recorded and played > back in slight temporal disorder.. > > My hardware has very experimental drivers, and I have still not worked out all > necessary flags needed for perfect grabbing. I suffer from the latter > phenomenon; my card grabs frames on one field boundary, tags them slightly > wrong, and then they get messed up in the replay process. > > Vertical shaking is one trade mark of field reversal. Such lav files can be > corrected with yuvcorrect or similar tools. > > Scott, have you tried replaying the grabbed file with lavplay and the -x flag?
Just tried with and without -x on the lstest.avi file at http://www.scottwaye.com/lstest.avi Concentrating on frames 4,5, 6 with glav -F (without -x first) and using the arrow keys to step back and forward: Frame 4: looks ok Frame 5: comb effect above titles with the "comb" on pixel higher than frame 4 Frame 6: looks ok but the picture is 2 pixels higher than frame 4 with -x Frame 4: comb effect below the titles Frame 5: ok, but is higher than frame 4 Frame 6: comb effect below the titles and again is 2 pixels higher than frame 4 So concentrating on the first (without -x) I could get frames 4 and 6 steady by shifting frame 6 down 2 pixels - the code in y4mshift looks like it could be modified to do this based on the luminence of the top few lines fairly easily (wouldn't work if the background at the top of the frame was black too though). Frame 5 is a different problem however - looks like the fields need exchanging then the frame shifted down 1 pixel. ------------------------------------------------------- This SF.Net email is sponsored by: Sybase ASE Linux Express Edition - download now for FREE LinuxWorld Reader's Choice Award Winner for best database on Linux. http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click _______________________________________________ Mjpeg-users mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/mjpeg-users
