Hi Roman > > The very noticeable degradation comes about due to a bug in libdv which is > > used to do the encoding to DV from cinelerra. When I last looked into this > > about 2 years ago ffmpeg's DV codec wasn't as badly affected but it still > > wasn't great. AFAIK libdv is in deep maintenance mode now and I haven't > > seen any evidence that there's any work going on to find and fix the DV > > encoding bug. As for ffmpeg I haven't tested the DV encoding functionality > > of recent versions so I don't know what its status is. > > Since I'm the original author of the ffmpeg's DV codec -- I'd be very > curious to find out concrete evidence of why do you think encoder > produces inferior quality streams (and compared to what) since I am > in a position to fix this and improve it. > > Also, I remember that Dan Maas (CC'ed) once contributed a patch to > the ffmpeg's DV encoder that in his words made it comparable with > the Apple's DV encoder (which is supposed to be quite good based > on a number of reviews). > > I guess my point is simple enough -- don't assume anything based > on *releases* of ffmpeg -- try the latest SVN and decide for yourself.
I wasn't using an ffmpeg release; the tests I did at the time *were* based on current development snapshots. However, we are talking about tests which were done well over a year ago (which I clearly mentioned) and as I stated in my initial post I haven't revisited this issue since. This is because I've altered my workflow so I no longer require the use of DV as an intermediatory format. I actually have a vague recollection that you might have been in on the discussion about this at the time. I've just checked; I raised the issue in early April 2005 on the mjpeg-users mailing list. It then moved to the libdv-dev (subject "Possible bug encoding interlaced material?") which is where you came in. In essence ffmpeg's output was clearly better than libdv and for "real world" scenes did not appear to generate significant artifacts. However on CGI content, non-trivial artifacts did seem to be generated. I don't believe we came to any firm conclusion about this at the time. Note that the web location for the examples referred to in these old threads no longer exists. I could dig the files up again but it's proabably not applicable anymore because ffmpeg has moved on since those snapshots from 18 months ago. It would be best to repeat the tests on a new snapshot to see where we're at now, but I simply don't have the time for this at present and since I no longer need the DV encoder's output it's fairly low on my list of priorities. I am certainly happy to lend assistance where I can though. Regards jonathan ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ Mjpeg-users mailing list Mjpeg-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mjpeg-users
