On 8/24/07, Robert Tigelaar <[EMAIL PROTECTED]> wrote: > Allright, the test transcode of Pulp Fiction with -M2 was slighty better > then the without -M ones...in the end part, a/v desync was around 500-1000 > ms.
transcode has some known deficiencies in A/V sync field. Sadly, most of them are architectural, so fixing them isn't trivial nor quick. > I had looked trough the archives a bit already but couldnt find anything > related to the desyncs then...or so I thought. Handling desync is an hard issue per se. Transcode has some limited capabilities in handling the constant desync (A/V shift IIRC the exact naming), so if A/V desync is fixed and constant we can already handle it fairly. Things get real nasty in case of drifting (progressive, increasing A/V desync). transcode just can't do much on this case, yet. > - fixes in xvid export module; > encoder flush was implemented at time of closing module, and that > is supposed to fix issue #000017 ( > http://tcfoundry.hostme.it/mantis/view.php?id=17). > Please test and report problems or improvements :) > I've changed dvdrip/transcode sources to ~x86 (gentoo unstable) and am > recompiling them right now. When they're done, I'm going to test again and > I'll let you people know if there was any improvement. It will hardly be an improvement I guess, this fix regards an unrelared field. > By the way, I tried to transcode Children of the Korn 4 today, too. This > title has some *weird* a/v sync issues. The first 10 minutes, sync is ok, > then theres five minutes of heavy desync and then it's synced again...plain > weirdness all over. Definitively. And things get worse with NTSC (and it's insane framerate 30000/1001), and even worse with telecine. Sad but true, that's our situation right now. Bests, -- Francesco Romani // Ikitt
