improvement to results. inter-field comparison is now counterbalanced
with intra-field total (vertical) variation. this means that areas of
extreme high frequency content, which become aliased within individual
fields, will not interfere with field matching. examples: white noise
effects, small kanji, very small latin text, ...
may still need tweaking. please report regressions. this change will
likely be made optional in the future (right now it's enclosed in
"if (1)"...
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@15722 b3059339-0415-0410-9bf9-f77b7e298cf2
simplify logic
eliminate stupid alternative affinity calculations (gave bad results)
favor output of clean duration-3 over duration-2 plus broken-1
(will give a more steady 3:2 pattern during telecine, w/ no quality loss)
options to adjust strictness of tests (but no way to set them presently :)
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@11642 b3059339-0415-0410-9bf9-f77b7e298cf2
frames in output. this will make it easier for the caller to do timing
or framerate regulation.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@11629 b3059339-0415-0410-9bf9-f77b7e298cf2
core (one of the chroma planes could be used for much lower cpu load
at the expense of some accuracy) and turning off verbose mode.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@10957 b3059339-0415-0410-9bf9-f77b7e298cf2
incorrect field matching) and greatly improve selection logic. the
pullup core should be very accurate now, so try throwing tough samples
at it and report any failures! :)
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@10731 b3059339-0415-0410-9bf9-f77b7e298cf2
(pullup.[ch]) is not mplayer-specific and is designed to work well
with g2; vf_pullup.c is the g1 wrapper. see man page for details, and
keep in mind, this is a work in progress.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@10665 b3059339-0415-0410-9bf9-f77b7e298cf2
rfelker
rathann
alex
aurel