ffmpeg | branch: master | Peter Cordes <[email protected]> | Wed Mar 4 03:35:36 2015 -0400| [3ad3529b567e04a812c0ff103855ecba3efff900] | committer: Michael Niedermayer
avfilter/vf_mpdecimate: add comment about 8x8 SAD Signed-off-by: Michael Niedermayer <[email protected]> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=3ad3529b567e04a812c0ff103855ecba3efff900 --- libavfilter/vf_mpdecimate.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/libavfilter/vf_mpdecimate.c b/libavfilter/vf_mpdecimate.c index cb3e69f..7cc3254 100644 --- a/libavfilter/vf_mpdecimate.c +++ b/libavfilter/vf_mpdecimate.c @@ -120,6 +120,11 @@ static int decimate_frame(AVFilterContext *ctx, return 0; for (plane = 0; ref->data[plane] && ref->linesize[plane]; plane++) { + /* use 8x8 SAD even on subsampled planes. The blocks won't match up with + * luma blocks, but hopefully nobody is depending on this to catch + * localized chroma changes that wouldn't exceed the thresholds when + * diluted by using what's effectively a larger block size. + */ int vsub = plane == 1 || plane == 2 ? decimate->vsub : 0; int hsub = plane == 1 || plane == 2 ? decimate->hsub : 0; if (diff_planes(ctx, _______________________________________________ ffmpeg-cvslog mailing list [email protected] http://ffmpeg.org/mailman/listinfo/ffmpeg-cvslog
