On 2016-01-20, at 16:01, Moritz Barsnick wrote: > Regarding cropdetect: > > On Wed, Jan 20, 2016 at 11:31:50 -0500, Etienne Desautels wrote: >> [Parsed_cropdetect_1 @ 0x40b8900] x1:719 x2:0 y1:479 y2:0 w:-718 h:-478 >> x:720 y:480 pts:6006 t:0.066733 crop=-718:-478:720:480 > > I have noticed dimensions like this before. Are they actually useful? > What do they produce if fed to the crop filter (e.g. via metadata)? > > I realize they represent "empty" frames, where "everything" is cropped. > Does something need fixing here? (I started an attempt a few months > ago, but got distracted by life.)
Personally, I think the most important thing regarding this is to document this fact. When you know it, it's easy to discard negative values and understand them. Also those negative values gives information (frame is mostly black). I don't know a good alternative if we want to keep this information (black frame) in another form? The negative values could be converted to zeros, but what's the advantage? About the crop filter, I don't know. Just my 2 cents. Etienne Desautels _______________________________________________ ffmpeg-user mailing list ffmpeg-user@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-user
