Am 17.05.2010 13:08, schrieb Stefano Sabatini:
On date Sunday 2010-05-16 22:22:00 +0200, Stephen Hunt encoded:
Hi All,
Thinking of using ffmpeg libraries to write a program that attempts
to pinpoint editing cuts in a video clip. Would import this
information into an editing program to enable easier study of how a
film is edited.
Anyone have any pointers on how to approach this please? My C
language skills are quite rusty (wrote two programs, 10k lines each
in Uni, early 1990s). But very keen to get something like this
working. Also, has anyone perhaps done this already?
It should be relatively easy to port some of the existing code which
already do this to a libavfilter filter. I don't have time for this
right now, but you may file a feature request on roundup *or* try to
add it to FFmpeg yourself, I'll be glad to help in reviewing the patch
for the latter case.
Regards.
I've implemented a such algorithm. It basically substracts the values of
two images and "looks" whats left.
If the two pictures were from the same scene, this final image should be
near black. You have to do some testing to get a good threshold for this
black image. This will also fail if you have a scene with fast movement
(i.e. massive camera shake). But it worked with the mose videos I've
tested so far.
Best regards
Hendrik
_______________________________________________
libav-user mailing list
[email protected]
https://lists.mplayerhq.hu/mailman/listinfo/libav-user
