Hi All,

I'm trying to grab still images from an AVCHD stream. It originated from a
panasonic tm350 camcorder (NTSC version) which uses 3:2 pulldown to shoehorn
24p video into 29.97fps.

Needless to say, 2 frames out of every 5 are combed, and I need to devise a
mechanism to deal with this.

It seems to me that in order to know which fields to rejoin into which
source (24p) frame, i need to know or compute where the start of each
4source/5destination frame blocks is.

I've done some googling, and so far have not found anything to suggest that
ffmpeg has any way of knowing a) that my video is hard telecined and b)
where the start of each block is.

Does anyone know if ffmpeg (in particular with the h264 codec, or the m2ts
steam code) has anything to know about hard telecining?
If not, can anyone describe an algorithm i might implement to deduce the
syncronisation of frames? I guess I need to spot frames that exhibit combing
between fields, because 2 sequential combed frames would always correspond
to frames 3 and 4 in a 5 frame 3:2 pulldown block. What's a good mechanism
for detecting combing (in effect, spotting interlacing).

thanks for any help

slip
_______________________________________________
libav-user mailing list
[email protected]
https://lists.mplayerhq.hu/mailman/listinfo/libav-user

Reply via email to