[FFmpeg-user] FFMpeg problem using fieldmatch and decimate filters to IVTC

Nicholas Robbins nickrobbins at yahoo.com
Wed May 7 22:19:25 CEST 2014


> Nick,
>
> I can tell that there is some judder. It bothers me a little bit, but
> not enough to really get too bent out of shape about. I'm happy that
> everything is still in sync. Does the 60 fps conversion make a huge
> difference? I assume it'd go from: ABCDD EFGHH @ 30fps ... to AABBCCDD
> EEFFGGHH @ 60fps? Hmm, it might be worth it then to do a test-run and
> see if I could tell a difference at all.
>
>
> Steve

Ideally you go to 120fps. Then the parts that were telecined (orig 24fps) would be 
AAAAABBBBBCCCCCDDDDDD.....

The parts that were shot in interlaced NTSC (60 fields/s) would be
AAaaBBbbCCccDDddEEee..

Any parts that were shot in progressive 30fps (unlikely but possible for FX shots) would be
AAAABBBBCCCCDDDDEEEE....

This is b/c 120 is the least common multiple of 24, 30 & 60. If you had PAL mixed in (25fps) you would need to use 300 fps. 

I use 60fps b/c my STB can't handle 120fps. The stuff that was originally film is now AABBBCCDDD... (2,3,2,3,..) I don't notice it, where I did notice the ABCDD (1,1,1,2,...) 

You will have to add the dejudder filter if you do this, b/c otherwise you will get AABBCCDDDD (2,2,2,4) which isn't any better.

On a side note, my new headache is some cartoon from the 90's that were drawn at 12 fps, doubled to 24 fps and then telecined. I am using -vf fieldmatch,decimate and getting ok results.

Nick


More information about the ffmpeg-user mailing list