[FFmpeg-user] Treating video frame-based instead of time-based

[Wolfgang Hugemann]

Is ist possible to treat video frame-based instead of time-based,
analogously to a frameserver such as AVIsynth?

Concretely:
* Could I (easily) extract every nth frame from a video (in order to
dump the result as JPEGs)?
* Could I subtract each frame from its predecessor (in order to detect
minimal motion)?

At the moment, I mostly perform such tasks in AVIsynth, using the AVS
file as input for ffmpeg. Which has the shortcoming that I can only use
AVI containers as input for AVIsynth (or use the ffmpegsource to read
the original video container into AVIsynth).

So: Can I somehow get rid of AVIsynth in the mid?

Wolfgang Hugemann