[FFmpeg-user] Variable Quality Encoding within frames

Jeff Douglas jeff at integratedmonitoring.net
Fri May 31 00:53:43 EEST 2019


My use case is to take a video source with area(s) of particular interest
and preserve detail in those macroblocks at the expense of less important
portions of the frame.

This has become quite common for surveillance cameras. Some use AI styled
detection (i.e. favoring a 'face' over 'trees blowing in the wind'). Others
implementations allow box(es) to be defined; equivalent to specifying a
base -crf and several overlay -crf(s) cropped by rectangle.

Is there any equivalent function currently available within the ffmpeg
filter/encoder framework?

Cheers, Jeff


More information about the ffmpeg-user mailing list