Team Cocoon wrote:After talking to Gilles here is an example of command line for quantising ,
-maxrate 2500k -b 2000k -qmin 3 -qmax 5 -bufsize 4096 -mbd 2 -flags +4mv+trell -aic 2 -cmp 2 -subcmp 2 -g 300
I'm trying to break this down according to the currently published ffmpeg documentation. So we've got:
-maxrate 2500k (maximum video bit-rate in bits/second)
-b 2000k (average video bitrate in bits/second)?
-qmin 3 (minimum quantizer scale, whatever that means)
-qmin 5 (maximum quantizer scale)
-bufsize 4096 (video buffer verifier buffer size (in bits), have no idea what this is about or how the size is determined)
-mbd 2 (Macro block decision = rate control, have no idea what this is either)
-flags (doesn't exist in ffmpeg documentation)
+4mv (currently '-4mv' in ffmpeg documentation, (Use four motion vector by macroblock (MPEG-4 only))
+trell (doesn't exist in ffmpeg documentation)
-aic 2 (Enable Advanced Intra-coding (h263+). ffmpeg doc does not list or explain available settings )
-cmp 2 (doesn't exist in ffmpeg documentation)
-subcmp 2 (doesn't exist in ffmpeg documentation)
-g 300 (gop size, set the Group of Pictures size (# of frames)) (I'm guessing this is the same as what DivX calls the "Max Keyframe Interval")