Here are the avconv settings I like to use when converting 1080i HD movies from my camera:
|
1 |
-crf 21 -c:v libx264 -preset medium -maxrate 2000k -bufsize 1835k -c:a libfaac -b:a 128k |
I find this results in a video which is only very inferior from the one taken from the camera, but is generally about 1/8 – 1/10th the file size, which is much more appropriate for long term storage.
The downside is that on my humble core i3 processor, it takes about 3-4x the length of the video to do the conversion, i.e. a 5 minute clip would take 15-20 mins to encode. For the times when I can’t be bothered to wait this long, or crystal clear video is not as important I use the following, which gives a perfectly acceptable quality video, and encodes roughly in real-time – i.e. a 5 minute clip takes 5 mins:
|
1 |
-crf 21 -c:v libx264 -preset veryfast -maxrate 2000k -bufsize 1835k -c:a libfaac -b:a 128k |
The resultant file size is the same as it would have been using the slower options above.
Rotating
Quite often, videos are recorded ‘portrait’ rather than ‘landscape’. Rather than craning your neck 90° to watch the video, you can add the following to the options to rotate the video as part of the encoding process:
Rotate clockwise: -vf transpose=1
Rotate anticlockwise: -vf transpose=2