I had to trim some videos to a 30 second clip, and needed to do it in a way that kept the codec. Using straight ffmpeg, I couldn’t figure out how to get an mp4v2 output from an mp4v2 input. The file looks like this:

Test.mp4: ISO Media, MP4 v2 [ISO 14496-14]

Converting using ffmpeg changes the file to mp4 v1 though ☹.

ffmpeg -i Test.mp4 -vcodec copy -acodec copy -ss 0 -to 30 outputFile.mp4

Results in this filetype (running file outputFile.mp4):

outputFile.mp4: ISO Media, MP4 Base Media v1 [IS0 14496-12:2003]

So after the ffmpeg conversion, I used MP4Box -brand mp42 outputFile.mp4, and that seems to work. It doesn’t seem too lossy, but we’ll see.

I now need write a script to go through and convert a bunch of wmv files to these mp4v2’s. So far it looks like:
find . -iname '*.wmv'|while read filename; do ffmpeg -i "$filename" -strict -2 "$filename".mp4; done

And I’ll need to fix the filename correctly.

Here is my script. I think it’s pretty good: #!/bin/bash while read filename do name="${filename%.wmv}.mp4" if [ -f "$name" ]; then echo "File $name exists already sucka!" else (ffmpeg -i "$filename" -strict -2 "$name" && SAVEIFS=$IFS; IFS=$(echo -en "\n\b");MP4Box -brand mp42 $name;IFS=$SAVEIFS)& fi done < <(find . -iname '*.wmv') wait

This is even better using gnu parallel: find . -iname '*.wmv' | time parallel -j+0 --eta "ffmpeg -n -loglevel panic -hide_banner -i {} -strict -2 {.}.mp4"