ffmpeg is FLOSS (no cost), plus is the best tool to: transcode, trim, demux or mux. ffmpeg is on Windows&Linux&Android OS&iOS&OSX.

A few example use howtos

[This post allows all uses.] For Table of Contents, view on GitHub.

[Version of post is +posts/TranscodeMuxHowto.md#TableofContents · SwuduSusuwu/SubStack@904799e]

Tools compatible with this howto:

• For Android OS: FFmpeg Media Encoder - Apps on Google (SilentLexx) with much success; has a visual interface to ffmpeg but can process commands through console.

  • [Alternative (have not used): FFmpeg - Apps on Google (FFmpeg from CrossPlat).]

• For misc Linux OSs (suchas Ubuntu or Android OS +Termux): apt install ffmpeg.

• For iOS: https://shaunhevey.com/posts/how-to-use-ffmpeg-on-ios/.

• For OSX: brew install ffmpeg --with-fdk-aac --with-ffplay --with-freetype --with-libass --with-libvorbis --with-libvpx --with-opus (or alternatives).

• For Windows: FFmpeg - Official app in the Microsoft Store.

Howto

[Notice: versus stock Android, have moved default paths /Music/ to /Sounds/, /Movies/ to /Visuals/.]

[Notice: if /storage/emulated/0/ (directory root) is not found, replace with /sdcard/.]

[Notice: Can use examples with FFmpeg Media Encoder or Termux as-is (use absolute paths).]

Example visuals.mp4 was 4gb, to compress to 224mb used:

ffmpeg -i "/storage/emulated/0/Visuals/screen-20240629-045526.mp4" -framerate 30 -c:v libx264 -crf 32 -preset slower "/storage/emulated/0/Visuals/visuals.mp4"

. The libx264 codec compresses visuals best. -preset slower instructs it to compress more. You can replace -crf 32 with -b:v 2m to set an exact goal of “compress to 2mbps”.

[Notice: On some devices, Android OS’s permissions require to output to /storage/emulated/0/Download/]

Suppose you want to mux sounds.mp4 with visuals.mp4,

but you want to skip sounds.mp4’s 4 second intro, plus limit output to 2 minutes:

To demux sounds, pass -ss 4 to skip 4 seconds, pass -t 2:00 to output 2 minutes, pass -map 0:a:0 (zero-indexed) to demux first input as sounds, pass -c copy for instant process, output as .m4a:

ffmpeg -i "/storage/emulated/0/Download/sounds.mp4 -ss 4 -t 2:00 -map 0:a:0 -c copy "/storage/emulated/0/Sounds/demux.m4a"

Now sounds.m4a is 2 minutes, but visuals.m4a is much longer; pass -stream_loop -1 to mux sounds as loop to match visuals.mp4:

ffmpeg -i "/storage/emulated/0/Visuals/visuals.mp4" -stream_loop -1 -i "/storage/emulated/0/Sounds/demux.m4a" -map 0:v:0 -c copy -map 1:a:0 -shortest "/storage/emulated/0/Visuals/mux.mp4" 

Suppose you want the mix the sounds from visuals.mp4 with the loop from sounds.mp4:

ffmpeg -i "/storage/emulated/0/Visuals/visuals.mp4" -stream_loop -1 -i "/storage/emulated/0/Sounds/demux.m4a" -map 0:a:0 -map 1:a:0 -filter_complex amix=inputs=2:duration=shortest "/storage/emulated/0/Sounds/demux2.m4a"
ffmpeg -i "/storage/emulated/0/Visuals/visuals.mp4" -i "/storage/emulated/0/Sounds/demux2.m4a" -map 0:v:0 -c copy -map 1:a:0 -shortest "/storage/emulated/0/Visuals/mux2.mp4"

[Notice: -c copy is not compatible with -filter_complex; unless you want to reincode the visuals (slow), is 2 steps to do this]

Suppose you wish to produce a 10fps HD .gif from the first 24 seconds of visual.mp4:

ffmpeg -i "/storage/emulated/0/Visuals/visual.mp4" -map 0:v:0 -pix_fmt rgb24 -r 10 -s 1920x1080 -t 24 "/storage/emulated/0/Visuals/visual.gif"

External resources

Lists of commands&options which ffmpeg can use:

• ffmpeg Documentation

• complete list of ffmpeg flags / commands

How to use extra tools (which ffmpeg's GPLv2 version has): https://github.com/FFmpeg/FFmpeg/blob/master/LICENSE.md

visual.gif syntax was used to produce:

• https://www.deviantart.com/swudususuwu/art/BUD-Robos-hold-you-loop-v0-4-3-2-S-Simulator-1004148092

• https://www.deviantart.com/swudususuwu/art/Sakura-School-Simulator-howto-use-robot-props-loop-1019774750

mux.mp4/mux2.mp4 syntax was used to produce:

Video Transcoder - Apps on Google (a visual interface to ffmpeg) was cool versus most “video editor” apps -- but is not available for new versions of Android OS, can not loop (just has trim + convert (which can act as demux) + resize + compress), is slow (can not pass -c copy to ffmpeg, thus always reincodes inputs.)