Learn FFmpeg - Muxing, Demuxing & Containers

#Introduction

In episode 7 you learned to inspect file contents with ffprobe — now you can see that a video file is actually made up of many streams packed into one container. This episode dissects that container itself, and the two operations revolving around it — muxing (combining streams) and demuxing (splitting them).

This understanding matters because most "video conversions" people do aren't really conversions at all — just changing the envelope. Realizing this will save CPU time and preserve quality.

#Container: The Envelope for Codecs

Separate two things that are often confused: codec and container. A codec is how the data is compressed — e.g., H.264, AAC, VP9. A container is the wrapper that unites those codec streams into a single playable file — complete with metadata, subtitles, and chapters.

The easiest analogy: a codec is the language, a container is the envelope. An envelope can hold a letter in English or Indonesian; an MP4 container can hold H.264 video and AAC audio. But not every language fits a given envelope — and not every codec fits every container.

Mux (multiplex) means combining several streams into one container; demux means unpacking them. FFmpeg does both in one command — input is demuxed, output is muxed.

#Popular Container Formats

A rule you must remember: WebM only accepts VP8/VP9/AV1 + Opus/Vorbis. You can't force H.264 or AAC into WebM — the container rejects them. MKV, on the other hand, accepts almost every codec, which is why MKV is the favorite for archiving and ripping.

#Remux Without Re-encoding: -c copy

Most people run ffmpeg -i a.mkv b.mp4 and think it's a "conversion". In fact it's a remux: moving streams from one envelope to another without touching their content. The key is -c copy — copy the bitstream as-is, without decoding and re-encoding.

ffmpeg -i input.mkv -c copy output.mp4

This command is very fast — can be dozens of times faster than re-encoding — and more importantly: quality doesn't change at all because the data is never unpacked.

Why does this matter? Because lossy compression is cumulative. Every time you decode then re-encode H.264, quality drops a little — that's what's called generation loss. -c copy avoids an additional generation entirely. If the goal is just changing containers, never re-encode.

#Selecting Streams with -map

Without -map, FFmpeg selects streams based on default heuristics: one best video, one best audio, one subtitle. For full control, we use -map with a stream address in the file:type:index format.

The most important patterns to master:

• 0:v:0 — first input file, video stream, video index 0
• 0:a:1 — first input file, audio stream, audio index 1
• 1:s:0 — second input file, subtitle stream, subtitle index 0

ffmpeg -i input.mkv -map 0:v:0 -map 0:a:0 -c copy output.mp4

With the command above, you map exactly one video and one audio to the output — nothing more. This is the basic pattern that will become the backbone of all later muxing operations.

#Adding & Removing Streams

With -map, adding and removing streams are two sides of the same coin: what isn't mapped doesn't go to the output.

ffmpeg -i input.mp4 -map 0:v:0 -c copy silent.mp4

The command above maps only video — all audio and subtitles are dropped. The result is a silent file without re-encoding. To add a stream, just feed in a second input and map it:

ffmpeg -i video.mp4 -i music.mp3 -map 0:v:0 -map 1:a:0 -c:v copy -c:a aac output.mp4

Here video is copied (-c:v copy) and the song from the second file is encoded to AAC (-c:a aac). The -map 0:v:0 -map 1:a:0 pattern is the universal answer for "replace the soundtrack" or "add audio" without touching the video.

#Demuxing: Unpacking Streams

Demuxing — separating streams — is done by mapping only one stream and writing it to its own file. This technique is useful for extracting pure audio, subtitles, or raw video from a combined file.

ffmpeg -i input.mp4 -map 0:a:0 -c copy audio.m4a

The result is a separate audio file extracted from the MP4. Because of -c copy, this extraction is instant and doesn't change quality — exactly like taking a letter out of an envelope without opening it. In episode 11 you'll do the same for subtitles.

#Conclusion

In episode 8 you've understood the container's role as the envelope for codecs: the compatibility rules of MP4, MKV, and WebM, fast remuxing without re-encoding with -c copy, precise stream selection with -map, and how to add, remove, and separate streams without touching quality.

The key takeaway: the separation between codec and container is the key. Once you distinguish the two, most of FFmpeg's "weird" puzzles become clear — including why a file rejects a certain codec.

In the next episode 9 we scale up: processing many files at once with loops, parallel processing, and safe scripts — the skill that turns FFmpeg from a manual tool into a batch machine. See you in episode 9!