Merge videos with FFmpeg

ffmpeg -f concat -safe 0 -i list.txt -c copy merged.mp4

Reads list.txt (one line per clip: file 'clip1.mp4'), appends the packets of each file and copies them without decoding. A 12 s clip followed by a 4.5 s clip became a 16.52 s file with 495 frames, in well under a second.

Flags, explained

-f concat

Use the concat demuxer, which reads a text list of files instead of a media file.

-safe 0

Allow absolute paths and paths with special characters in the list. Without it, anything outside the current directory is rejected as unsafe.

file 'clip.mp4'

One line per clip in list.txt, in playback order. Single quotes around the path; escape a quote inside it as '\''.

-c copy

Copy every stream. Requires identical codecs, dimensions, frame rate, pixel format, sample rate and channel layout across all clips.

concat=n=2:v=1:a=1

The concat filter: n inputs, each providing one video and one audio stream, in the order of the labels before it.

xfade=transition=fade:duration=1:offset=11

Dissolve of 1 s starting 11 s into the first input. Other transitions: wipeleft, slideup, circleopen and about 50 more.

acrossfade=d=1

The audio counterpart: a one-second crossfade between the two audio streams.

Variants

ffmpeg -i first.mp4 -i second.mp4 -filter_complex "[1:v]scale=1280:720:force_original_aspect_ratio=decrease,pad=1280:720:(ow-iw)/2:(oh-ih)/2,fps=30,setsar=1[v1];[1:a]aresample=48000[a1];[0:v][0:a][v1][a1]concat=n=2:v=1:a=1[v][a]" -map "[v]" -map "[a]" -c:v libx264 -crf 23 -preset medium -c:a aac -b:a 128k merged.mp4

ffmpeg -i first.mp4 -i second.mp4 -filter_complex "[0:v][1:v]xfade=transition=fade:duration=1:offset=11[v];[0:a][1:a]acrossfade=d=1[a]" -map "[v]" -map "[a]" -c:v libx264 -crf 23 -preset medium -c:a aac -b:a 128k merged.mp4

Pitfalls we hit

• The concat demuxer does not check that clips match. Joining a 1280x720 30 fps clip with an 854x480 25 fps clip with -c copy produced a file that reported 15.64 s and threw 6 decoder errors on playback. Different sources go through the concat filter.
• A relative path in list.txt is resolved from the directory of the list file, not from where you run ffmpeg.
• Timestamps must be continuous. Clips cut with -c copy that do not start on a keyframe can leave a gap or a frozen frame at each join; re-encode those clips before joining them.
• xfade needs the offset in seconds from the start of the first clip. A wrong offset silently shortens or lengthens the output instead of failing.

The same edit as a typed operation

merge_videos probes every input, normalises size, frame rate and audio to one target, then renders the join in a single pass, with an optional crossfade whose offsets are computed from the probed durations. Mismatched clips are the normal case, not an error.

{
  "inputs": [
    { "id": "main", "url": "https://example.com/intro.mp4" },
    { "id": "clip-2", "url": "https://example.com/demo.mp4" }
  ],
  "options": { "aspect_ratio": "16:9" }
}

FAQ