Add a watermark to a video with FFmpeg

ffmpeg -i input.mp4 -i logo.png -filter_complex "[1:v]scale=240:-1[wm];[0:v][wm]overlay=W-w-24:H-h-24" -c:v libx264 -crf 23 -preset medium -c:a copy watermarked.mp4

Scales the logo to 240 px wide, keeps its alpha, and draws it 24 px from the right and bottom edges on every frame. W and H are the video size, w and h the logo size. The bottom-right corner of the 1 s frame dropped from an average luma of 147 to 107: the logo is there. Audio is copied.

Flags, explained

-i logo.png

The watermark is a second input. PNG keeps transparency; a JPEG is opaque unless you cut alpha with format and colorchannelmixer.

[1:v]scale=240:-1[wm]

Scale the second input to 240 px wide with proportional height and label the result wm. -1 is fine here because the overlay is not encoded on its own.

overlay=W-w-24:H-h-24

Place the top-left corner of the overlay at x and y. W,H are the main video dimensions, w,h the overlay dimensions, so this is the bottom-right corner with a 24 px margin.

colorchannelmixer=aa=0.5

Multiply the alpha channel by 0.5, i.e. 50 % opacity for the whole overlay.

enable='between(t,2,8)'

Apply the filter only while the expression is true; t is the timestamp in seconds.

drawtext=text=...:x=w-tw-24:y=h-th-24

Render text with libfreetype; tw and th are the text width and height, w and h the frame size.

Variants

ffmpeg -i input.mp4 -i logo.png -filter_complex "[0:v][1:v]overlay=24:24" -c:v libx264 -crf 23 -preset medium -c:a copy watermarked.mp4

ffmpeg -i input.mp4 -i logo.png -filter_complex "[1:v]scale=240:-1,format=rgba,colorchannelmixer=aa=0.5[wm];[0:v][wm]overlay=(W-w)/2:(H-h)/2" -c:v libx264 -crf 23 -preset medium -c:a copy watermarked.mp4

ffmpeg -i input.mp4 -i logo.png -filter_complex "[1:v]scale=240:-1[wm];[0:v][wm]overlay=W-w-24:H-h-24:enable='between(t,2,8)'" -c:v libx264 -crf 23 -preset medium -c:a copy watermarked.mp4

ffmpeg -i input.mp4 -vf "drawtext=text='kinopipe.com':fontsize=36:fontcolor=white@0.8:x=w-tw-24:y=h-th-24" -c:v libx264 -crf 23 -preset medium -c:a copy watermarked.mp4

Pitfalls we hit

• A watermark is a re-encode: -c:v copy is impossible because the pixels change. Budget the encode time and keep -crf at 18 to 23 to avoid degrading the source.
• Logos scaled with -1 can get an odd height; that is fine for the overlay but breaks if you ever encode the scaled logo alone. Use -2 when in doubt.
• The overlay is positioned in source pixels. A 240 px logo looks right on 1080p and huge on a 480p clip; derive the size from the frame with scale=iw*0.2:-1 on the main input width, or compute it from ffprobe first.
• drawtext depends on libfreetype and a font. Builds without it say "Filter not found", and a missing font falls back to whatever fontconfig picks, which may not be what you expect.

The same edit as a typed operation

add_watermark_to_video takes a named position, a width in pixels and an opacity, then builds the scale, alpha and overlay graph itself and renders it with the rest of the recipe in one pass. Margins and alpha handling are validated before the encode starts.

{
  "inputs": [
    { "id": "main", "url": "https://example.com/video.mp4" },
    { "id": "watermark", "url": "https://example.com/logo.png" }
  ],
  "options": { "position": "bottom-right", "width": 240, "opacity": 0.85 }
}

FAQ