Draw a box on a video with the drawbox filter API

Burn a review box onto a clip with the drawbox filter in Sume's video filter. drawtext is not allowed, so labels need captions. Check first, free.

4 min readSume
All posts

Use the drawbox filter in POST /v1/video-filter: it is on the allowlist, and the encode returns a new MP4 with the box burned in. Text is different: drawtext and subtitles are not allowed, so a label on the box needs another route such as captions.

Runway's changelog (Comments on Assets, dated Aug 7, 2026, read 2026-10-01) describes pinning comments to an exact spot or moment in a video. That is a feature inside Runway. Sume has no comment layer; the closest thing is a mark baked into a copy of the clip.

What can drawbox and its neighbors do here?

Compositing filters such as drawbox, drawgrid and overlay are on the list; the docs name only a few exclusions.

Marking options in the video filter, from the docs and code read 2026-10-01.
NeedFilterAllowed?
A rectangle outline or filldrawboxYes
A grid for alignmentdrawgridYes
Text burned indrawtextNo
Subtitle filesubtitlesNo

How do I draw a box?

Put the rectangle in filtergraph. The filter takes ffmpeg's own arguments, so a red outline looks like the example below. The box is on every frame; this filter does not appear in the docs as limited to a time range, so a mark for one moment means cutting the clip first with video trim.

curl -X POST https://api.sume.com/v1/video-filter/check \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "video_url": "https://media.sume.com/artifacts/artf_demo/cut.mp4",
    "filtergraph": "drawbox=x=100:y=80:w=320:h=180:color=red:t=4"
  }'

How do I find the spot to mark?

Pull a still at the timestamp with video frames, whose program takes at[] times, then read the coordinates off the image. The docs say stills come back at source size.

What does it cost?

The docs list $0.02 per encode job and say the check is free. The source must be a workspace media.sume.com clip of 300 seconds or less. For hiding an area instead of marking it, see blur part of a video.

Sources

Related posts

More in Use cases

All Use cases posts

Written by Sume