Vertical video subtitles: BBC's 3-line rule and Sume settings

The BBC's subtitle guidelines recommend a maximum of three lines for vertical 9:16 video, against two for 16:9, 4:3 and 1:1, with text allowed to fill up to 90% of the frame width and placed a little higher than the bottom, still in the lower third (read 2026-10-03). Sume's burned-in captions let you set the width and the vertical position per request with design.typography.safe_width_ratio and design.placement.anchor_ratio.

The BBC numbers describe authored subtitle files for its own platforms, so treat them as sound starting values and check a rendered frame.

What does the BBC say about vertical subtitles?

Version 1.2.5 of the guidelines (March 2026) states that for landscape or square video a maximum of two lines is recommended, and for vertical 9:16 video three lines, with extra lines acceptable only when no important picture information is covered. For online delivery the line width is 68% of the video width for 16:9 and 90% for 4:3, 1:1 and 9:16. As a guide, the 37 characters allowed in a 75% region on 16:9 equal about 25 characters in a 90% region on 9:16.

On position, the BBC says it is common in vertical videos to place subtitles a little higher, though still generally in the lower third, because faces are generally in the top half of the screen and bottom placement makes it harder to read the text and see the person speaking.

How do I set width and height in Sume?

safe_width_ratio is the fraction of frame width a caption line may occupy, so 0.9 mirrors the BBC's 90%. anchor_ratio is the caption line centre as a fraction of frame height for portrait frames; the BBC's three-line example region is 24% of the frame high (three lines at an 8% line height). Centring that region at 0.78 puts it between roughly 0.66 and 0.90 of the height, which is the lower third with room under it for platform UI. That 0.78 is arithmetic on the BBC numbers, not a BBC figure, and platform overlays differ, so confirm against the app you post to.

Both fields merge over the style's own value, out-of-range numbers return a 400 at request time, and design is not supported on the punch and tiktok-green styles (video captions). The request below uses slam.

curl -X POST https://api.sume.com/v1/video-captions \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: vertical-caption-bbc-001" \
  -d '{
  "video_url": "https://media.sume.com/artifacts/artf_demo/vertical.mp4",
  "style": "slam",
  "language": "en",
  "design": {
    "typography": {
      "safe_width_ratio": 0.9
    },
    "placement": {
      "anchor_ratio": 0.78
    },
    "phrasing": {
      "max_words": 5,
      "max_chars": 40
    }
  }
}'

Why is there no line-count setting?

Sume groups words into short on-screen phrases, controlled by the phrasing fields, rather than laying out a fixed block of subtitle lines. The controls you do have are max_words (1 to 12), max_chars (4 to 60, characters per on-screen phrase) and pause_seconds, the silence that breaks one phrase into the next. If a phrase wraps to more lines than you like, lower max_words or max_chars and render again. For platform-specific placement, see the YouTube Shorts safe zone post; for size and outline, see the typography fields.

What should I check on the rendered clip?

Burned-in text cannot be switched off or resized by the viewer's player, so a bad position stays bad everywhere the clip is reposted. Look at one frame before you batch.

• Faces and mouths stay clear of the caption block, which is the BBC's main reason for lifting the text.
• Any on-screen text from the source video, such as names or prices, is not covered.
• The longest phrase in the clip still fits within the width you set, without shrinking to an unreadable size.
• The text is readable at phone size: the BBC notes that on small phones the authored size should be shown unscaled, so do not rely on a player shrinking or growing it.
• Re-style without paying for a second transcription by passing source_caption_id; see the restyle walkthrough.

What does it cost and where are the limits?

A standalone caption job is $0.20 per job for videos up to 60 seconds under the current fixed estimate, so confirm live pricing in the catalog. A restyle is still a render and is billed as one. The caption endpoint needs audible speech; silent clips fail with caption_no_speech, and you pass cues with text and times instead. If your clip is longer than 60 seconds, cut it first with video trim.