TikTok safe zone: no single number, so test a caption anchor ratio

There is no single TikTok safe-zone number I can give you from TikTok itself. On 2026-10-05 I could not retrieve a TikTok help page that states one, and third-party guides place the zones differently. What is certain is the mechanism: TikTok overlays its own interface (username, caption, buttons) on top of a 9:16 frame, so burned-in captions must stay clear of those areas. In Sume you control the vertical position of a caption line with design.placement.anchor_ratio, then check the result on a real still.

The control Sume gives you

Video captions takes a design object that overrides one style for one request. placement.anchor_ratio is the center of the caption line as a fraction of the frame height: 0 is the top edge, 1 is the bottom. Other placement fields, plus typography.safe_width_ratio, control how far the text may extend sideways. Two styles, punch and tiktok-green, do not support design, so use slam (Latin text) or a Hangul style when you need to move the line.

A test loop

Do not trust a number from a guide for your own frames; trust a still. Burn captions with a candidate ratio, pull a full-size still with video frames, and lay the platform's overlay over it in your editor. The check costs one caption job ($0.20 fixed estimate for clips of at most 60 seconds, per the docs) plus one frames job; the loop converges in two or three tries.

Import the clip first with POST /v1/media-imports, and send an Idempotency-Key on writes.

curl -X POST https://api.sume.com/v1/video-captions \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: caption-anchor-060" \
  -d '{
    "video_url": "https://media.sume.com/artifacts/artf_demo/short.mp4",
    "style": "slam",
    "design": {"placement": {"anchor_ratio": 0.6}}
  }'

Rules that save a retry

• Out-of-range design values return a 400 at request time, so a bad number never costs a render.
• Restyle without a second transcription by sending source_caption_id instead of video_url.
• Korean text on slam is refused with caption_hangul_text_latin_style; use a Hangul style.
• Third-party guides, such as Reap, cover length and size, not exact safe areas.

Width as well as height

Zones are not only vertical. typography.safe_width_ratio limits how much of the frame width a caption line may use. As an example only, a ratio of 0.8 on a 1080-pixel-wide frame is 864 pixels of text width, leaving 108 pixels of margin on each side; 0.9 would give 972 pixels and 54 pixels of margin. Pick a value, burn a test, and look at a still, because the right number depends on the font and the words.

Whichever numbers you choose, record them with the video. If you later restyle the same clip, send source_caption_id with a new design block; the price is the same because a restyle is still a render, and the transcript is not run again.

Where to find the real zones

Platform-owned zone diagrams exist, usually as downloadable templates in the platform's creator or ads help. Use those as the overlay in your test. A diagram you download from TikTok is a vendor source; a number copied from a blog is not. The Sume side of the workflow is only to move the text where the diagram says it is safe.

Keep a record of the ratio you settled on. A team that uses one anchor ratio and width for every clip gets a consistent look, and a restyle through source_caption_id is the cheap way to apply a new ratio to an old clip, since the transcript is not run again.