TikTok ad safe zone shrinks with caption length: place burned-in text

Yes, the TikTok ad safe zone moves, and the longer your ad caption is, the smaller it gets. TikTok's Ads Help says each key element, such as text and logos, has to sit inside the safe zone or it may be covered or cropped, and that the zone's size depends on the ad format (vertical, horizontal or square), on the caption length, and on any interactive add-on (read 2026-10-03). It does not publish one fixed pixel box on those pages, so the practical answer is to burn your own text into the part of the frame that stays clear in TikTok's preview, then check a frame before you ship.

Sume gives you the two tools for that: a caption job whose line position you can set per request, and a frame extractor to look at the result. Neither one knows TikTok's UI, so the safe zone itself still comes from TikTok's downloadable safe-zone files and its preview tool.

What TikTok actually says about the safe zone

Three statements matter. First, key elements must be inside the zone or they can be covered or cropped. Second, the zone varies by dimension, caption length and add-on. Third, TikTok recommends checking the preview tool for element placement and points to downloadable safe-zone files that differ by ad format and region (read 2026-10-03).

The caption rule feeds into the first two. On the auction in-feed page, a non-Spark ad caption shows at most four lines and is drawn in white with a uniform font, and the creative best-practices page tells advertisers to keep content visible within the UI safe zone (read 2026-10-03). A four-line caption takes room from the bottom of a vertical frame, so text you burned low in the picture is the first thing to collide with it.

Move burned-in captions with anchor_ratio

Sume's video captions endpoint burns captions onto a public HTTPS clip. Its design.placement.anchor_ratio field is the caption line's centre as a fraction of frame height, so a smaller number lifts the line toward the top and a larger one pushes it down. design is not supported on the punch and tiktok-green styles, so use slam (the Latin default) when you need to move the line.

The docs do not state a default ratio or a recommended range for TikTok, so treat the value in the sample below as a starting guess, not a rule. Pick it by comparing the output against TikTok's template, and burn a different ratio per ad if the on-platform caption lengths differ.

Check one still before you upload

After the caption job finishes, ask video frames for stills at a couple of instants where the caption is on screen. The endpoint returns durable image artifacts at the source size, so you can lay TikTok's safe-zone file over them in any image tool. Frames submit as 202, and each at value must be inside the clip's duration.

This script burns a caption at anchor_ratio 0.42 and pulls two stills. It needs SUME_API_KEY and a clip already on media.sume.com (import it first with POST /v1/media-imports).

import os, time, requests
BASE = "https://api.sume.com"
H = {"Authorization": f"Bearer {os.environ['SUME_API_KEY']}"}

def run(path, body, key):
    r = requests.post(BASE + path, headers={**H, "Idempotency-Key": key}, json=body)
    r.raise_for_status()
    job = r.json()["request_id"]
    while True:
        s = requests.get(f"{BASE}/v1/jobs/{job}/status", headers=H).json()["status"]
        if s in ("completed", "failed", "canceled"):
            break
        time.sleep(3)
    return s, (requests.get(f"{BASE}/v1/jobs/{job}/result", headers=H).json() if s == "completed" else None)

clip = "https://media.sume.com/artifacts/artf_demo/ad.mp4"
status, res = run("/v1/video-captions", {
    "video_url": clip, "style": "slam",
    "design": {"placement": {"anchor_ratio": 0.42}}}, "safe-zone-cap-001")
print(status, res and res.get("video_url"))
if res:
    status, frames = run("/v1/video-frames",
        {"video_url": res["video_url"], "at": [1.5, 4]}, "safe-zone-frames-001")
    print(status, frames)

What Sume cannot do here

Sume does not draw TikTok's caption block, interactive add-ons or the app UI over your clip, so it cannot show you where they will land. A caption job costs $0.20 for videos up to 60 seconds under the current fixed estimate (confirm in GET /v1/catalog), so decide the placement from one test render rather than iterating on the full set.

If your ad later gets a longer caption in Ads Manager, the zone shrinks and a burned-in line that fit yesterday can collide today. Keep the burned-in text short, keep it out of the bottom band, and re-run the still check whenever the caption or add-on changes.