TikTok ad static image rule: 50% of the video, and AI slideshows

No. TikTok's Ad Format and Functionality policy says ad content must be dynamic, and that static images "should not occupy more than 50%" of the video. An AI slideshow built from stills with a voiceover is the pattern that rule is aimed at (read 2026-10-03).

The same page asks for audio that is not unclear or muffled and for visuals that are legible. This post lists what the page says, how to check a clip against it, and where Sume's tools stop: Sume can show you stills and tell you whether a track exists, but it does not measure how much of a video is static.

What does the TikTok ad policy page say about still images and audio?

The table is a reading of TikTok's Ad Format and Functionality page, not a spec sheet for every placement. Placement-specific pages can differ, so confirm the one you buy.

Does an image-to-video clip count as static?

TikTok's page gives the 50% line for static images and one example of a violation, and it does not define a measuring method. Read it as a review standard applied by people and systems at TikTok, not a number your export can be checked against by script.

A short AI clip that animates a product image keeps the picture moving, so it differs from a slideshow where each image sits still for seconds. A clip that renders a still with only a zoom is the closer call. If more than half of the runtime reads as a held picture, assume it is at risk and add real motion or cut shorter.

Length also differs by page. The ad review FAQ I read on the same day lists a video length of 5 seconds to 10 minutes, while the format policy above says 60 seconds. When the two disagree, check the placement spec for the ad type you are buying, as in TikTok in-feed ad specs.

How do I check an AI ad clip before uploading it with Sume?

Video inspect reads one clip already on media.sume.com. It returns a probe, stills as image artifacts, and, if you ask, a transcript. The docs tell you to check probe.has_audio first, because a transcript request on a silent clip fails with inspect_source_has_no_audio.

For motion, ask for stills at a fixed rate. frames takes { "fps": n } with 0 < n <= 2, capped at 24 stills a call, so one still per second covers a clip up to 24 seconds. Flip through them: if the first frames in a row show the same picture, that stretch is static.

Silent clips have a second trap. The video captions docs say speech-to-captions needs audible speech and a silent clip fails as caption_no_speech; authored cues skip speech-to-text. That fixes the on-screen text but not the missing audio, which the TikTok page says an ad needs.

curl -X POST https://api.sume.com/v1/video-inspect \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: ad-motion-check-001" \
  -d '{
    "video_url": "https://media.sume.com/artifacts/artf_demo/ad.mp4",
    "frames": { "fps": 1 }
  }'

A pre-upload checklist for AI slideshow-style ads

Run this on every variant, since a batch shares its weak spots.

• Open the stills from one inspect call and mark any run of near-identical frames. Add up the seconds against total length.
• Confirm the clip has an audio track and listen for muffled or clipped sound.
• Check the length against the placement you picked, not just the 60-second figure.
• Make sure any required disclosure line is visible on screen. See preflight an AI UGC ad for TikTok.