TikTok Shop wants motion: no slideshows or screenshots

A slideshow of product photos does not meet TikTok Shop's video guidance. Two seller-university pages say to avoid still-frame content (static images, screenshots, slideshows, screen recordings), and one sets a floor of 3 seconds of dynamic content per video with no still images or looping visuals. The fix is to generate real motion: Sume's video generation turns a prompt, and optionally reference images, into a clip.

What the two pages say

Read the Must include row carefully

The high-quality-video page lists items a video must include: real-time speech or sign language, the creator's face and multi-angle product demos. A generated clip with no person and no voice may not match that list, whatever its motion. Treat the page as the standard for a TikTok Shop video and check your clip against it.

For LIVEs the page prohibits AI-generated voices and pre-recorded audio outright, so generated clips do not fit a LIVE.

Generating motion with Sume

The Video Generation docs describe an asynchronous flow: submit to POST /v1/videos, poll, then download the content. Models differ in supported_durations, supported_aspect_ratios and supported_resolutions; read them from GET /v1/videos/models. Most catalog models top out at 15 seconds, which is well above TikTok's 3-second floor.

A prompt that describes a product turning, a hand picking it up or a close-up tilt gives motion across the clip. Review the output frame by frame, because a generation that mostly holds one pose is still a still frame in practice. Compare against the listing so the product stays accurate.

Quick screen

• Scrub the clip for long holds on one pose; they are still frames in practice.
• At least 3 seconds of real motion.
• No screenshots or screen recordings as the main picture.
• Check for a person, speech and multi-angle demos, as the page lists.