UGC-style AI video ad: model checklist for 9:16, audio, product photo
For a vertical UGC-style ad, check four things on the model: 9:16, audio, a first-frame or reference photo, and the clip length. Sume rows that pass.
A UGC-style vertical ad needs four things from the model: a 9:16 frame, sound, a way to keep your product photo exact, and enough seconds. On Sume, Seedance 2.5, Wan 3.0, MiniMax H3 and Gemini Omni Flash 1.1 cover all four; Kling 3 covers them without reference images; Grok Imagine lacks audio and an aspect ratio field.
The four checks
"First frame" is image_url on the Video Router or a frame_images entry on /v1/videos.
| Model id | 9:16 | Audio | Product photo | Max (s) |
|---|---|---|---|---|
| seedance-2.5 | yes | optional | first frame or references | 30 |
| wan-3.0 | yes | optional | first frame or references | 30 |
| minimax-h3 | yes | always on | first frame or references | 15 |
| gemini-omni-flash-1.1 | yes | always on | first frame or references | 10 |
| kling-3 | yes | optional | first frame only | 15 |
| grok-imagine-video-1.5 | follows the first frame | none | first frame required | 15 |
Keep the product exact
A first-frame image fixes how the product starts, but the model still animates it. For label and logo fidelity, keep camera moves small, avoid prompts that rotate the product and check the first and last frames of every take before it goes into an ad set. Reference images guide appearance without fixing the first frame.
Sound and compliance
If the model creates speech, transcribe it with a video inspect step and check the words before publishing. Platforms also label AI-made video, so keep the generation record for each clip.
Sources
Related posts
More in Use cases
- UK AI ad disclosure: no blanket rule, two CAP questions to ask
The ASA says the UK has no blanket rule to disclose AI in ads, and gives two questions. How to run them on a Sume video: stills, transcript, a human sign-off.
- Unity Ads video specs: 60 s, H.264 MP4, 100 MB cap
Unity's user acquisition docs set video at 60 seconds or less, H.264 in MP4, 16:9 or 9:16, 100 MB max, and an end card. What Sume's trim step can and cannot do.
- Upload AI-made sound to Freesound or Pixabay: tags and labels
Freesound allows generative AI sounds with a GenAI tag and the model named; Pixabay wants an AI-generated checkbox. How to record the model from a Sume job.
- Use a trending TikTok as a reference for an original YouTube Short
Study a trend without re-uploading it. Sume's trending search returns TikTok metadata, not video files, which fits YouTube's Oct 2026 originality push.
Written by Sume