Which entries publish one value, not a ladder
Most entries publish a ladder of resolutions because each step is priced. Three publish one value: 1280x720, 720 by 480 with others refused, and 1216x704. Each describes a delivery format rather than a configurable output. As of 2026-09-22.
| Item | What is published | Detail |
|---|---|---|
| 1280x720 | A user guide | Stated with a single frame rate, 24 fps |
| 720 x 480 | A model card | Other resolutions are stated as unsupported |
| 1216x704 | A model card | With a rule that resolutions divide by 32 |
Inclusion rule. Entries whose own page states a single output resolution with no second value offered. A page that names one value and mentions others elsewhere is not counted. Order. As the pages were read: guide first, then the two cards.
1A fixed size removes a decision and a lever
With a ladder, exploration happens at the bottom and delivery at the top, and the difference is paid once. With one value, exploration costs what delivery costs, which changes what a campaign of forty variants adds up to.
It also simplifies delivery. Every request returns the same kind of file, so an upload check is answered once for the whole campaign rather than per tier, and nothing drifts between pieces.
2Two of the three are open-weight releases
The pattern is not a coincidence. A model card describes what released weights produce; a product page describes what can be bought. Tiers exist to be charged for, and where nothing is charged the underlying fixed behaviour surfaces instead.
The third is a vendor user guide, and it reads the same way: one resolution, one frame rate, a length that moves in increments. A fixed format offered as a product rather than as a research artefact.
3One of them fails a published floor
The smallest of the three is 720 by 480, and the surface in this register that publishes minimum dimensions accepts vertical from 540x960, horizontal from 960x540 and square from 640x640. That frame clears none of the three.
Everywhere else in this register the pixel minimums are cleared at the cheapest rung and stop mattering. This is the one row where they bite, and the card rules out buying a way past them.
- Video resolution1280x720the single value in the table
- Inference resolution720 x 480, no support for other resolutionsfrom the model card table
- What the released model produces30 FPS videos at a 1216x704 resolution
4From brief to placement
For ad work around “Single values”, SceneMixer's pages explain how its ad projects are produced:
The final cut ends on an end card: your main image plus four lines — Brand, Product name, Tagline, Call to action. The first two are read from your photo and your brief; the tagline and call to action are written with the script.
SceneMixer, The End Card (read 2026-10-13)
On the end card, yes, along with the product name, the tagline and the call to action, in the language of the ad.
SceneMixer, AI Ad Video Generator — “Will my brand name appear in the video?” (read 2026-10-13)
5Sources
Format requirements come from Google Ads Help, read 2026-09-12, and tool limits from each vendor page listed on published limits, read 2026-09-22. The rule against mixing the two is on method. Nearby: Published defaults, A minute in one call, Ceilings on inputs.