Which pages these figures were read from
Four entries in this register were read from a priced page and eleven from documentation: API references, model cards, user guides and usage limits. A rate card prints what is charged for; a reference prints what a parameter accepts. The columns they fill follow from that. As of 2026-09-22.
| Item | What is published | Detail |
|---|---|---|
| Rate cards | Four entries | Resolution almost always; length and shape rarely |
| API references | Five entries | Length, shape and resolution as request parameters |
| Model cards | Two entries | Exact pixel figures, frame counts and frame rates |
| Guides and limits | Four entries | Fixed formats, nested ceilings and upload caps |
Inclusion rule. Entries grouped by the kind of page each was read from. The counts are of entries in this register, not of pages a vendor publishes overall. Order. Most common source type first.
1A priced page prints what varies with the rate
Resolution appears on every rate card here because each step carries a different charge. Length appears on two of them and shape on one, because neither usually changes what a request costs in a way the page needed to express.
That is why two entries in this register have nothing in the length column at all. Both were read from rate cards, and a schedule cannot be built from a page written to answer a cost question.
2A reference prints what a parameter accepts
The documentation entries fill the length, shape and resolution columns because each is a field a caller supplies. They also carry the conditions: a model identifier for a longer run, a version for a duration rule, a tier for the top of a range.
What they omit is price. Two of them state a per-second basis and the rest say nothing, so a plan assembled from documentation knows what it can ask for and not what asking costs.
3Model cards answer a third question entirely
The two model cards publish exact pixel pairs, frame counts and frame rates, because those are the quantities someone running weights has to budget. They are also the only entries in the frame rate column alongside one user guide.
None of the three document types is more reliable than the others. Each is accurate about the question it was written to answer, and these sheets record which question that was so a gap can be read as a gap rather than as a limitation of the product.
- How generation is billedCharged per second of output, with gen4_turbo at 5 credits per second and gen4.5 at 12
- Video duration6 second increments, up to two minutesfrom the model table
- Resolutions720p, 1080p and 4Kpriced per step
- What the released model produces30 FPS videos at a 1216x704 resolution
4Making the ad itself
For ad work around “Where figures were read”, SceneMixer's pages explain how its ad projects are produced:
Every character, scene and prop card can carry a short list of the other names your script uses for the same thing.
SceneMixer, Also Called: Nicknames, Old Names, and How the Storyboard Knows Who Is Who (read 2026-10-13)
The spoken lines and the end card follow your interface language across fifteen dialogue languages, with the cast and setting adapted to that market. The end card falls back to a language it can typeset when a script needs shaping or right-to-left layout.
SceneMixer, AI Ad Video Generator — “Which languages can the ad be in?” (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 2026-09-12 and 2026-09-22. The rule against mixing the two is on method. Nearby: Length tied to a tier, Who publishes a rate, How long output stays.