Google Ads allows ratios between 16:9 and 9:16: Sume models for 3:4
Google Ads lists 16:9, 1:1 and 9:16 and allows ratios between them. Seedance, Wan 3.0 and MiniMax H3 render 3:4 and 4:3 on Sume; Kling 3 and Omni do not.

Google Ads allows any aspect ratio between 16:9 and 9:16, and it names three as recommended: 16:9, 1:1 and 9:16. Seedance (all four models), Wan 3.0 and both MiniMax H3 models can render the in-between ratios 4:3 and 3:4 on Sume. Kling 3 lists only 16:9, 9:16 and 1:1, and Gemini Omni Flash only 16:9 and 9:16.
What Google Ads says
The Google Ads Help page on video asset specifications for Video and App campaigns recommends horizontal 16:9, square 1:1 and vertical 9:16. It then states: "Aspect ratios between these settings are allowed. On the YouTube app, the video player will automatically adjust to ratios between 16:9 and 9:16."
The same page warns that campaigns may not serve if they only have vertical videos, because vertical assets no longer serve in YouTube in-feed environments such as the homepage feed, Search results feed and Watch Next feed. A 3:4 or 1:1 asset is not vertical 9:16, so it is one way to avoid a vertical-only set without producing a separate 16:9 version.
Which ratios are in range
Width divided by height for 16:9 is 1.78 and for 9:16 is 0.56. Any ratio with a value between those two is in the allowed band. 4:3 is 1.33, 1:1 is 1.00 and 3:4 is 0.75, so all three are inside it. 21:9 is 2.33, which is wider than 16:9, and falls outside the band on Google's wording.
Seedance has fixed sizes for each ratio and resolution, so the in-between formats come with known pixel counts.
| Ratio | Width / height | Inside 16:9 to 9:16 | Seedance 720p size | Seedance 1080p size |
|---|---|---|---|---|
| 16:9 | 1.78 | Edge | 1280x720 | 1920x1080 |
| 4:3 | 1.33 | Yes | 1112x834 | 1668x1251 |
| 1:1 | 1.00 | Yes | 960x960 | 1440x1440 |
| 3:4 | 0.75 | Yes | 834x1112 | 1251x1668 |
| 9:16 | 0.56 | Edge | 720x1280 | 1080x1920 |
| 21:9 | 2.33 | No | 1470x630 | 2205x945 |
Which Sume models take which ratio
These lists come from the Sume video catalog. Always confirm in GET /v1/videos/models before a batch, because limits differ per model.
- Seedance 2.5, 2.0, Fast and Mini: 21:9, 16:9, 4:3, 1:1, 3:4 and 9:16, plus
autoandadaptiveon 2.5. - Wan 3.0: 16:9, 4:3, 1:1, 3:4 and 9:16, plus
autoandadaptive. - MiniMax H3 and H3 Max: 21:9, 16:9, 4:3, 1:1, 3:4 and 9:16, plus
adaptive. - Kling 3: 16:9, 9:16 and 1:1 only.
- Gemini Omni Flash 1.1: 16:9 and 9:16 only.
A practical use of the band
Take a campaign that has a 16:9 video and a 9:16 video. Add a 3:4 clip at 720p on Seedance 2.0 for 15 seconds. 834x1112 is 927,408 pixels against 921,600 for 720x1280, so the token count is 326,042 against 324,000, and the Sume price is $5.71 against $5.67 for the 9:16 version. A 1:1 clip at 960x960 has exactly the 9:16 pixel count and costs $5.67.
So the ratio barely moves the cost on Seedance at 720p. The tier and the length move it far more.
What to check
- The page does not say that every placement serves every in-between ratio, only that the ratios are allowed and that the YouTube app player adjusts between 16:9 and 9:16.
- The page does not list a length or file size. Use Google's other pages for those.
- Do not choose 3:4 for a model that does not list it; check each model's
supported_aspect_ratiosfirst.
Sources
Related posts
More in Models
- Google image model dates on Oct 7, 2026: Nano Banana and Imagen ids
One table of Google's image model release and shutdown dates on Oct 7, 2026, with the replacement Google names for each Nano Banana, preview and Imagen 4 id.
- Google image model shutdown dates, Oct 7 2026: Imagen 4 to 2.5 Flash
Google's deprecations table lists Imagen 4 as shut down Aug 17, 2026 and gemini-2.5-flash-image as shutting down March 15, 2027. Newer ids have no date yet.
- Google Shorts ad lengths 6-60 s vs Sume model limits: jobs per ad
Google lists 6-60 s for Video Reach and 10-60 s for Video Action, App and PMax Shorts ads. How many Sume jobs a 30 s or 60 s ad takes on each video model.
- OpenAI names Flare or Sunburst to replace gpt-image-1: which to pin?
OpenAI's deprecations page lists gpt-image-2.5-sunburst or -flare for gpt-image-1. Sume serves both at one price, so the choice is quality. A script to A/B.
Written by Sume