Meta Advantage+ and 3+ formats: one product clip, three shapes

Meta tells advertisers that adding 3 or more creative formats can make the difference, and that the more formats you provide, the more ways Advantage+ can match creative to Reels, Stories or Feed. With Sume you can make one product clip and conform it to three output shapes using the output block of POST /v1/video-trim.

Meta's claims are from its holiday creative page, read 2026-10-01. Sume behavior is from the video trim, Timeline 1.0 and video generation docs. Meta's page does not list pixel sizes, so take those from Meta's own ad specs.

What does Meta say about formats?

The page says 86% of global holiday shoppers use social media for inspiration across Reels, Stories and Feed, and that there is a 2.1x higher purchase intent when shoppers engage with diverse formats. It says the more formats you provide, the more ways Advantage+ can match the right creative to the right person. It also says repurposing comes first: "Repurpose first, produce second, refresh throughout".

How do I make three shapes from one clip?

Video trim takes one media.sume.com clip, a start, exactly one of end or duration, and an optional output of { width, height, fps }. The docs call it a conform on the way out. It works with precision: "exact" only; with keyframe it fails video_trim_output_requires_exact. Run the call three times with different output values.

curl -X POST https://api.sume.com/v1/video-trim \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: reel-cut-001" \
  -d '{
    "video_url": "https://media.sume.com/artifacts/artf_demo/product.mp4",
    "start": 0,
    "duration": 15,
    "output": { "width": 1080, "height": 1920, "fps": 30 }
  }'

What if the shapes have different aspect ratios?

The trim docs do not say how a source of one aspect ratio is fit into a different output size, so check the result before shipping. Timeline 1.0 states it: video[].fit is cover (default), contain, stretch or blur, and output.width / height are even integers from 256 to 2160. For a reframe, a one-slot Timeline render with an explicit fit is the better-defined route. If a model supports the shape natively, supported_aspect_ratios on the models endpoint lists it, for example 21:9, 16:9, 4:3, 1:1, 3:4 and 9:16 on the Seedance 2.0 example in the docs.

How many versions can I queue?

If you drive the work through a Format, a bulk request queues up to 100 Format runs with a concurrency window. Direct media calls like trim are single jobs; submit each with its own Idempotency-Key. Pair the three shapes with a phase plan from holiday ad creative by phase.