Product photo to holiday video ad: attach the photo to a Format run
Turn a product photo into a video ad with the sume-product-commercial Format. Attach up to 30 images by URL or asset id; media URLs in input share that budget.

To turn a product photo into a holiday video ad, call a Format from the Sume catalog and attach the photo to the run. The catalog includes sume-product-commercial, and the call is POST /v1/formats/sume/sume-product-commercial/runs with an instruction and an attachments array of input_image entries. The run returns a receipt with an arun_ id right away, and the finished video appears at primary_output_url once the run is completed.
What you can attach
A run can carry up to 30 images the agent can look at, each as a public HTTPS image_url or as an asset_id from the Assets API. Sume fetches a URL when the run is created, so it must be reachable without a login. Media URLs placed inside input share the same budget of 30 files per run.
| Item | Limit or rule |
|---|---|
attachments | Up to 30 images; type is input_image |
| Image source | Public image_url or an uploaded asset_id |
Media URLs in input | Share the 30-file budget (max 30 images, 10 videos, 10 audio) |
instruction | Up to 8000 characters |
input | JSON object, at most 64 top-level keys and 2 MiB |
| Over a limit | 400 invalid_attachment |
| Unreachable image | 502 attachment_fetch_failed with details.index |
Steps
The run is a single request, but a few habits keep it cheap and repeatable.
- Pick a clean photo with the product fully visible, and host it at a public URL or upload it to get an
asset_id. - Write an
instructionthat names the offer and the mood, such as a calm gift-box reveal, and keep promo text out of the photo prompt if you will burn it in later. - Send the request with an
Idempotency-Keyand ageneration_spend_cap_usdso a bad run has a ceiling. - Poll the receipt until the status is terminal, then read
primary_output_urland theartifactslist.
curl -X POST https://api.sume.com/v1/formats/sume/sume-product-commercial/runs \
-H "Authorization: Bearer $SUME_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: gift-box-hero-001" \
-d '{
"instruction": "Calm gift-box reveal, warm light, 9:16.",
"attachments": [
{"type": "input_image", "image_url": "https://example.com/gift-box.jpg"}
],
"generation_spend_cap_usd": 20
}'What Sume does not do
The Format catalog page lists the slug, but the exact look and the input fields vary by Format, so read the Format with GET /v1/formats/sume/{slug} before you depend on a field. Sume does not promise that the product in the video matches the photo to the pixel; check the result before it goes into an ad account.
A replay of the same idempotency key with a different image list returns 409 idempotency_conflict, while a true replay does not fetch your images again. That protects you from a double charge, and it also means a changed photo needs a new key.
Shot list for a seasonal ad
For a holiday offer, the photo carries most of the result, so spend your effort there. Use a sharp shot on a plain background, avoid heavy text on the packaging if you can, and send a second angle as another attachment when the product has a back or a label that matters. More images give the agent more to work from, up to the 30 allowed, but each extra file is another thing that must be fetchable, and one unreachable image fails the attachment step with its index in the error.
Keep the promotional words, such as a percentage or a code, out of the prompt if they must be exact, and burn them in afterwards with a caption job using cues. That way a wrong digit is caught in text you wrote and not in text a model drew.
Sources
Related posts
More in Use cases
- Qwen Image edit with 10 references on Sume: four ratios for 10 cents
qwen/qwen-image on Sume bills about $0.025, takes up to 10 references and lists 13 ratios. Four ratio variants of one product shot cost 10 cents. Steps inside.
- Recap clip after a demo call: a 30 second avatar video for $5.52
Send a 30 second avatar recap after a live demo: $5.52 on Sume standard, $7.35 on plus, $16.50 on max. Script plan, scene split and the 60 second limit.
- Recast a 14-second spokesperson clip: $5.25 at 768p, $7.88 at 1080p
A 14-second clip sits one second under the 15-second shot cap of H3 Max Recast. Cost on Sume is $5.25 at 768p and $7.88 at 1080p; here is how the limits apply.
- Recast a 22-second explainer host: $8.25 at 768p, $12.38 at 1080p
Replace the host in a 22-second explainer video with H3 Max Recast on Sume: $8.25 at 768p, $12.38 at 1080p, plus the cut-every-15-seconds rule to check first.
Written by Sume