AI magazine cover generator API: covers with room for type

To generate a magazine-cover-style image with the Sume API, run the catalog Format sume-magazine-cover-campaign at POST /v1/formats/sume/sume-magazine-cover-campaign/runs, with your brief in instruction and the portrait or product photos in attachments. Its description names clean campaign typography zones, which reads as room for type: plan to set the masthead and cover lines yourself.

The facts below come from Sume's Format catalog, Create a run, Structured output, and Image API docs, read on 2026-09-27. The Format's description is quoted from its catalog entry.

What does the magazine cover Format make?

sume-magazine-cover-campaign (catalog title: Sume Magazine Cover Campaign) is a catalog image Format. Its description reads: "Create a finished magazine-cover-style campaign image with a bold portrait, designed negative space, and clean campaign typography zones. Use when the user asks for beauty launches, fashion campaigns, cover reveals, and branded editorial social posts. Not for: animated or motion deliverables."

That is the Format's stated aim, not a promise about a given image, and it makes a still rather than a moving cover. GET /v1/formats/sume/sume-magazine-cover-campaign returns the description; the recipe body is not returned, because it reaches the agent, not the caller. Any key with formats:write may call the Format.

How do I brief a cover?

A cover run carries the same fields as any Format run. The ones that shape a cover:

• Attach the portrait and the product as input_image items with public HTTPS image_urls, up to 30 per run. filename is the label the agent sees.
• Say where your type will go in instruction, such as a clear top third for the masthead. It accepts 8000 characters, and about the first 4000 reach the run.
• Sume publishes no field list for input; the Format's recipe reads the keys it recognizes. No documented field carries masthead text, so keep the words on your side.
• Bind an output_schema with a SumeMediaFile# field for the cover, and make any text field nullable. On the projection path, text that would come from your brief comes back null.

curl -sS -X POST "https://api.sume.com/v1/formats/sume/sume-magazine-cover-campaign/runs" \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: acme-spring-cover-v1" \
  -d '{
    "instruction": "Spring launch cover. Keep the top third clear for a masthead.",
    "attachments": [
      { "type": "input_image", "image_url": "https://example.com/model.jpg", "filename": "model.jpg" },
      { "type": "input_image", "image_url": "https://example.com/cream.png", "filename": "cream.png" }
    ],
    "output_schema": {
      "name": "acme/cover/v1",
      "schema": {
        "type": "object",
        "additionalProperties": false,
        "required": ["cover_image", "alt_text"],
        "properties": {
          "cover_image": { "$ref": "SumeMediaFile#" },
          "alt_text": { "type": ["string", "null"] }
        }
      }
    },
    "primary_output_key": "cover_image",
    "generation_spend_cap_usd": 10
  }'

How do I get the cover back?

The create answers 202 with a run receipt. Poll GET /v1/format-runs/{run_id}, or send communication.webhook_url and receive one signed format.run.terminal POST. With the schema above, primary_output_url is the cover. Its media.sume.com URL does not expire, and it is public to anyone holding it.

If nothing the run made satisfies your schema, output is null, output_error says why, and over the API the run ends failed; artifacts[] still lists what the run made. Then lay out the masthead and cover lines in the zones, in your own design step.

Can I generate a cover with the Image API instead?

Yes, when you want to choose the model and write the prompt. POST /v1/images takes the portrait and product as input_references with public HTTPS URLs, and the options below shape a cover. A model accepts only the values its catalog descriptors list, so read GET /v1/images/models first; a parameter the model does not list is refused with 400 unsupported_parameter.

Ratios and sizes are covered in AI image generation API aspect ratios, and 4K and image counts in 4K AI image generation API.

What are the limits?

From the same docs pages:

• Attachments are JPEG, PNG, WebP, GIF, or AVIF, up to 30 MB per image and 500 MB per run. input_image is the only attachment type.
• Format generation is metered at API pricing rates under generation_spend_cap_usd: up to $500, null runs at $500, and 0 is rejected.
• Image API billing is all-or-nothing: a completed generation is billed in full, a failed one is not billed. Its data[].url is a Sume-hosted signed URL.
• The Format's description rules out animated or motion deliverables. For a campaign film, see AI fashion editorial video API.