AI fashion video generator API: editorial films, lookbooks

To generate an AI fashion editorial video with the Sume API, run the catalog Format sume-fashion-editorial at POST /v1/formats/sume/sume-fashion-editorial/runs with the look's photos attached and the styling brief in instruction. For a lookbook, queue one run per look in a single bulk request; to control each step, make a cover-pose still on POST /v1/images and animate it on POST /v1/videos.

The facts below come from Sume's Format catalog, Create a run, Bulk runs, Image API, and Video generation docs, read on 2026-09-27. Putting one specific garment on a person is a different job, covered in Virtual try-on video API.

Which Sume Format fits a fashion video?

Three catalog Format descriptions use the word “fashion”, and they split the work three ways: a styled campaign film, a static cover, or a supplied item on a person. The quoted wording is each Format's stated aim, not a guarantee about any one result.

How do I brief one editorial film?

Attach the look as input_image entries with public HTTPS URLs, such as styled shots or a mood reference: up to 30 images per run, 30 MB each. Write the styling in instruction: the pose, the set, the ratio, and the pace. The instruction is composed after the Format body and wins where they disagree. model on the run picks only the orchestrating LLM; the Format's tools choose the image and video models.

An API run is unattended: approvals a chat recipe would wait for are treated as granted, and a run that cannot finish comes back failed. Runs that make video take minutes, not seconds, so take the result by webhook or poll.

How do I make a whole lookbook?

Queue one run per look with POST /v1/formats/sume/sume-fashion-editorial/bulk-runs. items holds 1–100 ordinary run bodies in order, each with its own instruction and attachments, and concurrency (1–16) sets how many run at once. The queue has no webhook: poll GET /v1/format-run-queues/{queue_id}, or set communication.webhook_url on each item. Queue mechanics are covered in Sume Format bulk runs.

curl -sS -X POST "https://api.sume.com/v1/formats/sume/sume-fashion-editorial/bulk-runs" \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: fall-lookbook-v1" \
  -d '{
    "concurrency": 2,
    "items": [
      {
        "instruction": "Look 1: charcoal coat. 9:16 cover-story film.",
        "attachments": [{ "type": "input_image", "image_url": "https://example.com/look-01.jpg" }],
        "generation_spend_cap_usd": 20
      },
      {
        "instruction": "Look 2: ivory slip dress. 9:16 cover-story film.",
        "attachments": [{ "type": "input_image", "image_url": "https://example.com/look-02.jpg" }],
        "generation_spend_cap_usd": 20
      }
    ]
  }'

Can I build the editorial shot myself?

Yes, in two calls: make the cover-pose still on POST /v1/images, then send the still you pick to POST /v1/videos in frame_images as the first_frame. The image URL comes back Sume-hosted and signed, so host the still at your own public HTTPS URL first. Pick a portrait ratio that both your image model and your video model list, such as 3:4 or 9:16 on the docs' seedance-2 record. 4:5, which the docs call Instagram portrait (1080×1350), is an image ratio only.

What does it cost, and what are the limits?

Budget per run and per queue:

• Each run's generation is metered at the API pricing rates and capped by its generation_spend_cap_usd, up to $500. In a bulk request, set it on each item.
• A bulk request holds up to 100 items with at most 16 in flight, and every child still passes ordinary run admission: the wallet, workspace generation concurrency, and spend caps.
• Direct video jobs are reserved on submit at the provider's list price × 1.25, plus a 5.5% agent fee by default.
• The Format picks its own models, and its description is an aim, not a promise about any one film.