Runway Product Ad recipe vs Sume reference-to-video

Runway's Product Ad Recipe takes one to ten product photos plus optional style images, productInfo and a userConcept, and returns a finished ad video. On Sume there is no storyboard step: you send the photos as reference images to a video model that accepts them, and write the concept, product facts and camera direction into one prompt.

Recipe fields are from Runway's Product Ad page. Sume details are from the Video Router docs and Video generation, read 2026-10-01.

How do the Product Ad fields map?

Runway's page says the Recipe analyzes the product, builds a storyboard and renders a short video. Sume's closest shape is reference-to-video.

How do I point the prompt at a photo?

With reference_to_video, address media as <IMAGE_REF_0>, <IMAGE_REF_1> and so on, zero-based in list order. Write "the bottle in <IMAGE_REF_0> on a marble counter, slow dolly in" so the model knows which image is the product. On /v1/videos the equivalent field is input_references, which gives style or content guidance rather than exact frames.

Which models accept references?

Only models whose supported_input_references lists a type accept that type, and limits are per model. Read capabilities from GET /v1/video-router/models rather than assuming one envelope. Photos must be reachable over public HTTPS. Following Runway's own tip is also useful on Sume: clean, high-resolution product photos from several angles give the model more to work from.

What should I do next?

Start with one short clip and two or three angles of the product. For a luxury look see AI luxury product ad; if you already have an ad and only the product changes, use an edit instead, as in the product swap comparison.