Luma Scenes Ray 3.2 or Seedance 2: pin a model or use auto

Luma Scenes renders with Ray 3.2 or Seedance 2, in 9:16 or 16:9, and Luma says the two are not interchangeable: Ray 3.2 leans toward several short clips, Seedance 2 often returns one longer clip. On Sume the same choice is the model field. Pin a catalog id when you need a known clip length, or send sume/auto and accept that Sume does not say which family ran.

Luma's comparison is from its Scenes post. Sume facts are from the Video Router docs and Video generation, read 2026-10-01.

How do the limits differ per model?

The docs state that limits are per model, and that other catalog models are capped at 15 seconds, with a few exceptions listed on that page.

How do I see what a model accepts?

Read capabilities from GET /v1/video-router/models rather than assuming one envelope. Whether a given third-party model such as Ray 3.2 is listed is also answered by that catalog, so check it before planning around one.

When should I use sume/auto?

When you do not care which family renders. The response echoes sume/auto, and the docs say Sume does not disclose which family served the request, so do not infer it from the output. Auto create controls default to 720p and 8 seconds, with 3 to 10 second clips at 16:9 or 9:16.

How do I build a multi-clip storyboard?

Pin one model per scene if you need predictable length, and keep the aspect ratio the same across scenes. Approve stills first, as in approve keyframes, then render. Catalog limits for Luma's own models are compared in Ray 3.2 vs catalog limits.