How Do I Use First-and-Last-Frame Generation with the MiniMax H3 API?

Overview

First-and-last-frame control lets you supply the opening and closing images and have the model generate the motion between them, which is powerful for precise transitions. Confirm that your MiniMax H3 route supports it and how the two frames are passed in the request schema, since not every access path exposes this mode.

Source note: Verified 2026-08-06 against the MiniMax official H3 blog, MiniMax Video Generation API docs, and Hugging Face MiniMax-H3 model page.

Use it where predictability matters. By fixing both endpoints, you constrain the animation far more than a single prompt does, so it suits product reveals, controlled camera moves, and shots that must land on an exact final frame.

Choose two frames that share consistent lighting, framing, and subject placement; if the start and end disagree too much, the model has to invent a jarring path between them and the result looks unstable. The aspect ratio and resolution of both frames should match what the endpoint expects.

Before scaling this workflow, also check MiniMax H3 pricing for controlled routes so you can account for any route-specific cost differences.

Set the motion expectation in your prompt too, describing how the transition should feel — a slow morph, a camera push, a steady reveal — rather than leaving the in-between entirely to chance.

Once the request is built, it runs through the same asynchronous flow as any other generation: submit, store the task ID, poll for completion, then retrieve and save the finished clip.