MiniMax H3 Frame to Video

Direct motion from an exact first frame, optionally landing on an exact last frame. Stay prompt-only and never submit a generation job.

Confirm this workflow

Use this skill when the image is the actual first frame, not merely a style or identity reference. With two images, require the first to be the opening frame and the second to be the closing frame. Otherwise route to ../minimax-h3-reference-to-video/SKILL.md.

Recommend a 5–15 second clip, 768P for iteration or 2K for a final pass when the active workflow and hardware support it, and note that output aspect ratio should follow the first-frame image.

Build the motion bridge

1. State whether there is one first frame or a first/last pair.
2. Inventory visual invariants visible in the supplied frame: identity, object design, layout, lighting direction, text, and geometry.
3. Describe what begins moving, in what order, and how strongly.
4. Direct camera movement without contradicting the fixed starting composition.
5. With a last frame, describe a continuous causal path that naturally reaches it.
6. Add sound events and timing when useful.
7. Prohibit abrupt morphing, unmotivated cuts, duplicated elements, identity redesign, or geometry drift when those are likely risks.