MiniMax H3 game footage prompts

410 results

Third-person combat, ability activation, and environment traversal. Game footage fails in AI generation when the prompt describes a game engine instead of describing footage — "character uses skill" produces a UI animation, not a fight.

The pattern in these prompts: state the camera as a player would see it (over-the-shoulder, locked-on, orbiting), name the input that triggers the action, and specify what the environment reacts with — weapon trails, ground impact, screen shake amplitude. Describing the engine produces UI; describing the shot produces footage.

Keep the camera declaration at the very front. H3 weights early tokens more heavily, and a camera stated after the action gets partially overridden.