Minimax H3

MiniMax H3 is a truly multimodal AI video generator. Mix up to 9 images, 3 video clips, and 3 audio tracks in a single request. H3 reads your characters, motion, camera moves, voices, and style together, then blends them into one coherent video with native stereo sound. Reference a face to keep it consistent, a dance clip to copy its choreography, or a voice sample to clone its tone — all at the same time.

Features:

Change anything without re-shooting. MiniMax H3 can swap characters and objects, replace backgrounds, relight a scene, rewrite dialogue, and clone voices — all from a plain-language instruction. Turn day into night, replace a green screen with a fairytale forest, or hand a character new lines in the same voice. Everything you didn’t touch stays stable, so you can iterate on a video shot by shot like a director giving notes.