MiniMax H3 Reference

Video Models

referenceimage to videovideo to videocharactersubjectaudio

Overview

Generate 2K video with native stereo audio from reference material. Accepts up to 9 images, 3 videos, and 3 audio clips to carry subjects, motion, and voice into the shot.

Requires at least 1 input. Accepts up to 9 images, 3 videos, and 3 audio clips (max 12 inputs total, shared across all types). Reference videos and audio clips must each be 2-15 seconds.

Image
Video
Audio