MiniMax H3 image to video,The single complete reasoning process takes 5 minutes.Upload a single first-frame reference image + prompt text, and the model will generate subsequent dynamic shots.Output is up to 2K resolution, 24fps, and up to about 15 seconds.