It removes the need to film or animate short video concepts manually, letting creators prototype ads, social clips, or pre‑visualizations quickly with limited resources.
MiniMax H3 is a web‑based AI video generator that produces 4–15 second clips at up to 2K resolution. Users supply a textual prompt and optionally a first‑frame image, a last‑frame image, and up to nine reference images, three reference videos (total ≤15 s) and three audio files (≤15 s). The system concatenates these inputs into a multimodal representation, feeds it to a proprietary diffusion model, and renders a short video with controllable motion, camera moves, and visual style. Resolution can be set to 768 p or 2K, and duration is chosen from preset lengths. Credits are deducted per second of output (9 credits/s at 768 p, 14 credits/s at 2K). Generated files are downloadable without watermarks.
Have feedback for the maker?
Sign in to leave a review, report a bug, or suggest a feature.
Leo poppy
