MiniMax H3 AI Video Generator

Create up to 15-second 2K videos with native stereo audio using MiniMax H3. Guide each shot with text plus image, video, and audio context. Direct motion, sound, scene changes, and edits in natural language.

How to Use MiniMax H3 in DomoAI

Start with one short clip, add only the references it needs, then generate and compare results in Omni Reference.

A character portrait being dragged into an image, video, or audio upload area

Add the Right Context

Start with a prompt. Add supported image, video, or audio references only when they help define the clip. Explain what you want each reference to guide.

Prompt screen reading ‘Describe your idea in detail; the more specific your prompt, the better the result.’

Describe the Shot From Start to Finish

Describe the visual style, subject, setting, main action, camera movement, and desired ending. Keep one main action per short clip.

Three generated variations of a giant stone figure reaching toward a lone warrior in a misty landscape

Generate, Compare, Iterate

Generate a few versions. Change one prompt detail or reference at a time, then compare the motion, framing, and overall fit.

Minimax H3 reference-based video generation combining character, scene, object, and audio inputs

What Reference Inputs Does MiniMax H3 Support?

Choose one MiniMax H3 workflow. Omni Reference supports up to 9 images plus 3 video/audio files total . Frames to Video uses dedicated Start Frame and End Frame slots. A prompt is required, and the workflows cannot be combined.

Omni Reference: AI Music Video Shot

Use Image 1 for the singer, Image 2 for the stage, Video 1 for movement, and Audio 1 for lip-sync. Slowly arc clockwise from medium shot to close-up, ending with direct eye contact. No cuts, text, extra people, or identity drift.

First-to-last-frame product shot

Start exactly from the Start Frame, keeping the can unchanged. In one continuous macro shot, water curls around its base as the camera slowly arcs clockwise. Match and end exactly on the End Frame. No cuts, extra objects, label changes, or deformation.

Who Is MiniMax H3 For?

MiniMax H3 can help creators and teams explore short videos from text or supported image, video, or audio references.

Contemporary dancer in black performing amid motion-blurred figures and orange stage lights

Filmmakers and Story Creators

Explore short cinematic scenes and character moments with visual or motion references. Generate each 4–15-second shot separately, then select clips for a longer sequence.

Transparent retro electronic devices displayed as a glossy product concept under colored studio lights

Brand and Social Teams

Explore product visuals, campaign clips, and social concepts. Review product shape, labels, text, and movement before publishing.

Low-poly 3D character in a green cap and red scarf against a vivid blue sky

Music Video and VTuber Creators

Plan individual 4–15-second shots around a song, character, or visual concept. Use supported references to guide each clip, then arrange them into a longer sequence.

How to Write a Clear MiniMax H3 Video Prompt

A clear H3 prompt describes the visual style, subject, setting, main action, camera movement, and desired ending in order.

Minimax H3 text-to-video generation of a fashion editor working in a retro office

Set One Main Outcome

Choose one primary action and a clear ending. A focused prompt gives a short clip enough room to show motion and consequence.

Assign Each Reference a Job

Tell H3 what you want each reference to guide. For example, an image may guide appearance, while a video may guide action or camera direction.

Describe Cause and Effect

Name what starts the motion and how the environment responds. Describe water splashing, fabric trailing, dust rising, or objects reacting to impact.

Describe the Camera Direction and Ending

Name one camera movement and its pace, then describe the ending composition you want. When using a first or last frame, explain how the shot should develop from or toward it.

MiniMax H3 FAQ

What is MiniMax H3 in DomoAI?

MiniMax H3 is an optional AI video generation model in DomoAI Omni Reference. You can use it for text to video, image to video, or supported video and audio reference workflows.

How do I use MiniMax H3 in DomoAI?

Select MiniMax H3 in Omni Reference, add references, enter a prompt, and generate. For start/end frames, use Frames to Video.

How long can MiniMax H3 videos be?

MiniMax H3 generates videos from 4 to 15 seconds long in DomoAI Omni Reference.

What inputs does MiniMax H3 support?

Up to 9 images, 3 videos, and 3 audio files. Video and audio clips can be 2–15 seconds each, with a 15-second total limit per type. A prompt is required for every generation.

What resolution does MiniMax H3 support?

MiniMax H3 supports 768P and up to 2K output in DomoAI Omni Reference.

Can new users try MiniMax H3 for free?

Yes. New users receive 30 free credits after signing up, which they can use to try MiniMax H3.

More From DomoAI

Generate, stylize, and upscale in one place

Explore More