Source files

0/15

Preview

Available now

Create 2K audiovisual scenes with MiniMax H3

Use MiniMax H3, also known as Hailuo 03, to turn text, images, video, and audio references into complete audiovisual scenes. Create native stereo sound and video together, generate clips up to 15 seconds in 2K, transfer motion from reference footage, and refine selected content while preserving the wider scene. AI Image Editor is an independent service and is not MiniMax's official website.

Up to 15 secondsUp to 2KNative stereo audioMultimodal references
Availability
Available

Generate with MiniMax H3 directly in AI Image Editor.

Output
2K video + stereo audio

Generate video and synchronized native audio in the same result.

Context
Text · image · video · audio

Use multiple reference types to communicate subject, motion, style, and sound.

MiniMax H3 video generation features

MiniMax H3 brings multimodal context, audiovisual generation, reference control, and in-context editing into a unified video model.

Generate video with native stereo audio

Create visuals and sound together so dialogue, ambience, music, and effects can follow the timing and action of the scene.

Create clips up to 15 seconds in 2K

Build a complete short scene with enough time for staged action, camera movement, interaction, and an ending beat.

Use multimodal creative references

Combine text, images, video, and audio to define characters, objects, movement, visual language, and soundtrack direction.

Transfer motion from reference video

Guide performance and physical movement with source footage while adapting the subject, environment, or visual treatment.

Regenerate selected video details

Use contextual editing to change targeted content while preserving the surrounding scene, timing, composition, and sound where possible.

Render readable branded visuals

Direct product details, interface elements, signs, and short text while keeping brand-sensitive content subject to final review.

What you can create with MiniMax H3

Plan audiovisual advertising, reference-driven character scenes, product stories, social content, game concepts, and interface demonstrations.

Produce product and brand video concepts

Combine product references, campaign direction, scene timing, camera instructions, and sound cues for a cohesive commercial draft.

Define the product story

Describe the audience, benefit, visual sequence, product interaction, camera path, and final call to action.

Supply approved references

Add product views, logos, colors, packaging, style frames, motion examples, voice, music, and ambience.

Verify every brand detail

Review product shape, labels, on-screen text, audio, claims, rights, and required disclaimers before use.

Direct character action with motion references

Use character images and reference footage to establish appearance, performance, movement, camera behavior, and scene rhythm.

Build a consistent reference set

Define the character, clothing, props, environment, palette, lighting, and voice or sound direction.

Choose clear motion footage

Use source action with readable body movement, timing, framing, contacts, and spatial relationships.

Inspect reference fidelity

Check faces, hands, anatomy, distant subjects, object persistence, interactions, and motion continuity after generation.

Create social and interactive product stories

Prototype social shorts, game scenes, app demonstrations, ecommerce stories, and other audiovisual content with ordered action.

Choose the delivery format

Set the channel, aspect ratio, length, pacing, safe areas, and room for captions or interface overlays.

Write synchronized visual and audio beats

Pair each action and camera change with the intended dialogue, effect, ambience, or musical cue.

Finish in post-production

Confirm continuity and rights, then add verified copy, captions, branding, mix adjustments, and final edits.

How to plan a MiniMax H3 generation

01

Write the scene and sound direction

Describe subjects, action, environment, camera, lighting, timing, dialogue, ambience, effects, and music in chronological order.

02

Prepare the relevant references

Organize approved images, video, and audio that clearly define identity, motion, composition, style, and sound.

03

Generate, inspect, and finish

Review visual continuity, reference fidelity, text, motion, audio synchronization, and rights before completing the edit.

MiniMax H3 FAQ

What is MiniMax H3?

MiniMax H3 is a multimodal video generation model from MiniMax. It is also presented as Hailuo 03 and is designed to understand text, image, video, and audio references in a unified context.

Is MiniMax H3 available in AI Image Editor?

Yes. MiniMax H3 is available in AI Image Editor for text-to-video, image-to-video, and multimodal reference-to-video generation.

What generation modes does MiniMax H3 support?

MiniMax H3 supports text-to-video, image-to-video, reference-to-video, motion transfer, and contextual video regeneration for targeted creative changes.

How long can MiniMax H3 videos be?

MiniMax H3 can create audiovisual clips from 4 to 15 seconds, giving you flexibility for short motion tests, ads, and complete social scenes.

What resolution does MiniMax H3 support?

MiniMax H3 supports 768P and 2K video output. Choose the resolution in the generator before submitting your request.

Does MiniMax H3 generate audio?

Yes. MiniMax H3 can generate native stereo audio and video together, including dialogue, ambience, sound effects, and music direction.

How many references can I use?

Reference-to-video supports up to 9 images, 3 videos, and 3 audio files, so you can guide identity, motion, style, composition, and sound in one scene.

Are reference results always identical to the source?

No. Reference generation can vary, especially with distant subjects, complex movement, or many competing inputs. Review identity, anatomy, motion, sound, and legal permissions before publishing.

Is AI Image Editor the official MiniMax H3 website?

No. MiniMax H3 is developed by MiniMax. AI Image Editor is an independent service and is not affiliated with or endorsed by MiniMax.

Create with MiniMax H3

Choose a generation mode, add any reference assets, set the duration and resolution, then generate your MiniMax H3 video.

Available StatusUp to 15 seconds Duration