2K AI video creation with MiniMax H3
Turn every creative starting point into a cinematic video
MiniMax H3 AI Video Creator turns a written idea, one image, a pair of opening and closing frames, or image, video, and audio references into one video with native stereo sound. Each clip runs 4 to 15 seconds at 720p by default or 2K on request, for ads, ecommerce, brands, games, and social feeds. MiniMax H3 AI Video Creator works in Claude Code, Codex, Cursor, and other agents that run Skills.
Install MiniMax H3 AI Video Creator and tell me when I need to sign in: https://beatra.ai/skills/minimax-h3-ai-video/install.md
For agents:Install guideMarkdown version
Send it to WorkBuddy, Codex, Claude Code, or another agent to install. Sign in once the first time you use it.GitHub ↗Docs ↗
From first idea to a finished H3 video
Create from any spark
Start with an idea, a still image, opening and closing visuals, or a mix of image, video, and audio inspiration.
2K scenes with native stereo sound
Build polished movement and rich sound together for a cinematic result that feels complete from the first play.
Made for brands and creators
Shape camera movement, visual atmosphere, brand presentation, and story beats for the places your audience watches.
How it is made
Start with words, images, key frames, or creative references
Share the subject, destination, and visual feeling you want. MiniMax H3 turns different creative starting points into one focused video direction.
Written idea or scene · Product, character, or world image · Opening and closing visual · Visual, motion, and sound references
Shape movement, atmosphere, brand presence, and sound
Guide the camera, subject action, light, pacing, and soundscape so the video carries a clear point of view.
Readable camera movement · Distinct visual atmosphere · Focused subject action · Sound that supports the story
Bring the whole direction together in polished 2K
MiniMax H3 unites visual references, motion ideas, and native stereo sound into a finished clip with cinematic energy.
Polished 2K video · Native stereo sound · Strong brand presentation · Coherent reference-led storytelling
Move from one standout clip to a consistent creative series
Use the finished direction across advertising, ecommerce, brand, entertainment, and social storytelling.
Advertising and product launches · Brand and campaign stories · Dynamic posters and title moments · Game and social content
Built for commercial stories and standout content
AI advertising and ecommerce videos
Turn product ideas and hero images into attention-grabbing launch clips, product moments, and campaign creative.
Brand films and dynamic posters
Build memorable reveals, expressive type moments, atmospheric motion, and sound-led brand stories.
Game UI and film titles
Create cinematic interface motion, world-building beats, opening imagery, and title sequences with a strong visual rhythm.
Social stories and launches
Make striking short-form video for announcements, creator posts, campaign hooks, and feed-ready storytelling.
Inputs and outputs
Input · Text — A written scene: subject, action, camera, light, and sound
Input · Image — An image to animate, opening and closing frames, or reference images
- Formats
- JPG, PNG, WEBP
- Specifications
- up to 5 images per request
Input · Video — Reference videos for movement or style (optional)
- Formats
- MP4, MOV
- Specifications
- up to 3 videos per request; up to 15 s
Input · Audio — Reference audio for sound or pacing (optional)
- Formats
- MP3, WAV
- Specifications
- up to 3 audio clips per request; up to 15 s
Output · Video — Video with native stereo sound
- Formats
- MP4
- Specifications
- 1 video per request; 4–15 s; 720p, 2K; aspect ratios 21:9, 16:9, 4:3, 1:1, 3:4, 9:16
| Input or output | Type | Formats | Specifications |
|---|---|---|---|
| Input | Text — A written scene: subject, action, camera, light, and sound | — | — |
| Input | Image — An image to animate, opening and closing frames, or reference images | JPG, PNG, WEBP | up to 5 images per request |
| Input | Video — Reference videos for movement or style (optional) | MP4, MOV | up to 3 videos per request; up to 15 s |
| Input | Audio — Reference audio for sound or pacing (optional) | MP3, WAV | up to 3 audio clips per request; up to 15 s |
| Output | Video — Video with native stereo sound | MP4 | 1 video per request; 4–15 s; 720p, 2K; aspect ratios 21:9, 16:9, 4:3, 1:1, 3:4, 9:16 |
Specifications last checked · Sources: Model catalog · Documentation · Terms of Service
Questions
What can I create with MiniMax H3?
Create 2K advertising videos, ecommerce product moments, brand films, dynamic posters, game UI motion, film titles, and social clips with cinematic movement and native stereo sound.
Can I turn a written idea into video?
Yes. Describe the subject, action, setting, camera feeling, light, and soundscape, and the written concept becomes one video with native stereo sound.
Can I animate a product, character, or artwork image?
Yes. Start from the image and describe how the subject and camera should move, with the subject from your image at the center of the scene.
Can I create a video between opening and closing frames?
Yes. Use the two visuals to shape a deliberate beginning, transformation, and ending for product reveals, story beats, and scene changes.
Can image, video, and audio references guide one creation?
Yes. Give each reference a clear role for subject, style, movement, pacing, or sound, and they are combined into one video.
Related Skills
