Generate from Text
Describe the subject, action, setting, lighting, and camera movement. MiniMax H3 uses your prompt to generate a short video.
Create 2K Videos from Text, Frames, and References
Create 5–15 second 2K videos from text, frames, and reference media. Sign up to receive 90 credits and start generating online.
90 credits after signup
Text-to-video generation
Starting and ending frames
Image, video, and audio references
5–15 seconds in 2K
Creative starting points
Watch MiniMax H3 AI video examples built from text prompts, frame controls, and multimodal references across ads, games, and creative work.
Flexible creative controls
Explore MiniMax H3 features for text-to-video, first and last frame control, 2K output, reference media, motion, camera, and visual style.
Describe the subject, action, setting, lighting, and camera movement. MiniMax H3 uses your prompt to generate a short video.
Upload a starting frame, an ending frame, or both. These images help guide how the video begins or ends.
Add supported images, video, or audio to provide more context for the subject, motion, camera movement, and visual style.
Describe pans, zooms, tracking shots, static shots, and other camera directions in your prompt.
Choose a duration from 5 to 15 seconds and generate the final video in 2K.
References provide more context for important visual details. Results vary by source quality and scene complexity.
Four-step workflow
Learn how to use MiniMax H3 online. Choose a workflow, write a clear video prompt, add supported reference media, then generate a 2K video.
Sign up and get 90 credits to start creating videos.
Select text-to-video, first/last-frame, or reference-based generation.
Write a clear video prompt. Add supported frames or reference media when you need more control.
Choose a 5–15 second duration. Review the credit estimate, generate your 2K video, and download the result.
Example prompt: A silver watch rests on dark stone. Soft light moves across the surface as the camera slowly pushes forward. Premium product commercial style.
Choose the right input
Compare text-to-video, first and last frame video, and multimodal reference generation to choose the right MiniMax H3 workflow for your project.
Workflow
What you provide
Good for
Text-to-Video
A written prompt
Exploring a new scene from scratch
First and Last Frame Video
A prompt and one or two frame images
Animating a still image or guiding a transition
Multimodal Reference Video
A prompt with supported images, video, or audio
Adding context for the subject, motion, camera, and visual style
Real project workflows
See how marketers, creators, filmmakers, game teams, and motion designers use MiniMax H3 to explore short-form 2K video concepts.
Marketers, ecommerce teams, and product designers
Turn a product image into a short ad concept without arranging a full product shoot.
Store owners and social media managers
Create short lifestyle clips from existing product photography.
Creators, social teams, and fashion brands
Explore vertical concepts for Reels, Shorts, and TikTok.
Directors, cinematographers, and agencies
Test shot direction, framing, movement, and scene mood before filming.
Game designers, concept artists, and trailer teams
Turn character art into a short movement and environment study.
Brand and motion designers
Explore animated poster and campaign directions with layered depth and controlled movement.
Review all generated text, logos, packaging, faces, and brand details before publishing the final video.
Account credits
Get 90 credits when you sign up. MiniMax H3 uses 14 credits per billable second, with reference video duration included when applicable.
Each second of output uses 14 credits. A 5-second video uses 70 credits. A 10-second video uses 140.
Add the total reference video duration to the output duration. Multiply the result by 14 credits.
3s reference video + 5s output = 8 billable seconds
8 × 14 = 112 credits
Frequently asked questions
Find quick answers about MiniMax H3 video creation, 2K output, frame controls, reference files, supported formats, and credit use.
MiniMax H3 is an AI video model. It creates short 2K videos from text, frames, and supported multimodal reference media.
Start with a text prompt, first or last frame, or multimodal reference media. Create a 5–15 second 2K video with MiniMax H3.