TEXT · IMAGE · KEYFRAMES · REFERENCES

MiniMax H3 AI Video Generator

Create 5- to 15-second videos from text, images, keyframes, video, or audio references, with up to 2K output and synchronized native stereo audio.

Upload images

Upload a start frame, an end frame, or both.

15 credits

Example videos

Get inspired

Inspiration

Night drive

Inspiration

Glass and water

Inspiration

Fashion motion

Inspiration

Architectural light

Read the full MiniMax H3 tutorial for prompts, settings, and tips.

QUICK START

How to Use the MiniMax H3 AI Video Generator

Choose an input, add any required media, describe the shot, review your settings and credit estimate, then generate.

  1. Choose an Input & Add Media

    Choose a workflow, then upload frames or references if needed. Skip uploads in Text to Video.

  2. Describe the Shot

    Write the subject and action first, then add camera movement, lighting, visual style, and audio. When you upload media, focus the prompt on what should change over time.

  3. Review Your Settings

    Choose the duration, resolution, and aspect ratio, then check the estimated credit cost before starting the generation.

  4. Generate & Review

    Generate the video, then review the result in Your Creations. If something needs adjustment, keep what worked and change one instruction at a time.

Choose InputUpload MediaWrite PromptSet OutputGenerateReview

INPUT MODES

Choose the Right Input Mode

The generator has three main input tabs. Choose the one that matches what you already have and the level of control you need.

Image to Video

Upload a start frame, an end frame, or both, then describe the motion, camera behavior, timing, and audio you want.

Start FrameDefines how the shot begins.
End FrameGuides the target ending.
Start + EndControls both the opening and final composition while the model generates the transition between them.

First & Last Frame is part of the Image to Video workflow when both frames are uploaded.

Explore Image to Video Learn about First & Last Frame

Text to Video

Start with a prompt only. Describe the subject, visible action, camera movement, lighting, visual style, and audio without uploading media.

Use this mode when the shot can be communicated clearly through text alone.

Explore Text to Video

Reference to Video

Use image, video, and audio references when separate files need to guide different parts of the result, such as appearance, motion, timing, camera behavior, or audio.

Up to 9 images · 3 videos · 3 audio references

Give each reference a specific role in the prompt so the generator knows what that file should influence.

Explore Reference to Video

GENERATOR SETTINGS

MiniMax H3 AI Video Generator Settings

Adjust the resolution, aspect ratio, duration, and generation settings for the shot, then review the estimated credit cost before you start.

Resolution

Start at 768p for faster drafts, or switch to 2K when you need higher-resolution output.

768pDefault2KHigher resolution

Aspect Ratio

Choose an aspect ratio based on where the finished video will be used.

21:916:94:31:13:49:16
Cinematic · Landscape · Square · Portrait

Duration

Choose a clip length from 5 to 15 seconds. Shorter generations are useful for faster iteration, while longer clips give the shot more time to develop.

5s10s15s

Native Audio

Describe dialogue, music, ambience, or sound effects directly in the prompt when you want them included in the generated video.

DialogueMusicAmbienceSound Effects

Credit Estimate

The generator shows the estimated credit cost before you start. Credit use can change with duration, resolution, model, and reference inputs.

Review the live estimate

GENERATION TIPS

Get Better Results from the Generator

Keep the shot focused and give each prompt or reference a clear job.

Keep One Main Action

Short clips are easier to control when the subject performs one clear action instead of several unrelated actions at once.

Give the Camera One Direction

Use one clear camera instruction, such as a slow push-in, lateral tracking shot, static frame, or controlled orbit.

Use References Intentionally

Each uploaded file should control something specific, such as appearance, motion, timing, camera behavior, or audio.

Describe What Changes

When you upload an image, spend less of the prompt describing what is already visible and more on what should move, change, or happen next.

Refine One Instruction at a Time

If the result misses one goal, keep the parts that worked and change only the instruction responsible for the problem.

Read the MiniMax H3 Prompt Guide

YOUR CREATIONS

Review Your Generated Videos

Completed generations appear in Your Creations, where you can preview the full clip, check the final frame and audio, and download the result.

GeneratePreviewReviewDownload

If you want another version, keep the useful prompt and inputs, change one major instruction, and generate again.

CREDITS

Know the Cost Before You Generate

The workspace calculates an estimated credit cost from the selected model, duration, resolution, and reference inputs before you start.

A standard 5-second 768p generation currently estimates 15 credits. Longer clips, 2K output, and some reference combinations may increase the estimate.

Always review the live estimate shown in the generator before starting.

See Full Pricing

MiniMax H3 AI Video Generator FAQ

Quick answers about creating videos with MiniMax H3, including text-to-video, image-to-video, references, 2K output, native audio, credits, and generation settings.

Do I need to sign in to use the MiniMax H3 AI Video Generator?

You can explore the input modes, write a prompt, and adjust video settings before signing in. Sign-in is required when you upload media or start a generation.

What resolution, video length, and aspect ratios does MiniMax H3 support?

The current workspace supports 5- to 15-second MiniMax H3 videos at 768p or up to 2K output. Supported aspect ratios include 21:9, 16:9, 4:3, 1:1, 3:4, and 9:16. Available settings may vary by generation mode.

Can I use MiniMax H3 for text-to-video generation?

Yes. MiniMax H3 Text to Video lets you generate a video from a prompt without uploading media. Describe the subject, action, camera movement, lighting, visual style, and sound you want in the shot.

How does MiniMax H3 Image to Video work?

Upload a start frame, an end frame, or both. MiniMax H3 Image to Video uses the uploaded image to guide the visual composition while your prompt controls motion, camera behavior, timing, and audio. Use both frames to define the opening and ending composition, then let the model generate the transition between them.

What files can I use with MiniMax H3 Reference to Video?

The current MiniMax H3 Reference to Video workspace supports up to 9 images, 3 videos, and 3 audio references. Include at least one image or video. Use each file for a specific purpose, such as appearance, motion, timing, camera behavior, or sound.

Does the MiniMax H3 AI Video Generator create native audio?

Yes. MiniMax H3 can generate synchronized stereo audio with the video, including dialogue, music, ambience, and sound effects described in your prompt.

How much does the MiniMax H3 AI Video Generator cost?

The generator shows an estimated credit cost before you start. MiniMax H3 credit usage can vary by duration, resolution, generation mode, and reference inputs. A standard 5-second 768p generation currently starts at 15 credits. Check the live estimate before generating.

Where are my MiniMax H3 videos saved?

Completed generations appear in Your Creations, where you can preview the video, review the audio and final frame, and download the finished result.

ABOUT THE MODEL

Looking for MiniMax H3 Model Details?

Explore the MiniMax H3 overview for model capabilities, real video examples, supported workflows, limitations, and comparisons.

Explore the MiniMax H3 Overview