MiniMax-H3 Studio

VIDEO + SOUND · CREATIVE STUDIO
Model docs ↗
Connecting to service
Make the scene yours

Your vision. In motion.

Describe a scene, mix your references, and make it yours with LoRAs.

MiniMax-H3 · 24 FPS

01

VISUAL + SOUND
Ctrl / ⌘ + Enter to generate

02Reference media

OPTIONAL

Start from text alone, or use a first and last frame to guide the shot.

Add first frameDrop or choose an image
Add last frameDrop or choose an image

A new first frame automatically selects the closest supported aspect ratio. You can then change Canvas manually. Images are cropped to fit the selected canvas.

03Sampling & LoRAs

INDEPENDENT LAYERS
Your acceleration LoRA sets the recommended sampling steps.

Style & character LoRAs 0

Stack compatible H3 LoRAs for character or style, each with its own weight. These are separate from Turbo acceleration.

No style or character LoRAs added.

Use a Hugging Face repo ID or a .safetensors file URL. Files download on first use.

Shot settings

Frame alignment may slightly change the final duration.

More steps usually take longer.

Waiting for the service…

Preview

NO OUTPUT YET

Your next scene starts here

Start with a scene, a visual, or a sound. Your finished video and synchronized soundtrack will appear here.

Synchronized audioMixed referencesCustom LoRA
Ready to create

Queue status, progress, and the final parameters will appear here.

Prompts and media are processed by the configured conditioner. Media selections stay in this page; other settings are saved in this browser. Generation time depends on canvas, duration, steps, and GPU quota.