50 credits per video.
From nothing, from a photo, or from your own material.
No upload needed. Describe the scene, the camera move and the sound you want, and H3 composes all three together.
Bring a still image to life. Add a second image and it becomes the closing frame, so you control where the shot begins and where it lands.
Feed in several references and say what each one is for: this one for the talent, this one for the product, this one for the mood. They all land in the same shot.
It stopped being a slot machine and started taking direction.
The place, the action, the mood, the sound. H3 reads all of it as one instruction and composes the shot around it, so the range of what you can make is the range of what you can describe.
Write the clip as beats — what the camera does at the start, what it lands on by the end — instead of re-rolling a one-line prompt and hoping something usable falls out.
Drop in reference images and tell the model what each is for — your face, your product, your street — and they carry into the scene you wrote.
Vertical, and the sound is generated with the picture rather than added afterwards — so a clip comes out finished instead of silent and waiting on an edit.
A sentence is enough to start. Add a photo or reference images when you want more control.
Try MiniMax H3 →50 credits per video.