MiniMax H3: Video With Sound and Speech Made Together, Up to 15 Seconds and Up to 2K

MiniMax H3, from the company behind Hailuo, makes the picture and the sound, speech included, in the same pass. Up to 15 seconds, up to 2K, and a mode where you give it up to 9 images, 3 clips and 3 sounds. What it can do and how to work with it.

MiniMax is the company behind Hailuo, and H3 is its video engine. What sets it apart is that the sound, speech included, is made together with the picture in the same pass rather than added afterwards. Mind the name: H3 is the video engine, not M3, the same company's language model.

Three Ways to Work With H3

• From text: describe the scene, and the engine builds it with sound
• From an image: give a first frame, and a last frame too if you like
• From elements: up to 9 images, 3 clips and 3 audio files, up to 12 files in all, from which the engine builds one video

Length, Resolution and Ratio

Any whole number of seconds from 4 to 15. Resolution is 768P or 2K. In text mode you can choose a ratio: 21:9, 16:9, 4:3, 1:1, 3:4 or 9:16. Each run makes one video.

How to Get the Best Out of It

• Short dialogue. Too many words in one shot come out fast and rushed
• When you give a reference clip, write that it is for motion only. Otherwise the face may drift
• Write the prompt in English, or in Hebrew and it is translated before it reaches the engine
• Do not ask for Hebrew lettering inside the video: Hebrew titles and captions are added in the edit

A street musician in a rainy alley plays a slow tune on a saxophone, then looks up and says: "This one is for you." Neon reflections on the wet ground, handheld camera slowly pushing in

Questions and Answers

Is the sound really made together with the picture? Yes, in the same pass, speech included.

Does the video belong to me? Yes. What is made in your account belongs to you, for personal, business or commercial use.

MiniMax H3 on Cadabra AI

On Cadabra AI the MiniMax H3 tile sits in the Video tab, with the three modes, every length from 4 to 15 seconds, and 768P or 2K. It is on the same balance as the image, voice and music engines, so you can create a character, put them in a video and add Hebrew captions in the same place. The price is shown before every run, and a generation that fails is refunded automatically. And for anyone starting out: 10 free credits on sign-up, no credit card, at livinginmovie.com/en/start.

Where to go next

• 105 image codes, and here they run
• 61 codes that turn a picture into a clip
• More guides on the blog