MiniMax · video
MiniMax H3
MiniMax H3: native audio, frames, an audio track to follow and references (5 images, 3 videos, 3 audio clips), up to 4K.
- Length
- 5 to 15 s
- Resolution
- 480p, 768p, 2K, 4K
- Aspect ratio
- 16:9, 9:16, 1:1, 4:3, 3:4, 21:9
- References
- Up to 11 audio or image or video files
About MiniMax H3
MiniMax H3 is the base model of MiniMax's H3 family, priced at $0.13 per second. It runs as text to video, image to video, start and end frame and references to video, with native audio, from 5 seconds to 15 seconds, at 480p, 768p, 2K and 4K and 16:9, 9:16, 1:1, 4:3, 3:4 and 21:9.
Two things set it apart in the H3 family here. It is the only H3 with a reference mode, which mixes images, video clips and audio clips in a single request. And every other mode accepts an optional audio file that becomes the clip's soundtrack. A typical generation takes about 3 min.
Good for
- Shots built from mixed references, images for the subject and style, video clips for the motion and audio clips for the sound
- Putting your own soundtrack on a generated clip
- Delivering at 4K, the highest setting of the MiniMax models here
Getting good results
- In reference mode, cite each file by its type and order, as the provider's format does: Image, Video or Audio followed by its number.
- An audio file added to text, image or frame mode replaces the generated soundtrack; a longer file is trimmed to the video, keeping the beginning.
- Only 480p and the step above it are rendered natively, according to the provider; the two higher settings upscale that base render. Compare $0.75 at the default resolution with $2.00 at 4K.
MiniMax H3 questions
What can the reference mode take?
Images, short video clips and audio clips together, 11 files at most, with a separate cap for each kind.
How is the price set?
Per second, by resolution: $0.75 for the default clip, $2.25 at 15 seconds.