MiniMax-H3 Test: Generating Videos Using Only Reference Images and Audio

dilinjabass · reddit · 2026-08-04

A creator shared a video demonstration generated using the MiniMax-H3 model.

The creator noted that the project required only a few reference images and audio clips, along with some minor post-editing, highlighting the model's potential for multimedia-driven generation without complex prompting.

Related event: MiniMax Open-Sources Omni H3 Model, Stunning Multimodal Control & Generation in Tests(24 posts)→

Original post →

More from Multimodal

Multimodal channel →