MiniMax H3 text to video + upscale

MiniMax H3 is an open AI video model that generates video with sound in one pass: voices, sound effects and ambience come out already synced to the picture. This version adds a 2x latent upscale pass, so you get the normal clip and a sharper upscaled one.

Great for:

• Higher-quality AI influencer videos
• Cartoon and animated shorts that need crisp detail
• Clips you want to post in better resolution

How it works

1. Write your scene prompt.
2. Run the base generation.
3. The upscale pass refines it and saves both versions with audio.

Make it yours with LoRAs

This workflow supports LoRAs. Grab character, style, anime or cartoon LoRAs from CivitAI or Hugging Face, drop them into the LoRA loader, and keep your AI character looking the same in every shot.