Instructions to use MiniMaxAI/MiniMax-H3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use MiniMaxAI/MiniMax-H3 with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("MiniMaxAI/MiniMax-H3", dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Inference
- Notebooks
- Google Colab
- Kaggle
Almost VR support
I have found that H3 can create 2 side by side views of an scene with some x-shift but it does not seem that it can create equidistant fisheye views using T2V.
Test prompt
"A woman is leaning against an handrail on a balcony in the Caribbean looking out over the beach with palm trees. She have long brown hair and a summer dress and is holding an cocktail in her hand. An light breeze is moving her dress and hair.
She takes a sip from her drink and enjoys the scenery.
First-person stereoscopic dual views captured simultaneously from natural human binocular vision mounted on a static tripod using parallel optical axes and identical field of view with strong equidistant fisheye distortion extending to the image edges."
Maybe someone else can find the "magic word" where H3 can generate an true VR video... that would be great :)