Instructions to use lightx2v/Minimax-h3-Turbo with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use lightx2v/Minimax-h3-Turbo with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline from diffusers.utils import load_image, export_to_video # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("lightx2v/Minimax-h3-Turbo", dtype=torch.bfloat16, device_map="cuda") pipe.to("cuda") prompt = "A man with short gray hair plays a red electric guitar." image = load_image( "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/guitar-man.png" ) output = pipe(image=image, prompt=prompt).frames[0] export_to_video(output, "output.mp4") - Notebooks
- Google Colab
- Kaggle
Will this Lora work if I use 8 steps? And if so, should I set the Lora strength to 0.5 instead of 1?
Will this Lora work if I use 8 steps? And if so, should I set the Lora strength to 0.5 instead of 1?
Have you tried it? Can you share results to compare
Have you tried it? Can you share results to compare
In my case it doesn't work
Here it works fine with more then 4 steps. Here are a few video's to compare. I use this version in Comfy: https://huggingface.co/lightx2v/Minimax-h3-Turbo/discussions/3
Prompt:
[0s-3s] A shot from the front seat of a high-speed roller coaster in a theme park, featuring a massive steel loop directly ahead under a bright blue sky.
[3s-6s] The coaster plunges downward into a steep drop and rushes through a series of rapid turns.
[6s-10s] The track twists and coils in a continuous sequence of sharp turns, while the wind whips past and the distant landscape blurs by in a rhythmic motion.
Seed: 42
LoRA strength: 1.00
Steps: 4
LoRA strength: 0.50
Steps: 8
LoRA strength: 0.50
Steps: 10
If you are curious how it compares to the model without turbo LoRA and 20 steps: