Instructions to use microsoft/Mage-Flow with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use microsoft/Mage-Flow with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("microsoft/Mage-Flow", torch_dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- Draw Things
- DiffusionBee
Quality issues
Using the spaces at https://huggingface.co/spaces/hugging-apps/mage-flow-base with my standard test prompt
Full body photo of a young woman with long straight black hair, blue eyes and freckles wearing a corset, tight jeans and boots standing in the garden
I get only a white square. As this is a perfectly SFW prompt, there are people like that walking at the street in public, this is clearly overblocking.
Reducing that test to
Full body photo of a woman with long straight black hair, blue eyes and freckles wearing jeans and boots standing in the garden
I do get images. But the quality of those is unusable.
1024x1024:
2048x2048:
Is there a bug in the Hugging Face spaces demo? Or is this the real quality of the Mage-Flow model?
They look perfectly cromulent to me.

