Instructions to use Qwen/Qwen-Image with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use Qwen/Qwen-Image with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("Qwen/Qwen-Image", 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
- Local Apps Settings
- Draw Things
- DiffusionBee
What range of resolutions is the model actually trained in, in terms of megapixels?
#51
by DiffusionFanatic1 - opened
Like, what are the lowest and highest officially supported resolutions? There seems to be no information on this beyond the (quite strange) resolutions listed in the model card.
It seems to support really high resolution, without getting repetitive objects or malformed anatomy like SDXL get.
By really high I mean 3172x3172 and more (though it is extremely slow, of course)
I tested resolutions between 768x768 and 2048x2048, and they performed well; resolutions lower or higher might not be as practical or could generate very slowly