Instructions to use Comfy-Org/Qwen-Image-2.1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusion Single File
How to use Comfy-Org/Qwen-Image-2.1 with Diffusion Single File:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
Try this setting, the image quality is insane!
#7
by hanswurstwerner - opened
I have tried a lot and with these settings I have such clean and clear results that it is just insane and you maintain a large part of the speed.
8-12 steps with cfg 1.5 - 3.5 and the remaining steps with cfg1
2.5 MP
If the details get too strong, the cfg will go down a bit cfg 1.5-3.5
dpm++_2m_sde with simple is also great and sometimes has even more details, you have to try it out.
The standard workflows all use cfg1, which I just noticed is that with cfg1 the prompt follow and the quality of the images suffer a lot. This method combines quality with good prompt follow and good speed.
Other Settings
But, how good is it compared to cfg 1.5-3 cfg native?
@Rkss Hi, I have tested both and have the better results so far with 2 samplers, but natively it also works very well with er_sde + simple cfg 2.5-3.5 for me
hanswurstwerner changed discussion status to closed





