Instructions to use Qwen/Qwen-Image-2.1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use Qwen/Qwen-Image-2.1 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-2.1", 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
Diamond grid pattern caused by VAE
yep, there are some mitigations, GLSL node, a nodepack or re-encode to flux
https://www.reddit.com/r/StableDiffusion/comments/1wlv3tf/qwen_image_21_noisepatterningmoire_mild_workaround/
How can we determine whether this is caused by VAE or by the DIT part?
How can we determine whether this is caused by VAE or by the DIT part?
It's very well known qwen vae causes this. Even from the old 2511 qwen image
You can test by [load image > vae encode > vae decode > save image]. No other processing
So it's either save image node causes it (lol no), or the Qwen VAE does π
How can we determine whether this is caused by VAE or by the DIT part?
vae, it happens with krea too (it used qwen vae)
How can we determine whether this is caused by VAE or by the DIT part?
vae, it happens with krea too (it used qwen vae)
When I used some of the trained krea2, there were parallel stripes with an interval of 16px. However, it was clearly caused by the DiT part. Because Gemini helped me write a latent nyquist fix, it could be fixed.
