Instructions to use REPA-E/e2e-qwenimage-vae with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use REPA-E/e2e-qwenimage-vae with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline from diffusers.utils import load_image # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("REPA-E/e2e-qwenimage-vae", dtype=torch.bfloat16, device_map="cuda") prompt = "Turn this cat into a dog" input_image = load_image("https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/cat.png") image = pipe(image=input_image, prompt=prompt).images[0] - Notebooks
- Google Colab
- Kaggle
Bright Color Noise Artifacts in REPA-E Qwen VAE Reconstruction
#4
by LemonBranny - opened
Thank you so much for this excellent work. I've been testing your e2e-qwenimage-vae and I've found that it provides significantly more and better details compared to the original official Qwen VAE. This is truly fantastic!
While running tests with a standard diffusers script, I noticed a minor issue: the reconstructed images exhibit small, bright color noise artifacts. They appear as pixel-sized purple and red dots/noise, particularly in certain areas of the image.To be more specific, this color noise seems to appear exclusively on dark surfaces that are rich in fine details."

