Instructions to use Qwen/Qwen-Image-Edit-2511 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use Qwen/Qwen-Image-Edit-2511 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("Qwen/Qwen-Image-Edit-2511", 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] - Inference
- Notebooks
- Google Colab
- Kaggle
Multi-image edit (3 refs): artifacts at true CFG, fine on Lightning — reference-content dependent
Setup
Qwen-Image-Edit-2511 BF16 via ComfyUI
TextEncodeQwenImageEditPlus with 3 reference images (face close-up + body front + body back)
Output 1024×1536 (non-square 2:3)
Sampler: res_3m + bong_tangent (RES4LYF)
Behavior
With full CFG (2.7, 33 steps): generation reliably breaks on some reference sets — mixed artifacts (identity drift, color/texture corruption, anatomy distortion). Same parameters and seeds produce clean output on other reference sets.
With Lightning 4-step (true_cfg=1): every reference set is clean.
Pattern
1 reference (face only) → always clean, both modes
3 references → clean on some characters, broken on others — content-dependent
All references are Z-Image Turbo outputs, same prompt structure, identical dimensions
Failing sets tend to contain high-frequency content (curly hair, darker skin texture); working sets tend to be lower-frequency (straight hair, lighter skin). To be clear: this is about the rendered references, not the character identity itself.
What I've tried (no fix, or partial only)
I don’t even remember what I tried, but I tried a lot of things that seemed possible, and none of them worked. The workflow is below.
Question
Is this a known interaction between multi-ref token packing and the true-CFG noise_pred * (cond_norm / noise_norm) rescale path? Specifically:
Does Qwen2.5-VL's 384² downscale produce per-token norm outliers on high-frequency reference content that get amplified across denoising steps once true CFG is active?
Is multi-image reference (3+ refs) currently only stable at distilled-CFG / Lightning, or is there a recommended setup for full CFG?

Požehnané utorkové ráno, moja milovaná záhradná žena.
Nech dnešný deň začne tvojím krásnym úsmevom, pretože práve ten robí moje rána krajšími.
Prajem ti pokoj v srdci, radosť v duši a veľa nežných okamihov. Keby som bol pri tebe, pripravil by som nám voňavú rannú kávu, vzal by som tvoju ruku do svojej a s úsmevom by som ti zašepkal:
„Dobré ráno, láska moja. Si ten najkrajší dôvod, prečo sa každý deň teším na nové ráno.“
Krásny utorok, môj poklad. Ľúbim ťa. ❤️