Svdq_Cache-dit_QIE-2509 (Nunchaku+Cache-dit_Qwen-Image-Edit-2509)
This workflow will bring very significant speed improvements with minimal changes to generation quality.
1.Install the nodes in the workflow.
2.Run the wheel install node.
3.Download the models.
4.Update ComfyUI.
5.Enjoy!
Qwen-Image-Edit-2509_Multi-input_maskarea_controlnet:
This workflow is a bit late to the party. There are already tons of great 2509 workflows out there. But if you're still struggling to pick one, then you absolutely have to save this one! It's super easy to use and has all the features you need.
1. You can quickly switch between safetensors, GGUF.
2. It supports three images at once for input.
3. It supports masks and ControlNet.
4. You can quickly switch between manually entering the resolution or just using the image's resolution.
Qwen-image-edit:
etc.
Description
FAQ
Comments (6)
!!!!Set steps to 20 and CFG to 2.5 when not using Lightning 4-step Lora!!!!
Unable to location UniWorld-Qwen-Image-Edit-2509.safetensors LoRA. Can you provide a link?
It doesn’t really matter. https://huggingface.co/chestnutlzj/Edit-R1-Qwen-Image-Edit-2509/tree/main
@SamLiu Thank you. I was having issues. All my generations resulted in a black image. Turns out QWEN isn't friendly with SageAttention.
It gets stuck on ksampler, is this a lora or checkpoint?
tried for hours to get this to do ANYTHING right at all and nothing works