This SDXL focused Image2Image process utilizes your desired SDXL base model with the SDXL Refiner. To ease the process, all steps are automatically taken care of, all you have to do is change the denoising strength.
Please read the notes integrated in the flow.
About this version (V1.2)
Tidied up some logic nodes
Swapped the scale node with a "scale by side" node for easier automation
Changed default settings on FaceDetailer node
Description
Updated to UlralyticsDetectorProvider instead of MMDetDetectorProvider to follow Impact Pack updates.
Added Random Prompt option via boolean number selector.
Added multiple styles selectors. TWRI Styles will integrate your typed prompt into a style, WAS styles will append after your typed prompts.
FAQ
Comments (2)
Amazing workflow. I tried to activate Optional Post Processing, installed ComfyUI-Impact-Pack, connected image, but nodes are not processing anything. One node UltralyticsDetectorProvider has undefined model, is it missing models? where do I get them?
I just tried you workflow, thank you for this good work. I take this as my reference workflow for img2img. You may add the superscaler node as an optionnal scaling mode, since it is powerfull. And a dynamic prompt instead of a regular one, but that's a minor detail.