This workflow is designed to allow multiple prompt sets to be kept in a single workflow and then switched between easily, automatically changing the ReActor face swap image (if switched on) and output filename prefix. A detailed User Guide is included in the zip file.
A full set of wildcards specifically designed for use with this workflow can be found here:
ComfyUI Wildcard Pack
This workflow has the following features:
Support for SDXL, Flux Basic (Schnell/FP8), Flux Full Fat (Dev/FP16) and GGUF (Lower VRAM) Models
Selectable or Random - Portrait / Landscape Aspect Ratio (automatically adjusts upscale to suit)
8 Selectable or Random Switchable input prompt sets, each with:
2 Positive wildcard prompt boxes
ReActor face swap image loader
Output filename prefix
8 Selectable or Random Image Style Prompts
Shared negative prompt
Lora loader (4 for SDXL, 1 for Flux)
Reactor face swapper (switchable)
Face Detailer (switchable, mainly for SDXL and Flux Basic)
Simple upscaler (fast)
Image cropper for removing text at the bottom of an image (X pixels)
Image save with full MetaData
Description
Added a PAG (Perturbed Attention Guidance) node, which adds extra detail to images, by changing the model attention. This node is in the model control group (the prompt select has moved to the top of this group now). You can change the PAG number, but don't have to. It's crazy great. Also changed the KSampler to the advanced one, you don't need to do anything extra with this, it's mainly incase it is needed in the future. Updated User Guide, including note about ReActor installation issues using standard comfyui downloads and a solution.