This zip folder contains two example workflows for my Yolo model, Fdetailer. Extract the archive, then drag and drop either one of the json files into ComfyUI.
Both workflows have notes in them explaining various parts of the workflow. I've tried to cover the most basic features of Impact, but you can experiment a bit and add or remove things as you see fit. This isn't a comprehensive workflow for all of Impact's capabilities, but rather a simple workflow to get you up and running with minimal effort.
The simple workflow has a single detector and detailer node, showcasing the most basic way to use the model.
The advanced workflow contains two detector and detailer nodes, allowing you to have separate inpainting settings based on what objects are detected. This is my preferred and recommended way of using the model, so don't be turned off by the word "advanced" in the name. It's barely more complex than the simple version, I just wanted to show how you would use multiple detailer nodes to get more control over your final image.
Additionally, you can also check out the Impact Wildcard Documentation to see how the wildcard feature of Impact works, specifically the syntax related to the detailer node.
Description
FAQ
Comments (3)
Do you have one that does the same thing, but works with an already finished image?
Do you know if its possible to add prompts for each class?
Since theres 4 classes in your furry detection model for example.
And illd like to prompt for each. I know forge supports this, idk about comfy.
Yes. In the "wildcard spec" box inside the detailer node, you can specify a prompt for each class. Quick syntax example:
"[LAB]
[ALL] masterpiece, best quality,
[face] detailed eyes, blue eyes,"
Additional information can be found in the Impact Pack documentation: https://github.com/ltdrdata/ComfyUI-extension-tutorials/blob/Main/ComfyUI-Impact-Pack/tutorial/ImpactWildcard.md#special-syntax-for-detailer-wildcard