ComfyUI-Impact-Pack

Why we included this project

ComfyUI users who want to fix or refine details in generated images will find the building blocks here without having to hand-edit masks. The detector nodes locate faces and objects via bounding boxes and segmentation, and the detailer nodes re-inpaint just those regions at higher quality, so faces come out sharper instead of smeared. An iterative upscaler raises resolution in steps, and the pipe nodes let you chain detection, refinement, and upscaling into reusable workflows. SAM and SAM2 models are supported for segmentation and tracking, and you can apply ControlNet and IPAdapter to the detected regions. If you are putting together production image pipelines in ComfyUI where output quality and repeatability matter, this is one of the most widely used extension packs around.