Impact Pack

Comprehensive custom node suite for advanced features.

Installation

ComfyUI/custom_nodes/
  ComfyUI-Impact-Pack/

Install via ComfyUI Manager or:

git clone https://github.com/ltdrdata/ComfyUI-Impact-Pack.git

Core Nodes

DetectFaces

Detect faces in images.

Inputs

Input	Type	Description
image	image	Input image
face_detection	detection	Detection model

Outputs

Output	Type	Description
face	image	Cropped face
bbox	bbox	Bounding box
mask	mask	Face mask

Parameters

Parameter	Description
detection_model	Model to use
confidence	Detection threshold
max_faces	Maximum faces to detect

FaceDetailer

Enhance faces using specialized model.

Workflow

[Full Image] → [DetectFaces] → [FaceDetailer] → [Composite]
                                  ↓
                             Enhanced faces

Parameters

Parameter	Description
guide_size	Detection size
seed	Random seed
steps	Sampling steps
cfg	CFG scale
denoise	Denoise strength

Detailer

General image enhancement node.

Usage

1. Detect region of interest
2. Enhance with specified model
3. Composite back to original

Settings

Parameter	Default	Description
guide_size	512	Processing resolution
max_corrections	6	Max enhancement passes
core_padding	32	Edge buffer

BoundingBox Nodes

Crop bbox

Crop image using bounding box.

Expand bbox

Expand bounding box with padding.

Clamp bbox

Ensure box stays within image bounds.

Mask Nodes

Mask Detailer

Enhance masked areas.

Mask Composition

• Union
• Intersection
• Subtract

Feather Mask

Smooth mask edges.

Workflow Examples

Face Enhancement

LoadImage → DetectFaces → FaceDetailer → Composite
                ↓              ↓
           BoundingBox     Enhanced

Object Remove

LoadImage → Segment → Mask → Inpaint → Composite

Tips

1. Detection quality: Use appropriate detection model
2. Denoise balance: Lower for subtle, higher for major changes
3. Padding: Use enough to avoid edge artifacts

Next Steps

• ControlNet Nodes - Structure control
• LayerStyle - Visual effects
• Utils - Helper nodes