PerpNegGuider - ComfyUI Built-in Node Documentation

The PerpNegGuider node creates a guidance system for controlling image generation using perpendicular negative conditioning. It takes positive, negative, and empty conditioning inputs and applies a specialized guidance algorithm that computes all three noise predictions in a single batch for efficiency. This node is designed for experimental testing and provides fine control over the guidance strength and negative scaling.

Inputs

Parameter	Description	Data Type	Required	Range
`model`	The model to use for guidance generation	MODEL	Yes	-
`positive`	The positive conditioning that guides the generation toward desired content	CONDITIONING	Yes	-
`negative`	The negative conditioning that guides the generation away from unwanted content	CONDITIONING	Yes	-
`empty_conditioning`	The empty or neutral conditioning used as a baseline reference for perpendicular negative guidance	CONDITIONING	Yes	-
`cfg`	The classifier-free guidance scale that controls how strongly the conditioning influences the generation (default: 8.0)	FLOAT	Yes	0.0 - 100.0
`neg_scale`	The negative scaling factor that adjusts the strength of the perpendicular negative effect (default: 1.0)	FLOAT	Yes	0.0 - 100.0

Outputs

Output Name	Description	Data Type
`guider`	A configured guidance system ready for use in the generation pipeline	GUIDER

This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

Source fingerprint (SHA-256): 02e4b4e0ea413da31c280cb1e510bdf5caa7b5187eb6de94632e45a91e881820

​Inputs

​Outputs

Inputs

Outputs