PerturbedAttentionGuidance - ComfyUI Built-in Node Documentation

The PerturbedAttentionGuidance node applies perturbed attention guidance to a diffusion model to enhance generation quality. It adjusts the model’s denoising process during sampling by comparing the normal conditional prediction with one made using a simplified attention mechanism that only uses value projections, then adds the scaled difference back to the result. When the scale is set to 0, the node has no effect.

Inputs

Parameter	Description	Data Type	Required	Range
`model`	The diffusion model to apply perturbed attention guidance to	MODEL	Yes	-
`scale`	The strength of the perturbed attention guidance effect (default: 3.0). When set to 0, the node has no effect and returns the original denoised result.	FLOAT	Yes	0.0 - 100.0 (step: 0.01)

Outputs

Output Name	Description	Data Type
`model`	The modified model with perturbed attention guidance applied	MODEL

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): 1cf824486ae695a9e563c70a4798aaf4c9c067ae3b53172c9767e3c5093d0096

​Inputs

​Outputs

Inputs

Outputs