LatentApplyOperationCFG - ComfyUI Built-in Node Documentation

The LatentApplyOperationCFG node applies a latent operation to modify the conditioning guidance process in a model. It works by intercepting the conditioning outputs during the classifier-free guidance (CFG) sampling process and applying the specified operation to the latent representations before they are used for generation. When the sampler produces two conditioning outputs, the operation is applied to the difference between them, and the second output is then added back to the result.

Inputs

Parameter	Description	Data Type	Required	Range
`model`	The model to which the CFG operation will be applied	MODEL	Yes	-
`operation`	The latent operation to apply during the CFG sampling process	LATENT_OPERATION	Yes	-

Note: This node is marked as experimental. The operation is applied to the model’s conditioning outputs during the CFG sampling process. When two conditioning outputs are present, the operation is applied to the difference between the first and second output, and the second output is added back. When only one conditioning output is present, the operation is applied directly to it.

Outputs

Output Name	Description	Data Type
`model`	The modified model with the CFG operation applied to its sampling process	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): e383684a785878bfa4004c2fac78ae562d8e035fdfe081f8e4ebbb2c50161987

​Inputs

​Outputs

Inputs

Outputs