Skip to main content

Dual Model CFG Guider

This node allows you to use two different models during the guided CFG sampling process: one model for the positive (conditional) pass and a separate model for the negative (unconditional) pass. When no negative model is provided, it behaves like a standard CFG guider using a single model.

Inputs

ParameterDescriptionData TypeRequiredRange
modelModel used for the positive (conditional) pass.MODELYes
model_negativeModel used for the negative (unconditional) pass. Use the same model for ordinary CFG.MODELNo
positiveThe positive conditioning input.CONDITIONINGYes
cfgThe CFG scale value (default: 4.0).FLOATYes0.0 to 100.0 (step: 0.1)
negativeNegative conditioning run on the negative model. Leave unconnected for a text-free (image-only) unconditional pass.CONDITIONINGNo

Outputs

Output NameDescriptionData Type
GUIDERA guider object configured with the specified models and conditioning for use in sampling.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): a60803156e98d2ffe975d39922dfbeacafd1a2155d88dd2e285ac1426a1e7a33