Skip to main content
The CFGZeroStar node applies a specialized guidance scaling technique to diffusion models. It modifies the classifier-free guidance process by calculating an optimized scale factor based on the difference between conditional and unconditional predictions. This approach adjusts the final output to provide enhanced control over the generation process while maintaining model stability.

Inputs

ParameterDescriptionData TypeRequiredRange
modelThe diffusion model to be modified with the CFGZeroStar guidance scaling techniqueMODELYes-

Outputs

Output NameDescriptionData Type
patched_modelThe modified model with CFGZeroStar guidance scaling appliedMODEL
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): 4ec70f67eed8aaca964ce3dfb5b650af2f44d34bcde3b74c3886c91e1e8e8cc0