TCFG - ComfyUI Built-in Node Documentation

TCFG (Tangential Damping CFG) refines the unconditional (negative) predictions to better align with the conditional (positive) predictions during the sampling process. This technique improves output quality by applying tangential damping to the unconditional guidance, based on the research paper 2503.18137. The node modifies the model’s sampling behavior by adjusting how unconditional predictions are processed during classifier-free guidance.

Inputs

Parameter	Description	Data Type	Required	Range
`model`	The model to apply tangential damping CFG to	MODEL	Yes	-

Outputs

Output Name	Description	Data Type
`patched_model`	The modified model with tangential damping CFG 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): 6e0e1df09bdc29f14a4828ec54fd29f89926b9d8121bc8ba9f95ce066b899e95

SkipLayerGuidanceSD3 - ComfyUI Built-in Node Documentation

SetClipHooks - ComfyUI Built-in Node Documentation

⌘I

​Inputs

​Outputs

Inputs

Outputs