ModelSamplingSD3 - ComfyUI Built-in Node Documentation - ComfyUI

This node applies Stable Diffusion 3 style sampling settings to a model. It makes a copy of the model and replaces its sampling method with a flow-based sampling configuration that uses the given shift value, which controls how the sampling distribution is shaped.

Inputs

Parameter	Description	Data Type	Required	Range
`model`	The input model to apply SD3 sampling parameters to	MODEL	Yes	-
`shift`	Controls the sampling shift parameter (default: 3.0)	FLOAT	Yes	0.0 - 100.0 (step: 0.01)

Note: The shift value is applied together with a fixed internal multiplier of 1000. If the original model has a noise scale setting, that value is carried over to the modified model. The original model is not changed; a cloned and patched copy is returned.

Outputs

Output Name	Description	Data Type
`model`	The modified model with SD3 sampling parameters 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): a77e38c2cebf6f21f841a953ec5c59096eaf60ffc205c24f34f635e54c5718cb