SamplerDPMAdaptative - ComfyUI Built-in Node Documentation

The SamplerDPMAdaptative node implements an adaptive DPM (Diffusion Probabilistic Model) sampler that automatically adjusts step sizes during the sampling process. It uses tolerance-based error control to determine optimal step sizes, balancing computational efficiency with sampling accuracy. This adaptive approach helps maintain quality while potentially reducing the number of steps needed.

Inputs

Parameter	Description	Data Type	Required	Range
`order`	The order of the sampler method (default: 3)	INT	Yes	2-3
`rtol`	Relative tolerance for error control (default: 0.05)	FLOAT	Yes	0.0-100.0
`atol`	Absolute tolerance for error control (default: 0.0078)	FLOAT	Yes	0.0-100.0
`h_init`	Initial step size (default: 0.05)	FLOAT	Yes	0.0-100.0
`pcoeff`	Proportional coefficient for step size control (default: 0.0)	FLOAT	Yes	0.0-100.0
`icoeff`	Integral coefficient for step size control (default: 1.0)	FLOAT	Yes	0.0-100.0
`dcoeff`	Derivative coefficient for step size control (default: 0.0)	FLOAT	Yes	0.0-100.0
`accept_safety`	Safety factor for step acceptance (default: 0.81)	FLOAT	Yes	0.0-100.0
`eta`	Stochasticity parameter (default: 0.0)	FLOAT	Yes	0.0-100.0
`s_noise`	Noise scaling factor (default: 1.0)	FLOAT	Yes	0.0-100.0

All inputs are advanced parameters used to fine-tune the adaptive sampling behavior. All numeric inputs allow decimal values and accept a minimum of 0.0 and a maximum of 100.0, except order, which is limited to the integer values 2 or 3.

Outputs

Output Name	Description	Data Type
`sampler`	Returns a configured DPM adaptive sampler instance	SAMPLER

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): 07b2e5b9f21ec101eabccc6be245d043e64a996a14db10434b03eaae0a91b1d8

​Inputs

​Outputs

Inputs

Outputs