Skip to main content
This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub
The OptimalStepsScheduler node calculates noise schedule sigmas for diffusion models based on the selected model type and step configuration. It adjusts the total number of steps according to the denoise parameter and interpolates the noise levels to match the requested step count. The node returns a sequence of sigma values that determine the noise levels used during the diffusion sampling process.

Inputs

ParameterData TypeRequiredRangeDescription
model_typeCOMBOYes”FLUX"
"Wan"
"Chroma”
The type of diffusion model to use for noise level calculation
stepsINTYes3-1000The total number of sampling steps to calculate (default: 20)
denoiseFLOATNo0.0-1.0Controls the denoising strength, which adjusts the effective number of steps (default: 1.0)
Note: When denoise is set to less than 1.0, the node calculates the effective steps as steps * denoise. If denoise is set to 0.0, the node returns an empty tensor.

Outputs

Output NameData TypeDescription
sigmasSIGMASA sequence of sigma values representing the noise schedule for diffusion sampling