AlignYourStepsScheduler - ComfyUI Built-in Node Documentation

The AlignYourStepsScheduler node generates sigma values for the denoising process based on different model types. It calculates appropriate noise levels for each step of the sampling process and adjusts the total number of steps according to the denoise parameter. This helps align the sampling steps with the specific requirements of different diffusion models.

Inputs

Parameter	Description	Data Type	Required	Range
`model_type`	Specifies the type of model to use for sigma calculation (default: “SD1”)	COMBO	Yes	`"SD1"` `"SDXL"` `"SVD"`
`steps`	The total number of sampling steps to generate (default: 10)	INT	Yes	1 to 10000
`denoise`	Controls how much to denoise the image, where 1.0 uses all steps and lower values use fewer steps (default: 1.0)	FLOAT	Yes	0.0 to 1.0

Note: Each model type has a built-in noise level schedule containing 11 sigma values (for 10 steps). When denoise is 0.0, the node returns an empty sigma tensor. When denoise is between 0.0 and 1.0, the effective number of steps is calculated as round(steps × denoise), and only the corresponding last portion of the sigma schedule is used. If the requested steps value does not match the built-in schedule length, the noise levels are log-linearly interpolated to match the requested step count. The final sigma value is always set to 0.

Outputs

Output Name	Description	Data Type
`sigmas`	Returns the calculated sigma values for the denoising process	SIGMAS

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): 3adbe1016c1ff4b9b7ad3737f50b168f54444d4ca355488e60537d1136f85d3f

​Inputs

​Outputs

Inputs

Outputs