AlignYourStepsScheduler

Documentation

• Class name: AlignYourStepsScheduler
• Category: sampling/custom_sampling/schedulers
• Output node: False

The AlignYourStepsScheduler node is designed to adjust the noise levels (sigmas) for a given model type over a specified number of steps, ensuring that the diffusion process is aligned with the model's requirements. It dynamically interpolates or selects predefined noise levels to match the step count, facilitating a tailored diffusion process.

Input types

Required

• model_type
  • Specifies the model type for which the noise levels are to be adjusted, affecting the selection or interpolation of noise levels.
  • Comfy dtype: COMBO[STRING]
  • Python dtype: List[str]
• steps
  • Determines the number of steps for the diffusion process, influencing the interpolation or selection of noise levels.
  • Comfy dtype: INT
  • Python dtype: int

Output types

• sigmas
  • Comfy dtype: SIGMAS
  • A tensor of noise levels (sigmas) adjusted to align with the specified steps and model type.
  • Python dtype: torch.FloatTensor

Usage tips

• Infra type: GPU
• Common nodes: unknown

Source code

class AlignYourStepsScheduler:
    @classmethod
    def INPUT_TYPES(s):
        return {"required":
                    {"model_type": (["SD1", "SDXL", "SVD"], ),
                     "steps": ("INT", {"default": 10, "min": 10, "max": 10000}),
                      }
               }
    RETURN_TYPES = ("SIGMAS",)
    CATEGORY = "sampling/custom_sampling/schedulers"

    FUNCTION = "get_sigmas"

    def get_sigmas(self, model_type, steps):
        sigmas = NOISE_LEVELS[model_type][:]
        if (steps + 1) != len(sigmas):
            sigmas = loglinear_interp(sigmas, steps + 1)

        sigmas[-1] = 0
        return (torch.FloatTensor(sigmas), )