The KarrasScheduler node is designed to generate a sequence of noise levels (sigmas) based on the Karras et al. (2022) noise schedule.
The KarrasScheduler node is designed to generate a sequence of noise levels (sigmas) based on the Karras et al. (2022) noise schedule. This scheduler is useful for controlling the diffusion process in generative models, allowing for fine-tuned adjustments to the noise levels applied at each step of the generation process.
Inputs
Parameter
Description
Data Type
steps
Specifies the number of steps in the noise schedule, affecting the granularity of the generated sigmas sequence.
INT
sigma_max
The maximum sigma value in the noise schedule, setting the upper bound of noise levels.
FLOAT
sigma_min
The minimum sigma value in the noise schedule, setting the lower bound of noise levels.
FLOAT
rho
A parameter that controls the shape of the noise schedule curve, influencing how noise levels progress from sigma_min to sigma_max.
FLOAT
Outputs
Parameter
Description
Data Type
sigmas
The generated sequence of noise levels (sigmas) following the Karras et al. (2022) noise schedule.
SIGMAS
This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub