ComfyUI Nodes
K Sampler Advanced

KSampler (Advanced)

comfyUI节点-KSampler(Advanced) K采样器(高级)|ComfyUI组件节点


  • Class name: KSamplerAdvanced
  • Category: sampling
  • Output node: False

The KSamplerAdvanced node is designed to enhance the sampling process by providing advanced configurations and techniques. It aims to offer more sophisticated options for generating samples from a model, improving upon the basic KSampler functionalities.

Input types

ParameterComfy dtypeDescription
modelMODELSpecifies the model from which samples are to be generated, playing a crucial role in the sampling process.
add_noiseCOMBO[STRING]Determines whether noise should be added to the sampling process, affecting the diversity and quality of the generated samples.
noise_seedINTSets the seed for noise generation, ensuring reproducibility in the sampling process.
stepsINTDefines the number of steps to be taken in the sampling process, impacting the detail and quality of the output.
cfgFLOATControls the conditioning factor, influencing the direction and space of the sampling process.
sampler_nameCOMBO[STRING]Selects the specific sampler to be used, allowing for customization of the sampling technique.
schedulerCOMBO[STRING]Chooses the scheduler for controlling the sampling process, affecting the progression and quality of samples.
positiveCONDITIONINGSpecifies the positive conditioning to guide the sampling towards desired attributes.
negativeCONDITIONINGSpecifies the negative conditioning to steer the sampling away from certain attributes.
latent_imageLATENTProvides the initial latent image to be used in the sampling process, serving as a starting point.
start_at_stepINTDetermines the starting step of the sampling process, allowing for control over the sampling progression.
end_at_stepINTSets the ending step of the sampling process, defining the scope of the sampling.
return_with_leftover_noiseCOMBO[STRING]Indicates whether to return the sample with leftover noise, affecting the final output's appearance.

Output types

ParameterComfy dtypeDescription
latentLATENTThe output represents the latent image generated from the model, reflecting the applied configurations and techniques.