Nodes/ComfyUI_SmartModelLoader/Eclipse KSampler (Pipe)
ComfyUI Node

Eclipse KSampler (Pipe)

Eclipse KSampler using context pipe (PIPE). Performs sampling, VAE decoding, and updates the pipe with the final latent, image, and (optionally) sampler settings.

By r-vage·Created 9 days ago·Updated 2 days ago· 2
Eclipse KSampler (Pipe)
  • pipe
  • positive
  • negative
  • latent
  • image
  • pipe
  • latent
  • image
allow_overwritefalse
steps8
cfg1.0
sampler_nameres_multistep
schedulersimple
denoise1.00
tiled_decodefalse
tile_size512
preview_modePreview
seed42
Category🌒 Smart Model Loader/ Sampler

Inputs (15)

NameTypeDefaultDescription
pipePIPEThe pipe dictionary from Smart Model Loader or other compatible pipe nodes.
allow_overwriteBOOLEANfalseWhen enabled, allows values from the pipe to take priority over/overwrite local widget settings.
positiveCONDITIONINGThe positive conditioning (required, e.g. text prompt).
negativeCONDITIONINGThe negative conditioning (required, e.g. text prompt).
stepsINT81–10000The number of steps used in the denoising process.
cfgFLOAT1.00–100The Classifier-Free Guidance scale.
sampler_nameCOMBOres_multistepThe sampling algorithm.
schedulerCOMBOsimpleThe scheduler algorithm.
denoiseFLOAT1.000–1The amount of denoising applied.
tiled_decodeBOOLEANfalseEnable tiled VAE decoding to save VRAM on large images.
tile_sizeINT51264–4096The size of the tiles used for tiled VAE decoding.
preview_modeCOMBOPreviewShow the step-by-step rendering process during sampling and display the final decoded image at the end (Preview), or hide both (None) to keep the node layout clean.
seedINT42-3–18446744073709550000The random seed used for creating the noise. Use -1 for random, -2 to increment, -3 to decrement.
latentoptLATENTOptional input latent to denoise. Either this or 'image' must be connected/provided in the pipe.
imageoptIMAGEOptional input image to VAE-encode and denoise. Either this or 'latent' must be connected/provided in the pipe.

Outputs (3)

NameTypeDescription
pipePIPEThe updated pipe dictionary containing the model, clip, vae, latent, image, and sampler settings.
latentLATENTThe denoised latent.
imageIMAGEThe decoded image.