KSampler (Multi-Seed+)

Integrated sampling pipeline: Creates or encodes latent → multi-seed sampling → decodes to images. For denoise=1.0 uses width/height to create empty latent. For denoise<1.0 uses input_image.

By Pirog17000·Created about a year ago·Updated 9 months ago· 3

model
vae
positive
negative
input_image

IMAGE

◄denoise1.00►

◄steps20►

◄cfg8.0►

◄seed_count1►

◄seed0►

◄sampler_name▾►

◄scheduler▾►

◄width512►

◄height512►

◄noise_typevanilla►

◄injected_noise0.00►

◄vertical_splits1►

◄horizontal_splits1►

◄overlap64►

◄tile_supersampling1.0►

◄supersampling_min_resolution512►

◄supersampling_max_resolution2048►

◄dd_enabledfalse►

◄detail_amount0.10►

◄dd_start0.20►

◄dd_end0.80►

◄dd_bias0.50►

◄dd_exponent1.00►

◄dd_start_offset0.00►

◄dd_end_offset0.00►

◄dd_fade0.00►

◄dd_smoothtrue►

Categorypirog/sampling

Inputs (32)

Name	Type	Default	Description
model	MODEL		The model used for denoising.
vae	VAE		The VAE model used for encoding/decoding.
positive	CONDITIONING		The conditioning describing the attributes you want to include in the image.
negative	CONDITIONING		The conditioning describing the attributes you want to exclude from the image.
denoise	FLOAT	1.000–1	The amount of denoising applied. 1.0=new image, <1.0=img2img.
steps	INT	201–10000	The number of steps used in the denoising process.
cfg	FLOAT	8.00–100	Classifier-Free Guidance scale.
seed_count	INT	11–1000	The number of seeds to generate images with.
seed	INT	00–18446744073709550000	The starting random seed. It will be incremented for each image in the batch.
sampler_name	COMBO		The algorithm used when sampling.
scheduler	COMBO		The scheduler controls how noise is gradually removed.
width	INT	51216–16384	The width of the generated image in pixels (used when denoise=1.0).
height	INT	51216–16384	The height of the generated image in pixels (used when denoise=1.0).
noise_type	COMBO	vanilla	🎲 Noise Generation Method: • vanilla: Standard ComfyUI noise (torch.randn) - reliable baseline • spectral-diverse: Frequency-controlled noise with pink/blue/hybrid patterns - enhanced diversity • hierarchical: Multi-scale latent-aware noise with statistical modeling - maximum quality Advanced methods produce more diverse and potentially higher quality results.
injected_noise	FLOAT	0.000–1	Strength of noise injection for variation generation. 0.0=disabled, >0.0=blend base and variation noise.
vertical_splits	INT	11–8	Number of vertical splits (1 = no splitting, 2+ = process image in tiles)
horizontal_splits	INT	11–8	Number of horizontal splits (1 = no splitting, 2+ = process image in tiles)
overlap	INT	640–256	Fixed pixel overlap for tile borders (multiple of 8 recommended). Helps reduce seam artifacts between tiles.
tile_supersampling	FLOAT	1.01–4	Supersampling factor for each tile before processing. The tile is scaled by this amount, processed, and then scaled back down.
supersampling_min_resolution	INT	512256–4096	The minimum resolution for the longest side of a tile after supersampling. If smaller, the tile will be upscaled to this size.
supersampling_max_resolution	INT	2048256–8192	The maximum resolution for the longest side of a tile after supersampling. If larger, the tile will be downscaled to this size.
input_imageopt	IMAGE		Input image for img2img (used when denoise<1.0).
dd_enabledopt	BOOLEAN	false	—
detail_amountopt	FLOAT	0.10-5–5	Overall strength of the detail adjustment.
dd_startopt	FLOAT	0.200–1	Start of the adjustment curve as a fraction of total steps.
dd_endopt	FLOAT	0.800–1	End of the adjustment curve as a fraction of total steps.
dd_biasopt	FLOAT	0.500–1	Curve bias; >0.5 peaks later, <0.5 peaks earlier.
dd_exponentopt	FLOAT	1.000–10	Exponent for the curve shape.
dd_start_offsetopt	FLOAT	0.00-1–1	Adjustment multiplier before the curve starts.
dd_end_offsetopt	FLOAT	0.00-1–1	Adjustment multiplier after the curve ends.
dd_fadeopt	FLOAT	0.000–1	Fade the entire effect in or out.
dd_smoothopt	BOOLEAN	true	Apply smoothing to the curve.

Outputs (1)

Name	Type	Description
IMAGE	IMAGE	Generated images, one for each seed.