EasyKSampler (Full)(samplerFull)
EasyKSampler (Full) is a flexible sampling node that combines the standard KSampler settings with the Easy-Use pipeline system.
sampler Full
Description
EasyKSampler (Full) is a flexible sampling node that combines the standard KSampler settings with the Easy-Use pipeline system. It reads a processing pipeline from pipe, applies the selected sampler using steps, cfg, sampler_name, scheduler, and denoise, and outputs the generated image plus all of the relevant components for further processing.
The optional inputs make this node especially useful when you want to override only part of the pipeline without rebuilding a full PIPE_LINE. The xyPlot input also allows it to work with XY plot workflows, and the image_output options give you direct control over previewing, saving, and sending the result image.
Inputs
Required
- pipe (
PIPE_LINE): The pipeline that supplies the model, conditioning, latent, VAE, and other sampling components. The node reads from this pipe unless an optional override input is connected. - steps (
INT, default: 20, min: 1, max: 10000): Number of sampling steps. - cfg (
FLOAT, default: 8, min: 0, max: 100): Classifier-free guidance strength. Higher values generally make the result follow the prompt more closely. - sampler_name (
COMBO): The sampling sampler to use. Available samplers depend on the current ComfyUI installation. - scheduler (
COMBO): The scheduler to use for the sampler. Available schedulers depend on the current ComfyUI installation. - denoise (
FLOAT, default: 1, min: 0, max: 1, step: 0.01): How much of the latent to denoise. A value of1performs full denoising. Lower values preserve more of the original latent structure and are common for image-to-image or refinement workflows. - image_output (
COMBO): Controls what happens with the generated image. Options areHide,Preview,Save,Hide&Save,Sender,Sender&Save, andNone. Save-related options usesave_prefix; Sender-related options uselink_id. - link_id (
INT): Required when using aSenderorSender&Saveimage output mode. It identifies the receiver to send the image to. - save_prefix (
STRING, default:ComfyUI): File name prefix used when saving the output image.
Optional
- seed (
INT): Seed for reproducible sampling. If connected, the same seed can be used to produce the same result. - model (
MODEL): Optional model override. When connected, this model is used instead of the model from the pipeline. - positive (
CONDITIONING): Optional positive conditioning override. - negative (
CONDITIONING): Optional negative conditioning override. - latent (
LATENT): Optional latent override. - vae (
VAE): Optional VAE override. This is used for decoding the sampled latent into an image. - clip (
CLIP): Optional CLIP override. - xyPlot (
XYPLOT): Optional XY plot input. Connect an XY plot definition to produce a grid or series of samples based on plot values. - image (
IMAGE): Optional image input. Connect it when a workflow needs to pass an image into the node.
Outputs
- PIPE_LINE: The updated pipeline, including the sampling results and the relevant components used during sampling.
- IMAGE: The generated image or images.
- MODEL: The model used for sampling.
- CONDITIONING: The positive conditioning used.
- CONDITIONING: The negative conditioning used.
- LATENT: The latent after sampling.
- VAE: The VAE used for decoding.
- CLIP: The CLIP model used.
- INT: The seed value used for the sampling run.
Usage Notes
- If an optional input is connected, it takes precedence over the value carried in
pipe. If left unconnected, the value from the pipeline is used. - To reproduce a result, connect the
INTseed output to downstream nodes that need the same seed, or feed the same seed back into theseedinput. - For partial denoising or image-to-image workflows, set
denoisebelow1.0. Lower values keep more of the original latent intact. - When using
SenderorSender&Save, make surelink_idmatches the corresponding receiver node in your workflow. - When saving images,
save_prefixis used as the base name for the output file.
Comments
Sign in with GitHub to join the discussion.