ComfyUI-Easy-Use

dynamicThresholdingFull

Patches a model with dynamic-thresholding CFG controls, including mimic scale, percentile threshold, scheduling modes and minima, channel separation, variability measure, and interpolation strength.

easy a1111Loader

Loads a checkpoint pipeline and encodes prompts with Automatic1111-compatible behavior, including CLIP skip, VAE, LoRA, ControlNet, resolution, and batch settings.

easy ab

Routes one arbitrary input to either output A or B according to a boolean and explicitly blocks execution of the unselected output.

easy anythingIndexSwitch

Lazily evaluates and returns the arbitrary-typed value selected by an index, leaving all unselected branches unevaluated.

easy anythingInversedSwitch

Routes one arbitrary input to the output slot selected by an index and places execution blockers on every other output.

easy applyBrushNet

Applies a selected BrushNet model to an Easy-Use pipeline using an image and mask, with dtype, scale, and sampling-interval controls. It updates model, conditioning, latent, and image state.

easy applyFooocusInpaint

Applies the Fooocus inpaint head and patch tensors to a model using the supplied inpaint latent, returning the patched model.

easy applyInpaint

Routes an image and mask through the selected inpaint mode, including normal VAE encoding, Fooocus, BrushNet, PowerPaint, or differential and noise-mask variants, and returns the updated pipeline.

easy applyPowerPaint

Applies PowerPaint inpainting or outpainting conditioning to an Easy-Use pipeline using its model, CLIP, image, and mask, with fitting, task, scale, timing, dtype, and memory controls.

easy batchAnything

Combines two arbitrary values using type-aware batching: tensors are resized and concatenated, latents are merged, scalar values become lists, and compatible containers are appended.

easy blocker

Passes an arbitrary value when continue is true and returns a ComfyUI execution blocker when false, preventing downstream execution.

easy boolean

Provides a configurable boolean value and returns it unchanged.

easy cascadeKSampler

Runs the simplified Stable Cascade path from a prepared pipeline, completes the later stage and decoding, and returns the updated pipe and image.

easy cascadeLoader

Loads Stable Cascade stage C, stage B, stage A VAE, and CLIP components, encodes prompts, and creates compressed stage latents for Cascade samplers.

easy ckptNames

Selects a checkpoint filename and returns that name for downstream loader or XY-plot configuration.

easy cleanGpuUsed

Forces the pack's GPU and model-memory cleanup routine, then passes the connected value through unchanged.

easy clearCacheAll

Clears all Easy-Use cache entries and passes the connected value through unchanged.

easy clearCacheKey

Removes the specified Easy-Use cache key and passes the connected value through unchanged.

easy comfyLoader

Loads a checkpoint-based pipeline using ComfyUI prompt encoding, with configurable VAE, CLIP skip, LoRA and ControlNet stacks, latent resolution, and batch size.

easy compare

Compares two values using a selected equality, inequality, ordering, or zero-test condition and outputs the result as a boolean.

easy conditioningIndexSwitch

Lazily evaluates and returns the conditioning input selected by an index, avoiding evaluation of unselected branches.

easy controlnetLoader

Applies one ControlNet image to a pipeline's conditioning with configurable strength and soft-weight scaling, returning the updated pipe and conditioning.

easy controlnetLoader++

Applies a ControlNet Union model to a pipeline and selects its union control type, with image, strength, timing, and soft-weight controls.

easy controlnetLoaderADV

Applies ControlNet conditioning to a pipeline with explicit start and end percentages in addition to image, strength, and soft-weight scaling.

easy controlnetNames

Selects a ControlNet filename and returns that name for downstream loader or XY-plot configuration.

easy controlnetStack

Builds or extends a ControlNet stack from up to three ControlNets, retaining each reference image, strength, start and end percentages, and soft-weight scale.

easy controlnetStackApply

Applies each configured ControlNet stack entry to a pipeline's positive and negative conditioning, including its image, strength, timing range, and soft-weight scale.

easy convertAnything

Converts an arbitrary input to a selected string, integer, float, or boolean type and returns the converted value.

easy detailerFix

Executes the detailer configuration stored in an Easy-Use pipeline to refine detected or masked regions. It returns the updated pipe, final image, and cropped refined or alpha-enhanced region outputs.

easy float

Provides a configurable floating-point value, rounded to three decimal places at output.

easy fluxLoader

Builds a Flux generation pipeline from checkpoint, VAE, prompt, latent size, batch, LoRA, and optional ControlNet settings, while accepting model, CLIP, and VAE overrides.

easy forLoopEnd

Closes a counted or directory-image loop, increments its index, compares it with the total, and feeds carried values into the next iteration until complete.

easy forLoopStart

Starts a counted loop, emitting the current index and carried values through the pack's while-loop graph machinery for the configured total iterations.

easy fullCascadeKSampler

Runs full Stable Cascade stage-C sampling, prepares stage-B latent and model state, and returns an updated pipeline plus stage-B inputs.

easy fullkSampler

Runs a complete configurable sampling pass from a pipeline or supplied component overrides, decodes and optionally saves or previews images, and returns the updated pipe plus generation components.

easy fullLoader

Builds a complete Easy-Use generation pipeline from checkpoint, config, VAE, CLIP skip, prompts, resolution, batch, LoRA, and ControlNet settings. It also exposes all major pipeline components.

easy globalSeed

Controls a workflow-wide seed value and update action, supporting increment, decrement, randomize, and fixed behaviors while exposing the last seed to frontend state.

easy hiresFix

Upscales an image with a selected upscale model, optionally resizes or crops the result, encodes it through a VAE, and writes the image and latent into an Easy-Use pipeline for high-resolution refinement.

easy humanSegmentation

Segments selected human parts from an image using the configured parsing method and confidence threshold. It returns the cropped or isolated image, mask, and bounding box.

easy hunyuanDiTLoader

Loads a HunyuanDiT checkpoint pipeline with VAE, positive and negative conditioning, resolution, batch size, LoRA stack, and optional ControlNet stack.

easy icLightApply

Patches a model with IC-Light foreground or background lighting conditioning derived from an input image. It can remove the background, synthesize directional or ambient lighting references, and returns the patched model and lighting image.

easy if

Legacy/deprecated conditional node that returns the if input when the arbitrary condition is truthy and the else input otherwise.

easy ifElse

Selects and outputs either the on-true value or the on-false value according to a boolean input. Its branch inputs are lazy, so only the selected branch is requested for evaluation.

easy imageBatchToImageList

Splits an image batch into a list of one-image tensors while preserving each frame's image shape.

easy imageChooser

Presents batch images for interactive selection and returns the chosen image subset. It can pause for a new choice every run or reuse the last selection, and supports rescaled previews.

easy imageColorMatch

Matches a target image's color distribution to a reference using the selected transfer method, then previews, saves, or hides the corrected result.

easy imageConcat

Concatenates two image batches horizontally or vertically, resizing the second image according to the selected size-matching policy before joining them.

easy imageCount

Returns the number of images in the input image batch.

easy imageCropFromMask

Finds mask bounding boxes, smooths their size and center across a batch, expands them by configurable multipliers, and returns image crops, mask crops, and bounding-box metadata.

easy imageDetailTransfer

Transfers image detail from a source to a target using the selected frequency or blending mode, blur radius, and blend factor. An optional mask limits where detail is applied.

easy imageHSVMask

Compact description unavailable for this node.

easy imageIndexSwitch

Lazily evaluates and returns the image input selected by an index, avoiding evaluation of unselected image branches.

easy imageInsetCrop

Crops an image inward from its four edges using pixel or percentage measurements. Crop offsets are snapped to eight-pixel boundaries and invalid bounds are rejected.

easy imageInterrogator

Runs the selected image-interrogation backend on an input image and returns generated prompt text, with an option to use the lower-memory execution path.

easy imageListToImageBatch

Combines an input list of image tensors into one batch, resizing mismatched items to the first image's spatial dimensions before concatenation.

easy imagePixelPerfect

Calculates a pixel-perfect preprocessing resolution from the input dimensions and selected resize, inner-fit, or outer-fit rule, and displays the source dimensions with the result.

easy imageRatio

Reduces an image's width and height by their greatest common factor and returns the aspect ratio as integer and float pairs.

easy imageRemBg

Removes image backgrounds with a selected supported segmentation model and returns the foreground image plus mask. It can refine foreground colors, add a background, and preview or save the result.

easy imageSave

Saves an image batch with ComfyUI's normal image saver or routes it through preview-only output, using a configurable filename prefix.

easy imageScaleDown

Resizes every image in a batch to a specified width and height using Lanczos resampling. When center-crop mode is selected, it first crops the batch to the target aspect ratio.

easy imageScaleDownBy

Resizes an image batch by a fractional scale factor while preserving aspect ratio, using the pack's Lanczos scaling path.

easy imageScaleDownToSize

Scales an image batch down so either its longest or shortest side reaches the requested size. It preserves aspect ratio and never enlarges the source.

easy imageScaleToNormPixels

Scales an image by a multiplier using a selected ComfyUI resize method, rounding the resulting width and height down to multiples of eight.

easy imagesCountInDirectory

Counts supported JPG, JPEG, PNG, and WebP files in a directory for an image-loading loop, accounting for the starting index or explicit limit.

easy imageSize

Reads an image tensor's width and height, returns both dimensions as integers, and displays them in the node UI.

easy imageSizeByLongerSide

Returns the longer of an input image's width and height as a single resolution value.

easy imageSizeBySide

Returns either the longest or shortest pixel dimension of an input image, according to the selected side mode.

easy imageSplitGrid

Divides each input image into a requested row-by-column grid and returns all grid cells as one image batch.

easy imageSplitList

Distributes a batch into three interleaved groups when its length is divisible by three, or two interleaved groups when divisible by two.

easy imageSplitTiles

Cuts an image into an overlapping tile grid and returns tiles, per-tile masks, overlap metadata, and total tile count for indexed processing or reconstruction.

easy imagesSplitImage

Exposes individual images from a batch through separate image outputs so the first batch members can be wired independently.

easy imageSwitch

Returns image A when its boolean is true and image B when false.

easy imageTilesFromBatch

Selects one tile and its mask from a tiled batch by index and returns the tile plus its recorded x and y placement coordinates.

easy imageToBase64

Encodes an input image as PNG bytes and returns the resulting base64 string.

easy imageToMask

Legacy/deprecated converter that extracts the selected red, green, or blue channel from an image and returns it as a grayscale mask.

easy imageUncropFromBBOX

Resizes processed crops back into their recorded bounding boxes and blends them over the original images. It can create a softened rectangular blend mask or use a supplied mask.

easy indexAnything

Selects an indexed element from an arbitrary list, tensor batch, or container, supporting negative indices and clamping out-of-range values.

easy injectNoiseToLatent

Legacy/deprecated latent utility that mixes supplied noise into an image-encoded or existing latent, with strength, averaging, normalization, mask, and additional seeded-random-noise controls.

easy instantIDApply

Applies InstantID face identity conditioning to an Easy-Use pipeline using a face image, InsightFace provider, InstantID model, and ControlNet. It updates the model and conditioning and supports keypoint images, masks, strength, noise, and timing controls.

easy instantIDApplyADV

Runs the InstantID pipeline path while allowing explicit positive and negative conditioning to override the pipe's conditioning. It retains the face image, ControlNet, keypoint, mask, strength, noise, and timing controls.

easy int

Supplies a configurable integer value to the workflow, passing that value unchanged to an integer output.

easy ipadapterApply

Loads an IP-Adapter preset and applies an image reference to a model over a configurable sampling interval. It supports attention masks, tiled application, FaceID presets, and Flux or SD3-specific paths.

easy ipadapterApplyADV

Applies IP-Adapter image conditioning with advanced control over weight curves, embedding combination and scaling, negative images, masks, layer weights, batching, tiling, and sharpening.

easy ipadapterApplyEmbeds

Applies precomputed positive and optional negative IP-Adapter embeddings to a model, with controls for weight behavior, sampling interval, embedding scaling, and an attention mask.

easy ipadapterApplyEncoder

Encodes up to four weighted and optionally masked reference images into positive and negative IP-Adapter embeddings. It combines new and existing embeddings by concatenation, arithmetic, averaging, normalization, minimum, or maximum.

easy ipadapterApplyFaceIDKolors

Specializes the advanced IP-Adapter path for the FACEID PLUS KOLORS preset, exposing separate general, FaceID-v2, and Kolors weights plus batching, tiling, mask, and timing controls.

easy ipadapterApplyFromParams

Loads an IP-Adapter preset and applies an accumulated IPADAPTER_PARAMS collection to a model. It controls how regional embeddings are combined and scaled and accepts an optional negative image or loaded adapter.

easy ipadapterApplyRegional

Builds regional IP-Adapter parameters from a reference image, mask, image weight, and regional prompt text. It returns the region's conditioning and can append to an existing parameter set.

easy ipadapterStyleComposition

Applies IP-Adapter style and optional composition references with independent weights. It supports embedding-combination, attention-mask, negative-image, timing, and scaling controls.

easy isFileExist

Builds a path from directory, filename, and optional extension and returns whether it identifies an existing regular file.

easy isMaskEmpty

Returns true when a mask is missing or every mask value is zero, and false otherwise.

easy isNone

Tests absence by the pack's rules: None, an empty string, or numeric zero all return true.

easy isSDXL

Checks the CLIP supplied directly or through a pipeline and returns whether its conditioning model is an SDXL CLIP implementation.

easy joinImageBatch

Turns an image batch into one big image.

easy joyCaption2API

Sends an image to the BizyAIR JoyCaption2 endpoint and returns the generated caption. It supports caption styles and lengths, sampling controls, custom instructions, and downscales oversized inputs before submission.

easy joyCaption3API

Uses the JoyCaption3 BizyAIR endpoint with the same image-caption controls and preprocessing as the JoyCaption2 node, returning the service's generated caption text.

easy kolorsLoader

Builds a Kolors pipeline from UNet, VAE, and ChatGLM3 components, with prompt encoding, latent size, batch, LoRA stack, model overrides, and optional GPU cleanup.

easy kSampler

Runs the standard sampler using settings stored in the pipeline, decodes the latent, handles image output, and returns the updated pipe and image.

easy kSamplerCustom

Runs the pipeline's custom guider, sampler, noise, sigma, and latent objects and returns sampled and denoised outputs with the decoded image and pipe.

easy kSamplerDownscaleUnet

Temporarily patches selected UNet blocks for downscaled sampling over a configured step range, runs the sampler, and returns the updated pipe and image.