Detailed Explanation of ComfyUI Nodes

Learn about the CLIP Text Encode SDXL node in ComfyUI, which encodes text inputs using CLIP models specifically tailored for the SDXL architecture, converting textual descriptions into a format suitable for image generation or manipulation tasks.

Learn about the CLIP Text Encode SDXL Refiner node in ComfyUI, which refines the encoding of text inputs using CLIP models, enhancing the conditioning for generative tasks by incorporating aesthetic scores and dimensions.

Documentation for ComfyUI's CLIPTextEncodeHunyuanDiT node, including input/output types, methods, introduction to BERT and mT5-XL, and usage tips.

Learn about the Conditioning Set Timestep Range node in ComfyUI, which adjusts the temporal aspect of conditioning by setting a specific range of timesteps, allowing for more targeted and efficient generation.

Learn about the Conditioning Zero Out node in ComfyUI, which zeroes out specific elements within the conditioning data structure, effectively neutralizing their influence in subsequent processing steps.

Learn about the CLIP Loader node in ComfyUI, which is designed for loading CLIP models, supporting different types such as stable diffusion and stable cascade. It abstracts the complexities of loading and configuring CLIP models for use in various applications, providing a streamlined way to access these models with specific configurations.

Learn about the Load Checkpoint With Config (DEPRECATED) node in ComfyUI, which is designed for loading model checkpoints along with their configurations. It abstracts the complexities of loading and configuring model checkpoints for use in various applications, providing a streamlined way to access these models with specific configurations.

Learn about the Diffusers Loader node in ComfyUI, which is designed for loading models from the diffusers library, specifically handling the loading of UNet, CLIP, and VAE models based on provided model paths. It facilitates the integration of these models into the ComfyUI framework, enabling advanced functionalities such as text-to-image generation, image manipulation, and more.

Learn about the Dual CLIP Loader node in ComfyUI, which is designed for loading two CLIP models simultaneously, facilitating operations that require the integration or comparison of features from both models.

The QuadrupleCLIPLoader node is one of the core nodes of ComfyUI, initially added to support the HiDream I1 version.

Learn about the UNET Loader node in ComfyUI, which is designed for loading U-Net models by name, facilitating the use of pre-trained U-Net architectures within the system.

Learn about the Model Sampling Continuous EDM node in ComfyUI, which enhances a model's sampling capabilities by integrating continuous EDM (Energy-based Diffusion Models) sampling techniques. It allows for the dynamic adjustment of the noise levels within the model's sampling process, offering a more refined control over the generation quality and diversity.

Learn about the Model Sampling Discrete node in ComfyUI, which modifies a model's sampling behavior by applying a discrete sampling strategy. It allows for the selection of different sampling methods, such as epsilon, v_prediction, lcm, or x0, and optionally adjusts the model's noise reduction strategy based on the zero-shot noise ratio (zsnr) setting.

Learn about the Rescale CFG node in ComfyUI, which adjusts the conditioning and unconditioning scales of a model's output based on a specified multiplier, aiming to achieve a more balanced and controlled generation process. It operates by rescaling the model's output to modify the influence of conditioned and unconditioned components, thereby potentially enhancing the model's performance or output quality.

Learn about the Checkpoint Save node in ComfyUI, which is designed for saving the state of various model components, including models, CLIP, and VAE, into a checkpoint file. This functionality is crucial for preserving the training progress or configuration of models for later use or sharing.

Learn about the CLIPMerge Simple node in ComfyUI, which specializes in merging two CLIP models based on a specified ratio, effectively blending their characteristics. It selectively applies patches from one model to another, excluding specific components like position IDs and logit scale, to create a hybrid model that combines features from both source models.

Learn about the CLIPSave node in ComfyUI, which is designed for saving CLIP models along with additional information such as prompts and extra PNG metadata. It encapsulates the functionality to serialize and store the model's state, facilitating the preservation and sharing of model configurations and their associated creative prompts.

Learn about the ModelMergeAdd node in ComfyUI, which is designed for merging two models by adding key patches from one model to another. This process involves cloning the first model and then applying patches from the second model, allowing for the combination of features or behaviors from both models.

Learn about the ModelMergeBlocks node in ComfyUI, which is designed for advanced model merging operations, allowing for the integration of two models with customizable blending ratios for different parts of the models. This node facilitates the creation of hybrid models by selectively merging components from two source models based on specified parameters.

Learn about the ModelMergeSimple node in ComfyUI, which is designed for merging two models by blending their parameters based on a specified ratio. This node facilitates the creation of hybrid models that combine the strengths or characteristics of both input models.

Learn about the ModelMergeSubtract node in ComfyUI, which is designed for advanced model merging operations, specifically to subtract the parameters of one model from another based on a specified multiplier. It enables the customization of model behaviors by adjusting the influence of one model's parameters over another, facilitating the creation of new, hybrid models.

Learn about the VAESave node in ComfyUI, which is designed for saving VAE models along with their metadata, including prompts and additional PNG information, to a specified output directory. It encapsulates the functionality to serialize the model state and associated information into a file, facilitating the preservation and sharing of trained models.

Learn about the CLIPSetLastLayer node in ComfyUI, which is designed for modifying the behavior of a CLIP model by setting a specific layer as the last one to be executed. It allows for the customization of the depth of processing within the CLIP model, potentially affecting the model's output by limiting the amount of information processed.

Learn about the CLIPTextEncode node in ComfyUI, which is designed for encoding textual inputs using a CLIP model, transforming text into a form that can be utilized for conditioning in generative tasks. It abstracts the complexity of text tokenization and encoding, providing a streamlined interface for generating text-based conditioning vectors.

Learn about the CLIPVisionEncode node in ComfyUI, which is designed for encoding images using a CLIP vision model, transforming visual input into a format suitable for further processing or analysis. It abstracts the complexity of image encoding, offering a streamlined interface for converting images into encoded representations.

Learn about the ConditioningAverage node in ComfyUI, which is designed for blending two sets of conditioning data, applying a weighted average based on a specified strength. This process allows for the dynamic adjustment of conditioning influence, facilitating the fine-tuning of generated content or features.

Learn about the Conditioning(Combine) node in ComfyUI, which is designed for merging two sets of conditioning data, effectively combining their information. It provides a straightforward interface for integrating conditioning inputs, allowing for the dynamic adjustment of generated content or features.

Learn about the Conditioning(Concat) node in ComfyUI, which is designed for concatenating conditioning vectors, effectively merging the 'conditioning_from' vector into the 'conditioning_to' vector. It provides a straightforward interface for integrating conditioning inputs, allowing for the dynamic adjustment of generated content or features.

Learn about the Conditioning(SetArea) node in ComfyUI, which is designed for modifying the conditioning information by setting specific areas within the conditioning context. It allows for the precise spatial manipulation of conditioning elements, enabling targeted adjustments and enhancements based on specified dimensions and strength.

Learn about the Conditioning(SetAreaWithPercentage) node in ComfyUI, which is designed for adjusting the area of influence for conditioning elements based on percentage values. It allows for the specification of the area's dimensions and position as percentages of the total image size, alongside a strength parameter to modulate the intensity of the conditioning effect.

Learn about the Conditioning(SetAreaStrength) node in ComfyUI, which is designed for adjusting the strength of conditioning elements within a specified area. It allows for the specification of the area's dimensions and position as percentages of the total image size, alongside a strength parameter to modulate the intensity of the conditioning effect.

Learn about the StableZero123_Conditioning node in ComfyUI, which is designed for processing conditioning information specifically tailored for the StableZero123 model. It focuses on preparing the input in a specific format that is compatible and optimized for these models.

Learn about the StableZero123_Conditioning_Batched node in ComfyUI, which is designed for processing conditioning information in a batched manner specifically tailored for the StableZero123 model. It focuses on efficiently handling multiple sets of conditioning data simultaneously, optimizing the workflow for scenarios where batch processing is crucial.

Learn about the GLIGENTextBoxApply node in ComfyUI, which is designed for integrating text-based conditioning into a generative model's input, specifically by applying text box parameters and encoding them using a CLIP model. This process enriches the conditioning with spatial and textual information, facilitating more precise and context-aware generation.

Learn about the InpaintModelConditioning node in ComfyUI, which is designed for facilitating the conditioning process for inpainting models, enabling the integration and manipulation of various conditioning inputs to tailor the inpainting output. It encompasses a broad range of functionalities, from loading specific model checkpoints and applying style or control net models, to encoding and combining conditioning elements, thereby serving as a comprehensive tool for customizing inpainting tasks.

Learn about the StyleModelApply node in ComfyUI, which is designed for applying a style model to a given conditioning, enhancing or altering its style based on the output of a CLIP vision model. It integrates the style model's conditioning into the existing conditioning, allowing for a seamless blend of styles in the generation process.

Learn about the SD_4XUpscale_Conditioning node in ComfyUI, which is designed for enhancing the resolution of images through a 4x upscale process, incorporating conditioning elements to refine the output. It leverages diffusion techniques to upscale images while allowing for the adjustment of scale ratio and noise augmentation to fine-tune the enhancement process.

Learn about the SVD_img2vid_Conditioning node in ComfyUI, which is designed for generating conditioning data for video generation tasks, specifically tailored for use with SVD_img2vid models. It takes various inputs including initial images, video parameters, and a VAE model to produce conditioning data that can be used to guide the generation of video frames.

Learn about the WanFunControlToVideo node in ComfyUI, used to generate conditioning data for video generation tasks, specifically designed for use with Wan 2.1 Fun Control models. It accepts various inputs including initial images, video parameters, and a VAE model to generate conditioning data that can be used to guide video frame generation.

Learn about the EmptyImage node in ComfyUI, which is designed for generating blank images of specified dimensions and color. It allows for the creation of uniform color images that can serve as backgrounds or placeholders in various image processing tasks.

Learn about the ImageBatch node in ComfyUI, which is designed for combining two images into a single batch. If the dimensions of the images do not match, it automatically rescales the second image to match the first one's dimensions before combining them.

Learn about the ImageCompositeMasked node in ComfyUI, which is designed for compositing images, allowing for the overlay of a source image onto a destination image at specified coordinates, with optional resizing and masking.

Learn about the ImageInvert node in ComfyUI, which is designed for inverting the colors of an image, effectively transforming each pixel's color value to its complementary color on the color wheel. This operation is useful for creating negative images or for visual effects that require color inversion.

Learn about the ImagePadForOutpaint node in ComfyUI, which is designed for preparing images for the outpainting process by adding padding around them. It adjusts the image dimensions to ensure compatibility with outpainting algorithms, facilitating the generation of extended image areas beyond the original boundaries.

Learn about the LoadImage node in ComfyUI, which is designed to load and preprocess images from a specified path. It handles image formats with multiple frames, applies necessary transformations such as rotation based on EXIF data, normalizes pixel values, and optionally generates a mask for images with an alpha channel. This node is essential for preparing images for further processing or analysis within a pipeline.

Learn about the PreviewImage node in ComfyUI, which is designed for creating temporary preview images. It automatically generates a unique temporary file name for each image, compresses the image to a specified level, and saves it to a temporary directory. This functionality is particularly useful for generating previews of images during processing without affecting the original files.

Learn about the SaveImage node in ComfyUI, which is designed for saving images to disk. It handles the process of converting image data from tensors to a suitable image format, applying optional metadata, and writing the images to specified locations with configurable compression levels.

Learn about the EmptyLatentImage node in ComfyUI, which is designed to generate a blank latent space representation with specified dimensions and batch size. This node serves as a foundational step in generating or manipulating images in latent space, providing a starting point for further image synthesis or modification processes.

Learn about the LatentComposite node in ComfyUI, which is designed to blend or merge two latent representations into a single output. This process is essential for creating composite images or features by combining the characteristics of the input latents in a controlled manner.

Learn about the LatentCompositeMasked node in ComfyUI, which is designed to blend two latent representations together at specified coordinates, optionally using a mask for more controlled compositing. This node enables the creation of complex latent images by overlaying parts of one image onto another, with the ability to resize the source image for a perfect fit.

Learn about the LatentUpscale node in ComfyUI, which is designed to upscale latent representations of images. It allows for the adjustment of the output image's dimensions and the method of upscaling, providing flexibility in enhancing the resolution of latent images.

Learn about the LatentUpscaleBy node in ComfyUI, which is designed to upscale latent representations of images by a specified scale factor. This node allows for the adjustment of the scale factor and the method of upscaling, providing flexibility in enhancing the resolution of latent samples.

Learn about the VAEDecode node in ComfyUI, which is designed to decode latent representations into images using a specified Variational Autoencoder (VAE). It serves the purpose of generating images from compressed data representations, facilitating the reconstruction of images from their latent space encodings.

Learn about the VAEEncode node in ComfyUI, which is designed to encode images into a latent space representation using a specified Variational Autoencoder (VAE). It abstracts the complexity of the encoding process, providing a straightforward way to transform images into their latent representations.

Learn about the LatentAdd node in ComfyUI, which is designed for the addition of two latent representations. It facilitates the combination of features or characteristics encoded in these representations by performing element-wise addition.

Learn about the LatentBatchSeedBehavior node in ComfyUI, which is designed to modify the seed behavior of a batch of latent samples. It allows for either randomizing or fixing the seed across the batch, thereby influencing the generation process by either introducing variability or maintaining consistency in the generated outputs.

Learn about the LatentInterpolate node in ComfyUI, which is designed to perform interpolation between two sets of latent samples based on a specified ratio, blending the characteristics of both sets to produce a new, intermediate set of latent samples.

Learn about the LatentMultiply node in ComfyUI, which is designed to scale the latent representation of samples by a specified multiplier, allowing for fine-tuning of generated content or the exploration of variations within a given latent direction.

Learn about the LatentSubtract node in ComfyUI, which is designed for subtracting one latent representation from another. This operation can be used to manipulate or modify the characteristics of generative models' outputs by effectively removing features or attributes represented in one latent space from another.

Learn about the LatentBatch node in ComfyUI, which is designed to merge two sets of latent samples into a single batch, potentially resizing one set to match the dimensions of the other before concatenation. This operation facilitates the combination of different latent representations for further processing or generation tasks.

Learn about the LatentFromBatch node in ComfyUI, which is designed to extract a specific subset of latent samples from a given batch based on the specified batch index and length. It allows for selective processing of latent samples, facilitating operations on smaller segments of the batch for efficiency or targeted manipulation.

Learn about the RebatchLatents node in ComfyUI, which is designed to reorganize a batch of latent representations into a new batch configuration, based on a specified batch size. It ensures that the latent samples are grouped appropriately, handling variations in dimensions and sizes, to facilitate further processing or model inference.

Learn about the RepeatLatentBatch node in ComfyUI, which is designed to replicate a given batch of latent representations a specified number of times, potentially including additional data like noise masks and batch indices. This functionality is crucial for operations that require multiple instances of the same latent data, such as data augmentation or specific generative tasks.

Learn about the SetLatentNoiseMask node in ComfyUI, which is designed to apply a noise mask to a set of latent samples. It modifies the input samples by integrating a specified mask, thereby altering their noise characteristics.

Learn about the VAEEncodeForInpaint node in ComfyUI, which is designed for encoding images into a latent representation suitable for inpainting tasks, incorporating additional preprocessing steps to adjust the input image and mask for optimal encoding by the VAE model.

Learn about the LatentCrop node in ComfyUI, which is designed to perform cropping operations on latent representations of images. It allows for the specification of the crop dimensions and position, enabling targeted modifications of the latent space.

Learn about the LatentFlip node in ComfyUI, which is designed to manipulate latent representations by flipping them either vertically or horizontally. This operation allows for the transformation of the latent space, potentially uncovering new variations or perspectives within the data.

Learn about the LatentRotate node in ComfyUI, which is designed to rotate latent representations of images by specified angles. It abstracts the complexity of manipulating latent space to achieve rotation effects, enabling users to easily transform images in a generative model's latent space.

Learn about the CheckpointLoaderSimple node in ComfyUI, which is designed to load model checkpoints without the need for specifying a configuration. It simplifies the process of checkpoint loading by requiring only the checkpoint name, making it more accessible for users who may not be familiar with the configuration details.

Learn about the CLIPVisionLoader node in ComfyUI, which is designed to load CLIP Vision models from specified paths. It abstracts the complexities of locating and initializing CLIP Vision models, making them readily available for further processing or inference tasks.

Learn about the ControlNetLoader node in ComfyUI, which is designed to load ControlNet models from specified paths. It abstracts the complexities of locating and initializing ControlNet models, making them readily available for further processing or inference tasks.

Learn about the DiffControlNetLoader node in ComfyUI, which is designed to load differential control nets from specified paths. It abstracts the complexities of locating and initializing differential control nets, making them readily available for further processing or inference tasks.

Learn about the GLIGENLoader node in ComfyUI, which is designed to load GLIGEN models from specified paths. It abstracts the complexities of locating and initializing GLIGEN models, making them readily available for further processing or inference tasks.

Learn about the HypernetworkLoader node in ComfyUI, which is designed to load hypernetworks from specified paths. It abstracts the complexities of locating and initializing hypernetworks, making them readily available for further processing or inference tasks.

This node is designed to dynamically load and apply LoRA (Low-Rank Adaptation) adjustments to models and CLIP instances based on specified strengths and LoRA file names. It facilitates the customization of pre-trained models by applying fine-tuned adjustments without altering the original model weights directly, enabling more flexible and targeted model behavior modifications.

Learn about the LoraLoaderModelOnly node in ComfyUI, which is designed to load LoRA models without requiring a CLIP model, focusing on enhancing or modifying a given model based on LoRA parameters. It allows for the dynamic adjustment of the model's strength through LoRA parameters, facilitating fine-tuned control over the model's behavior.

Learn about the StyleModelLoader node in ComfyUI, which is designed to load style models from specified paths. It abstracts the complexities of locating and initializing style models, making them readily available for further processing or inference tasks.

Learn about the ImageOnlyCheckpointLoader node in ComfyUI, which is designed to load checkpoints specifically for image-based models within video generation workflows. It efficiently retrieves and configures the necessary components from a given checkpoint, focusing on image-related aspects of the model.

The CropMask node is designed to crop a specified area from a given mask. It allows users to define the region of interest by specifying coordinates and dimensions, effectively extracting a portion of the mask for further processing or analysis.

The FeatherMask node is designed to apply a feathering effect to the edges of a given mask, smoothly transitioning the mask's edges by adjusting their opacity based on specified distances from each edge. This creates a softer, more blended edge effect.

The GrowMask node is designed to modify the size of a given mask, either expanding or contracting it, while optionally applying a tapered effect to the corners. This functionality is crucial for dynamically adjusting mask boundaries in image processing tasks, allowing for more flexible and precise control over the area of interest.

The ImageColorToMask node is designed to convert a specified color in an image to a mask. It processes an image and a target color, generating a mask where the specified color is highlighted, facilitating operations like color-based segmentation or object isolation.

The ImageToMask node is designed to convert an image into a mask based on a specified color channel. It allows for the extraction of mask layers corresponding to the red, green, blue, or alpha channels of an image, facilitating operations that require channel-specific masking or processing.

The InvertMask node is designed to invert the values of a given mask, effectively flipping the masked and unmasked areas. This operation is fundamental in image processing tasks where the focus of interest needs to be switched between the foreground and the background.

The LoadImageMask node is designed to load images and their associated masks from a specified path, processing them to ensure compatibility with further image manipulation or analysis tasks. It focuses on handling various image formats and conditions, such as presence of an alpha channel for masks, and prepares the images and masks for downstream processing by converting them to a standardized format.

The MaskComposite node is designed to combine two mask inputs through a variety of operations such as addition, subtraction, and logical operations, to produce a new, modified mask. It abstractly handles the manipulation of mask data to achieve complex masking effects, serving as a crucial component in mask-based image editing and processing workflows.

The MaskToImage node is designed to convert a mask into an image format. This transformation allows for the visualization and further processing of masks as images, facilitating a bridge between mask-based operations and image-based applications.

None。

The PorterDuffImageComposite node is designed to perform image compositing using the Porter-Duff compositing operators. It allows for the combination of source and destination images according to various blending modes, enabling the creation of complex visual effects by manipulating image transparency and overlaying images in creative ways.

The SplitImageWithAlpha node is designed to separate the color and alpha components of an image. It processes an input image tensor, extracting the RGB channels as the color component and the alpha channel as the transparency component, facilitating operations that require manipulation of these distinct image aspects.

This node aims to provide a basic sampling mechanism to meet various application needs. It allows users to select and configure different sampling strategies to meet their specific requirements, thereby enhancing the adaptability and efficiency of the sampling process.

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.

The Sampler node is designed to provide a basic sampling mechanism for various applications. It enables users to select and configure different sampling strategies tailored to their specific needs, enhancing the adaptability and efficiency of the sampling process.

The SamplerCustom node is designed to provide a flexible and customizable sampling mechanism for various applications. It enables users to select and configure different sampling strategies tailored to their specific needs, enhancing the adaptability and efficiency of the sampling process.

The KSamplerSelect node is designed to select a specific sampler based on the provided sampler name. It abstracts the complexity of sampler selection, allowing users to easily switch between different sampling strategies for their tasks.

The SamplerDPMPP_2M_SDE node is designed to generate a sampler for the DPMPP_2M_SDE model, allowing for the creation of samples based on specified solver types, noise levels, and computational device preferences. It abstracts the complexities of sampler configuration, providing a streamlined interface for generating samples with customized settings.

The SamplerDPMPP_SDE node is designed to generate a sampler for the DPMPP_SDE model, allowing for the creation of samples based on specified solver types, noise levels, and computational device preferences. It abstracts the complexities of sampler configuration, providing a streamlined interface for generating samples with customized settings.

The BasicScheduler node is designed to compute a sequence of sigma values for diffusion models based on the provided scheduler, model, and denoising parameters. It dynamically adjusts the total number of steps based on the denoise factor to fine-tune the diffusion process.

The ExponentialScheduler node is designed to generate a sequence of sigma values following an exponential schedule for diffusion sampling processes. It provides a customizable approach to control the noise levels applied at each step of the diffusion process, allowing for fine-tuning of the sampling behavior.

The KarrasScheduler node is designed to generate a sequence of noise levels (sigmas) based on the Karras et al. (2022) noise schedule. This scheduler is useful for controlling the diffusion process in generative models, allowing for fine-tuned adjustments to the noise levels applied at each step of the generation process.

The PolyexponentialScheduler node is designed to generate a sequence of noise levels (sigmas) based on a polyexponential noise schedule. This schedule is a polynomial function in the logarithm of sigma, allowing for a flexible and customizable progression of noise levels throughout the diffusion process.

The SDTurboScheduler node is designed to generate a sequence of sigma values for image sampling, adjusting the sequence based on the denoise level and the number of steps specified. It leverages a specific model's sampling capabilities to produce these sigma values, which are crucial for controlling the denoising process during image generation.

The VPScheduler node is designed to generate a sequence of noise levels (sigmas) based on the Variance Preserving (VP) scheduling method. This sequence is crucial for guiding the denoising process in diffusion models, allowing for controlled generation of images or other data types.

The FlipSigmas node is designed to manipulate the sequence of sigma values used in diffusion models by reversing their order and ensuring the first value is non-zero if originally zero. This operation is crucial for adapting the noise levels in reverse order, facilitating the generation process in models that operate by gradually reducing noise from data.

The SplitSigmas node is designed to divide a sequence of sigma values into two parts based on a specified step. This functionality is crucial for operations that require different handling or processing of the initial and subsequent parts of the sigma sequence, enabling more flexible and targeted manipulation of these values.

The Note node is designed to provide a basic sampling mechanism for various applications. It enables users to select and configure different sampling strategies tailored to their specific needs, enhancing the adaptability and efficiency of the sampling process.

The Primitive node can recognize the type of input connected to it and provide input data accordingly. When this node is connected to different input types, it will change to different input states. It can be used to use a unified parameter among multiple different nodes, such as using the same seed in multiple Ksampler.

The Reroute node is designed to provide a basic sampling mechanism for various applications. It enables users to select and configure different sampling strategies tailored to their specific needs, enhancing the adaptability and efficiency of the sampling process.

The Terminal Log (Manager) node is primarily used to display the running information of ComfyUI in the terminal within the ComfyUI interface. To use it, you need to set the `mode` to **logging** mode. This will allow it to record corresponding log information during the image generation task. If the `mode` is set to **stop** mode, it will not record log information. When you access and use ComfyUI via remote connections or local area network connections, Terminal Log (Manager) node becomes particularly useful. It allows you to directly view error messages from the CMD within the ComfyUI interface, making it easier to understand the current status of ComfyUI's operation.