ComfyUI Custom Extensions
Browse 9,446 ComfyUI extensions and 73,215 custom nodes.
Make CLIPs / text encoders load on GPU device
This custom node resizes an image while forcing the final width and height to be multiples of a chosen value.
A custom node for ComfyUI providing high-quality preview sampling capabilities for image generation workflows. (Description by CC)
Custom utility nodes for ComfyUI by huyl2 (sort list v2, etc.)
Load upscale models directly from Hugging Face Hub in ComfyUI
Custom node pack that reduces repeated prompt assembly, reference image placement, and canvas normalization when creating game SD/chibi character assets in ComfyUI. (Description by CC)
A custom node for [a/OmniGen](https://github.com/VectorSpaceLab/OmniGen).
Three custom nodes for ComfyUI that allow you to encrypt and decrypt Python objects or text using simple XOR encryption.
Simple Custom Node To display two images and compare them
ComfyUI custom node for local translation using Tencent Hunyuan and HY-MT models.
A standalone LLM widget node for ComfyUI: ask a local CLIP/GGUF or an OpenAI/Gemini-compatible API a question about a connected image or video, without running the graph.
Loads an animated GIF/APNG/WEBP as a VIDEO stream, loops any VIDEO with variable playback speed, loads a frame range of a video file with a crop and a loop-point finder, and saves a VIDEO or image batch back out as an animated GIF (with a full set of color quantizers and dithering filters) or an animated WebP
This extension offers automatic drums extraction from audio files, as well as a few helper nodes to support my audio synchronization AnimateDiff workflows.
Utility nodes for ComfyUI including OpenAI Compatible Chat - Works with any OpenAI-compatible API endpoint including Lemonade, Ollama, OpenAI API, and LM Studio
Use to create masks in the middle of a workflow for image generation and inpainting in a single execution. Supports export to Krita and fetching once finished. This is a modified version of https://github.com/Cordux/ComfyUI-ImagePreviewPause
ComfyUI implementation of 'FitDiTx: Advancing the Authentic Garment Details for High-fidelity Virtual Try-on'
用ComfyUI玩小游戏合集的节点
A ComfyUI custom node that provides dynamic text substitution using wildcards and CSV files. Perfect for creating varied prompts with consistent relationships between terms.
Robust ComfyUI custom node for automated image-to-video pipelines with VRAM-aware memory management and intelligent fallback handling.
Smart tag filtering node for ComfyUI workflows with semantic category support for Danbooru/E621 tags.
A ComfyUI character prompt builder with high ethnicity adherence for Krea 2, conflict-aware randomization, and Identity Forge compatibility.
Bridge nodes for ComfyUI that send messages through WhatsApp (Twilio) and Discord when a workflow completes.
ComfyUI node to save, organize and reuse text presets with optional preview images.
A variety of random text encoder tools intended for use with ComfyUI and Qwen Image/Qwen Image Edit. More (may) be added as I try out various modifications to Qwen Image.
A conditioning node for ablated versions of Lumina2
A ComfyUI custom node for combining and stitching multiple images with flexible layouts including horizontal, vertical, and 2x2 grid arrangements.
A collection of pause-and-edit nodes for ComfyUI that let you interrupt workflow execution, inspect and modify values, then continue.
A custom node for ComfyUI that allows you to save generated images directly to your local computer via browser download, ensuring zero file footprint on the server's storage.
Video description nodes using Qwen3-VL and NVIDIA DAM models, plus StoryBoard JSON prompt generation nodes for structured scene and character prompts.
Video description and storyboard custom nodes for ComfyUI. Includes Qwen3-VL video analysis, JSON storyboard parsing, and character prompt generation.
Kling AI, ElevenLabs, and Google Gemini nodes for ComfyUI
Full-featured ElevenLabs integration for ComfyUI - TTS, STS, SFX, Music, Voice Cloning, STT, Dialogue + audio utilities
LLM chat node for ComfyUI — chat with Vision-Language models directly in the node UI, no workflow execution needed for generation.
Send generated 3D assets from ComfyUI to Blender, Maya, and Unreal Engine
A comprehensive suite of video quality assessment metrics for ComfyUI including fidelity metrics, temporal metrics, distributional metrics, and deep learning-based quality evaluation.
A ComfyUI node that randomly picks width and height from a user-defined list on every generation with support for weighted entries.
Custom ComfyUI nodes for practical image, LoRA, workflow timing, and LTX-2.5 motion-control workflows.
ComfyUI custom nodes for RenderZero Studio image and video workflows.
ComfyUI Nodes for OWL-ViT / OWLv2 using the HuggingFace Transformers implementation
Optional companion to ComfyUI-Ultimate-Auto-Sampler-Config-Grid-Testing-Suite. Looks up model and LoRA metadata (including trigger words) from the CivitAI API.
Optional companion to ComfyUI-Ultimate-Auto-Sampler-Config-Grid-Testing-Suite. Offloads VAE decoding to allowlisted HuggingFace Remote VAE endpoints.
Pack PBR material maps (e.g. from ComfyUI-Chord) into labPBR 1.3 and Bedrock (RTX / Vibrant Visuals) resource-pack textures.
A ComfyUI node that stores the last non-empty text input and reuses it when the input is missing or disconnected.
for comfyonline dynamic loader comfyonline is comfyui cloud website
Melite AutoRemesher (quad remesh budget classes) — a Melite first-party ComfyUI extension pack
Local prompt building, validation, workflow guidance, and run summaries for MiniMax H3 workflows in ComfyUI.
Real-time Audio-Reactive WebGL2 GPU Shader Log-Text Visualizer & Auto-Popup HUD for ComfyUI — giant foreground streaming logs tapering into a 3D trail, 4 shaders, Comfy Black & Yellow palette, zero pip dependencies, and 0 MB extra PyTorch VRAM.
ComfyUI custom nodes for VideoSeal video/image watermarking (embed & detect)
ComfyUI node pack that turns your workflows into models for Unplatform AI. Save a workflow with UPAI nodes, add one URL in the app, generate video and images on your own GPU.
API cost tracking, deterministic caching, and lazy execution bypass for ComfyUI cloud API workflows. Circuit-breaker spending control with persistent ledger.
Resolution-independent procedural fields for driving and masking effects. Perlin, simplex, value and Worley noise as pure functions of normalised coordinates, with a probability-integral-transform coverage control; analytic gradients shaped by a draggable stops-ramp widget; exact SDF shapes, tile lattices and scattered stamps; threshold, morphology, exact Euclidean distance, warp and combine nodes for reshaping; fields derived from the image itself via luma, edges and local contrast; and a mask-driven compositor that makes any global effect spatially varying.
Direct Google Gemini image generation for ComfyUI — use your own API key, bypass credits. Supports Gemini 2.0 Flash, 2.5 Flash, 2.5 Pro. Includes Prompt Studio and photographer style transfer.
A node-authoring copilot for ComfyUI: describe a node in plain language, NodeForge retrieves an existing one or authors a new one through a sandboxed write→test→verify loop, and installs it only after you approve the code, the test report, and a before/after preview. Surfaced as a sidebar panel.
Simple ComfyUI custom node that splits text into multiple prompt blocks using a configurable divider for multi-image generation in one queue run.
APIYI ComfyUI custom node plugin providing text-to-image and multi-image editing/blending nodes with support for multiple models and resolutions. (Description by CC)
Send generated images or videos to any HTTP webhook with optional parameters such as prompt-id and metadata payload.
Run selected ComfyUI nodes on rented cloud GPUs as a visible, editable Cloud Offload box, with live progress, previews, and errors relayed in place.
A ComfyUI node that scales and crops images and masks to target aspect ratios with letterboxing, cropping, stretching, and high-quality resampling filters.
A custom ComfyUI node designed for LoRA training that integrates local LLM APIs to automatically generate high-quality, multilingual captions in English, Traditional Chinese, and Simplified Chinese with customizable system prompts and configurable parameters. (Description by CC)
Stackable preset 'flakes' (prompt + LoRA + resolution + ControlNet) inserted between a checkpoint and a sampler.
We present LatentSync, an end-to-end lip-sync method based on audio-conditioned latent diffusion models without any intermediate motion representation, diverging from previous diffusion-based lip-sync methods based on pixel-space diffusion or two-stage generation. Our framework can leverage the powerful capabilities of Stable Diffusion to directly model complex audio-visual correlations.
ComfyUI custom nodes for multi-angle LoRA prompt generation. Provides 3D visual angle selector and simple angle selector with support for 8 azimuths × 4 elevations × 3 distances = 96 view combinations.
This plugin adds features in order to enable full functionality of WorkflowUI Application to which turns your workflows into apps and provides structures for organizing your generations.
Plan a stable subtitle region across video frames using protection masks and visual complexity.
Compose foreground and background videos with five mask shapes, positioning, feathering and borders.
This project contains some custom ComfyUI nodes for image processing, AI conversation, and utility tasks.
ComfyUI custom node for sub2api multimodal text, image generation, and image editing workflows.
Non-linear inference optimization for ComfyUI: 4-tier VRAM, speculative generation, precision routing
Batch-invariant inference nodes for guaranteed reproducibility in ComfyUI
UCDCAE AI Lab Werkraum: A SwarmUI-based Software Framework for Critical Educational and Scientific Exploration and Manipulation of Generative AI
ComfyUI nodes for the project AI for Arts Education
A custom node for SoX's mixdown function. Make sure that the sox command is installed and the path added before using this node.
Drop-in ComfyUI nodes that implement the jolly-cosmos3-comfyuinodes inference path from the NVIDIA model card, supporting Cosmos 3 architecture with Qwen2 chat templates and flow-matching.
A ComfyUI extension to allow textarea elements to be scrolled over. Useful when using a trackpad in order to prevent accidental forward/back navigation (two fingers horizontally on a Mac) when scrolling around the UI.
A ComfyUI extension that enhances touchpad navigation by redirecting two-finger scrolling over to the canvas, including over textareas. This can prevent accidental back/forward browser navigation when using horizontal touchpad gestures and provides smooth zooming and panning for Mac and other touchpad users.
A powerful ComfyUI custom node for local GGUF model storyboard generation with Vision-Language Model support.
Stop ComfyUI from reloading CLIP/UNet/VAE between branches when batching multiple prompts in one workflow.
ComfyUI audio nodes for IndexTTS 2 and IndexTTS 2.5 with voice cloning and five emotion-control modes.
Nodes:PhyCV - Phase-Stretch Transform (PST), PhyCV - VEViD, PhyCV - Page.
ComfyUI custom nodes for Bokeh Diffusion - Defocus Blur Control in Text-to-Image Diffusion Models.
A custom ComfyUI node bridging HY-Motion CLI backend with node-based workflows for fully local text-to-motion to FBX animation generation, designed for production pipelines.
Nodes:ImageCropCircle.
Word embedding utility nodes for ComfyUI. Load a pre-trained embedding model, explore neighbors, do analogies, and project any token/phrase onto 1D/2D/3D semantic axes with human‑readable summaries.
A custom ComfyUI node that uses Ollama vision-language models to generate or refine prompts from multiple input images and text.
Ollama comfyui node for using vl models to generate prompts with custom templates
ComfyUI Dynamic Pipe is a lightweight custom node extension that lets you bundle multiple arbitrary node connections into a single pipe connection, then unpack them later.
Collaboration-first workflow management for ComfyUI with personal workspaces, publishing, and published libraries.
A simple custom node to download models from Hugging Face URLs directly into your ComfyUI model folders.
Move nodes linearly.
Preview media file in model's dir.
Split latent frequency domanin into 3 Bands and Recombine with varying weights.
A ComfyUI node that applies painterly/brush-stroke effects to images, using either ImageMagick (Wand) or G'MIC (gmic-py) under the hood.
A ComfyUI node that will block images that flag custom set key words
Higgsfield API video nodes: Seedance 2.5 and Kling 4.0 (text-to-video and image-to-video) and Genjutsu motion transfer, with your own API key. Each node shows its price and stops before submitting if the estimate is above max_cost_usd.
ComfyUI image batch loader manager plugin with efficient batch upload, preview, and management functionality. (Description by CC)
Basic nodes for image ratios and size presets (all models)
A lightweight ComfyUI extension that highlights the currently executing node by inverting its colors.
Chat with a local NInfer LLM server from ComfyUI: streaming-free OpenAI-compatible chat, multi-image vision input, skill files loaded from disk, a chained options panel (temperature / max tokens / reasoning effort / top-k / top-p / seed / extra body), and process lifecycle control that starts the server on demand and shuts it down when idle so it can share one GPU with ComfyUI. English/Chinese UI.
Add presets for latent and adjust the image size.
A node pack for ComfyUI. Includes LM Studio integration for local LLM/VLM text generation (streaming, vision, custom endpoints), DaC (Divide and Conquer) algorithm v2 for optimal tile-based image upscaling with intelligent overlap and blending, and tile divide/select/combine utilities.