ComfyUI Custom Extensions
Browse 9,446 ComfyUI extensions and 73,215 custom nodes.
Reads A1111 and ComfyUI PNG metadata and applies editable generation settings to bound checkpoint, LoRA, CLIP text encode, and KSampler nodes.
This is a node that applies deepGHS's image restoration model within ComfyUI (original project link: https://huggingface.co/deepghs/image_restoration).
Nodes:SegToMask
Collection of nodes for the automation of workflows
Nodes: Convert String To Int, Convert String To Float
Multi-monitor preview with external window (zoom, pan, save), floating in-browser preview, and prompt monitoring. Features stable threading, customizable positioning, and Signal Scout for real-time prompt tracking.
A collection of custom nodes for ComfyUI
a simple prompter for video
Advanced workflow intelligence platform for ComfyUI that analyzes, understands, and optimizes ComfyUI workflows through architecture analysis, dependency mapping, and health diagnostics.
An artist-focused workflow documentation platform for ComfyUI that automatically analyzes workflows and generates detailed guides including workflow purpose, models used, main settings, performance tips, and key nodes used. (Description by CC)
Custom ComfyUI nodes to run SLM Vision models (DeepSeek-vl 1.3b chat, Qwen2.5-vl 3b, Gemma-3-4b-it) with optional RAG support.
Based on the original SaveImage node from ComfyUI This ComfyUI node provides functionality to save images in PNG format while simultaneously generating and returning their Base64 encoded strings. This is particularly useful for applications that require image data to be transmitted as strings, such as web applications.
Quickly search and get images from Gelbooru/Danbooru/E621 without leaving ComfyUI
A ComfyUI node that displays the type and contents of whatever is connected to the input. In the case of a Tensor object, it shows the shape instead of its value.
Pixel-Lock for Qwen-Edit: composite only the edited region back onto the original so unedited pixels stay pixel-identical (no zoom-drift, no whole-image degradation, multi-round safe). Auto-bypasses global/style edits. Tensor-native, zero deps. Validated across a 374-edit sweep.
A professional, high-fidelity post-processing node suite aiming for photorealism. Simulates precise physical analog camera behaviors (Bloom, Optics, Glare, Diffusion).
A custom node to create empty latents for Stable Cascade. features: width and height incrementation of 64 by default, possibility to lock the aspect ratio, switch width/height at execution
LordLoraQwenEdit is a ComfyUI custom node for Qwen Image Image Edit pipelines. It selectively applies Qwen Image edit LoRAs to specific UNet transformer layers based on keyword filters, keeping the rest of the network untouched for precise control.
Manage a whole stack of LoRAs on one ComfyUI node, with Civitai metadata, sample images and trigger words.
A powerful prompt management system for ComfyUI. Store, categorize, search, and combine prompts with full LoRA integration — all from a beautiful built-in web UI.
This repository fully replaces and extends the previous Hangover Nodes
This extension provides convenience nodes for batch processing.
ComfyUI custom node: rewrites a plain prompt into Ideogram 4's native structured JSON caption via the hosted magic-prompt API.
Put your own character into any photo, keeping its framing, lighting, grade and pose. Runs Nano Banana Pro (Gemini 3 Pro Image) on your own Google billing via AI Studio or Vertex AI.
Nano Banana Pro and other Gemini image models in ComfyUI on your own Google billing, via AI Studio or Vertex AI. No ComfyUI credits.
Strings should be easy, and simple. This extension aims to provide a set of nodes that make working with strings in ComfyUI a little bit easier.
Ten-image reference manager for Qwen Image.
Nodes:Enhanced Save Node
This collection of custom nodes for ComfyUI is specifically designed to automate and streamline the process of creating infinite videos with WAN2.1 VACE.
Pick a range of colours by Hue / Saturation / Lightness and get back a clean mask where every matching pixel is white and everything else is black.
Magical nodes that are meant for integration and science of course. ^^ Foundational Helpers and smart Containers that use automated functionalities to make room for creative use. A magical pack-synergy is at hand that does not require much extra clutter to make advanced techniques pop beautifully. The idea was to create universal artist's precision tools that do not care what you throw at them.
Helto all-in-one ComfyUI image generation nodes with managed privacy.
A node that can be dropped into a group. When a node is then clicked within that group, all nodes and connections in that group get greyed out and the connections from the clicked node go bright red.
Support nodes for my workflows. File management, video generation
A focused node collection designed for pipelines using real photographic images where resolutions are not standardized like they are in text-to-image generation.
VoxLock voice isolation for ComfyUI: keep one voice and remove the other voices in a recording, with the official VoxLock models (ONNX, CPU, about 12 times real time). Same output as the official voxlock package, bit for bit.
Standalone ComfyUI nodes for prompt rewriting with local GGUF models via llama-cpp-python.
A collection of nodes for manipulating LATENT in ComfyUI.
ComfyUI node to calculate optimal resolution to crop the image to (from a list of aspect ratios)
Tikpan 官方独家节点:支持 GPT-Image-2、Gemini、Grok、Qwen、Wan、Doubao、Suno、HappyHorse、Kling、Vidu3、Veo、Gemini Omni 等图像、视频与音频生成
Simple attention backend selector for ComfyUI supporting basic, sub_quad, split, pytorch, xformers, sage, and flash attention implementations.
A collection of custom nodes for ComfyUI that allow working with LoRAs and trigger words by index selection.
Fast image upscaling for ComfyUI using TensorRT acceleration (2-4x faster than standard methods). Auto-detects CUDA version and installs matching TensorRT wheels. Supports multiple upscaling models with automatic engine building and optimized inference for NVIDIA GPUs.
Open-weights lip-sync for music videos in ComfyUI. LSDub warps the singer's own mouth to the song (no new face, no face swap) and scores every shot, so you know which ones worked. Bring a clip and a song, or generate them with LTX-2.5, Qwen-Image-Edit and ACE-Step.
Universal domain-adaptive sampler for ComfyUI. Flow Matching + EDM with adaptive phase routing, momentum, SABER stabilization and perceptual sharpening
A comprehensive ComfyUI custom node suite for advanced Safe for Work prompt generation across all checkpoint types (Pony, SDXL, SD1.5, and more). This SFW edition is specifically designed for professional workflows, educational environments, and GitHub hosting compliance.
a comfyui custom node, which can select value from inputs
A collection of ComfyUI Nodes, most to make dynamic prompting a bit easier.
A ComfyUI custom node pack for interactive prompt building with checkbox toggling, weight control, and prompt utilities.
Fork of ComfyUI-QwenVL: Qwen-VL nodes (Qwen2.5-VL / Qwen3-VL, Transformers and GGUF backends) for text generation, image understanding, and video analysis. Adds local-only model discovery — models are picked from models/text_encoders and models/LLM with no automatic downloading — plus explicit mmproj selection, up to 3 image inputs on the Advanced nodes, a thinking-mode toggle, Gemma 4 / Qwen3.5 / Qwen3.6 / Qwen3.8 GGUF support, and MTP speculative decoding.
Facilitates model files audit and search
Download Youtube videos using ComfyUI
Node:Load Checkerboard Images for Calibrate Camera, Matrix and distortion coefficient to text, Undistort
Nodes for playing mini-games with ComfyUI.
ComfyUI wrapper for MatAnyone2 repository, requires cloning official MatAnyone2 and installing dependencies. (Description by CC)
Combines prompts from multiple categories and automatically advances through every combination across queue runs for batch variation generation.
IMGLAB custom nodes for ComfyUI
Creates before/after image comparison sliders from two images. Publishes to a shareable URL.
A mock of a possible implementation of for ComfyUI Core Video Nodes.
NODES: DrawTextNode, Qwen2.5VL_api, ...
Python script executor, powerful control panel (resource monitor, models management), pipe any, switch any (branch/if), file reading, diagram display (Graphviz, Mermaid, PlantUML) and other useful tools.
HTTP API for ComfyUI with webhook-based workflow execution.
The Webhook Notification plugin for ComfyUI, used to send webhook notifications when image generation is complete.
Execute shell commands/scripts with dynamic inputs, and result caching. Supports placeholders ({input0}, {seed}) and environment variables ($INPUT_0, $SEED, etc.).
Nodes:IW SaveString, IW PrintString, IW ReplaceString, IW StringNode, IW StringConcat, IW TokenizerVocab, IW JsonPickItem.
This workflow demonstrates a streamlined way to connect visual inputs with Large Multimodal Models (VLM) using a custom API node.
Synchronize ComfyUI PBR texture maps with Blender, Maya, and Unreal Engine over localhost TCP.
A ComfyUI custom node that generates precise mouth and lip masks from DWPose output with multiple mask modes for inpainting teeth, lips, or mouth regions.
ComfyUI Custom Node to Rotate Images, Img2Img node.
A ComfyUI node that loads existing images from the output directory by recency.
A full-screen ComfyUI output gallery for browsing images and workflow prompt metadata.
Editable reference media timeline for ComfyUI MiniMax H3 Reference to Video
**Random IMDb movie goof → AI video prompt → LTX-2 video clips → procedural interstitial → MusicGen film score → final stitched video.**
A collection of plugins for ComfyUI that provides a variety of useful nodes (batch path loading, rotate & crop, mask crop & composition, proportional scaling, slicing, string utilities, arbitrary type switching, image inference, etc.).
The Pony Character Prompt Picker node reads an Excel file specified by the user, allows manual selection of a tab, and randomly picks a cell value from a specified column, starting from row 3 to the end. The selected value is output as a string to the next node in the ComfyUI workflow.
A collection of utility nodes using Dictionary designed to optimize and manage workflows in ComfyUI.
Multi-turn conversational image editing powered by Google Gemini — persistent sessions, edit chaining, and visual gallery with full scene coherence across turns.
Midjourney image generation in ComfyUI via the LegNext.ai pool-based proxy — pay-as-you-go, no Discord required
Houdini-style reference copy for ComfyUI: clone any node and keep every parameter live-linked to the original. Edit one parameter on a clone to localize just that one; the rest stay linked. Pure frontend, zero graph nodes.
Simple wireless data transmission nodes for ComfyUI — send and receive any data type without visible wires using named channels. Inspired by Blackmagic Fusion.
OpenAI API based prompt rewriting agents for MiniMax H3 T2VA/I2VA/L2VA/FL2VA and Ref2VA workflows, using the official H3 skill files.
Speech synthesis, voice cloning and voice design nodes for Qwen3-TTS, patched for transformers 5.x (verified on 5.14.1 / ComfyUI 0.37.0 / torch 2.13.0). Fork of flybirdxx/ComfyUI-Qwen-TTS v1.0.7, which declares transformers>=4.57.0,<5.0.0. This fork adapts the four affected files so the plugin runs on transformers 5.x: pad_token_id no longer copied onto PretrainedConfig, ROPE_INIT_FUNCTIONS['default'] removed, create_causal_mask signature change, cache_position missing on decode (shape crash), and inv_freq left as uninitialised meta memory (silent failure producing wrong-sounding audio). The folder created under custom_nodes/ is named ComfyUI-Qwen-TTS-5x. Node CLASS names are identical to upstream (FB_Qwen3TTS*), so the two packages must not be enabled at the same time -- disable or remove one of them. If you need the original 4.x behaviour, install flybirdxx/ComfyUI-Qwen-TTS pinned to transformers 4.57.3. License: Apache-2.0 (code). Model weights are not included and remain under the Qwen3-TTS License Agreement from Alibaba Cloud.
Three ComfyUI nodes for generating structured WAN 2.2 video prompts, built specifically for chained multi-segment video pipelines.
Context-aware intelligent AI copilot inside ComfyUI. Explains workflows, analyzes nodes, and diagnoses runtime errors.
A custom node plugin for modifying ComfyUI workflow node IDs with right-click canvas editing, batch ID modification, and automatic workflow normalization features. (Description by CC)
Adds a node that outputs the workflow's name as a string
ComfyUI custom nodes for loading anime character prompts with aesthetic suffix support
This package provides a custom node to ComfyUI to send a message and image by means of a webhook
A collection of quality-of-life and convenience nodes.
Save native ComfyUI videos to a local folder with an in-node preview and overwrite protection.
ComfyUI custom nodes for Soprano TTS, a fast, lightweight text-to-speech model.
Wireless node toggle control for ComfyUI - control any nodes by title/color matching with radio-button restrictions
A fast, context-aware hotbox for ComfyUI: smart auto-wiring, model swapping, drag-to-pin bars, pro sliders, nodes saved as buttons and visual crop/paint editors. No Python nodes, zero dependencies.
This node utilizes the Groq.com API to enhance prompts. (Place API key and main system prompt in the groq_config.json)
Jacques' personal ComfyUI nodes (Gemini outpaint, padding, extras)
ComfyUI nodes for SoulX-Singer, a high-quality singing voice synthesis system.
Python script node for ComfyUI supporting math, calc, logic, and text editing with multi output. Handle conditional workflows, auto-select resolution, parse LLM outputs, and generate filenames — all in one flexible node.
按类别开关 LoRA(多标签 / 同类互斥 / 勾选联动 / 按子文件夹过滤 / 临时分类)+ 为每个 LoRA 挂载多套自定义提示词预设,勾选后输出为字符串
Shima - ComfyUI Workflow Creation and Sharing System