ComfyUI Custom Extensions
Browse 9,446 ComfyUI extensions and 73,215 custom nodes.
ComfyUI custom nodes for Alibaba Qwen3-TTS (Text-to-Speech) with Voice Clone, Design, and Custom Voice inputs.
Create new character voices from existing audio by adjusting persona axes such as Youth, Brightness, Energy, Softness, and Composure. Designed for TTS, voice cloning, and reference-based speech synthesis workflows such as Irodori-TTS.
Custom cycling KSampler with progressive upscale and more. Professional post-processing nodes.
ComfyUI-Kotari: File Manager for ComfyUI - Download models, clone repos, manage files entirely through the UI without terminal access
A lightweight ComfyUI custom node for generating high-quality masks and pose detection for virtual try-on applications. This node extracts only the essential masking functionality from FitDiT without requiring heavy diffusion models.
Elevenlabs-ComfyUI: Complete ElevenLabs API integration - TTS, STT, Voice Cloning, Dubbing, Sound Effects & More
ComfyUI custom nodes that wait on the OS clipboard for new image/text, in-graph replacement for the ComfyUI-Clipboard-Workflow-Automator script.
This is a simple ComfyUI node that will take in a string of 'color' (i.e. 'blue') and output a hex color format.
Apply a LoRA by bbox in Ideogram 4 — per-region character LoRAs in a single forward pass, on the MODEL path.
ComfyUI custom nodes for AlphaVAE — native RGBA transparent image generation using FLUX
ComfyUI node that adds Brightness, RGB channels, and Depth of Field to AI-generated image
GIS Processing Nodes for ComfyUI
A ComfyUI plugin for quick image sequence processing. This plugin allows users to manipulate frame sequences with various operations including frame insertion, deletion, and duplication.
This is a custom node that allows you to run musubi-tuner from ComfyUI.
Reusable resolution presets and aspect-ratio calculation for ComfyUI
converts SD 1.5 embeddings and hypernetworks into usable SDXL versions
把webui版的无边图像浏览移植到 ComfyUI,把 IIB 的核心体验带进 ComfyUI:接管左侧边栏的「资产」按钮,点击后打开一个与「模板」按钮同款的大窗口,用 ComfyUI 原生组件重建浏览、标签、搜索、统计、整理、对比、竖屏流、收藏、索引等能力,并把所有选项放进 设置 → 无限图像浏览。
Chat with an already-loaded Qwen2.5-VL / Qwen3-VL text encoder instead of loading a separate LLM -- multi-turn, vision-capable chat sharing the same VRAM. Exposes both a sidebar panel and an OpenAI-compatible /v1/chat/completions endpoint.
ComfyUI resolution, image, sampling, noise, latent, finishing and workflow utilities.
One ComfyUI loader for modified Qwen3-VL text encoders compatible with MiniMax-H3.
Text render node for ComfyUI
Smart Image Saving Nodes - Offers intelligent folder management and image saving capabilities, supporting flexible folder hierarchy control, multiple metadata sources, various image formats, and metadata embedding. It includes two nodes: SmartFolderManager and SmartImageSaver.
去github下载最新1.6.1版本!go github to download the new version 1.6.1 ! 简单好用的标签与自然语言提示词编写工具! Interactive prompt tag editor: tag chips with hover weight/delete, double-click edit, press-and-drag reorder, click to disable tags, Tencent Hy-MT2-1.8B GGUF translation, load image from path, a multi reroute node (1-20 pass-through rows with type auto-naming), and a Plus variant with dataset-driven tag suggestions while typing.
Local vision-language auto prompt builder for SCAIL/SCAIL-2 character and object replacement workflows. Generates a prompt from a reference image and sampled source-video frames, using local VLM models only.
ComfyUI custom node that applies the same Lora to two models while sharing a CLIP, supporting up to 6 Lora models with independent strength control. (Description by CC)
This custom node package provides seamless integration between ComfyUI and Google Cloud Platform (GCP) Cloud Storage, enabling you to run ComfyUI without relying on instance disk storage. decouple your storage needs from local disks by leveraging GCP's scalable and reliable Cloud Storage service. Models can be saved and loaded directly from Cloud Storage, while outputs and temporary files are automatically synced.
Node alignment panel for ComfyUI with reroute dot support and no color widget features.
A powerful workflow management extension for ComfyUI. This is beat version. Visit homepage and free download full version
MOSS-Image Prompt Weighting via DiT Attention V-Scaling for ComfyUI
Production-ready ComfyUI nodes for AI/ML API video and image endpoints, supporting text-to-image, image-to-image, and video generation.
A single ComfyUI node: image in → prompt text out, via any llama.cpp VLM.
Coarse-to-fine generation and interactive latent-token editing from Trajectory Forcing, as ComfyUI nodes. Needs CUDA 12 and its own Python environment -- read the README on GitHub before installing.
Custom node for extracting specific sections from LLM-generated complex prompts, designed for structured text extraction with clean output. (Description by CC)
All-in-one Discord bot node for AI image generation. Simple setup - just add token, select model, and queue. Includes style presets, rate limiting, and queue management.
A set of ComfyUI custom nodes to control LoRA (Low-Rank Adaptation) strength programmatically during the sampling process.
A ComfyUI node that breaks closed loops in masks to prevent inpainting models from modifying enclosed regions
ComfyUI nodes for Volcengine Ark Seedance video generation tasks.
string to hex converter in ComfyUI
Save clean JPG, WebP, and PNG images from ComfyUI with workflow metadata stored in separate sidecar JSON files.
A ComfyUI custom node for tiling images horizontally with configurable spacing
A ComfyUI custom node for tiling images horizontally with configurable spacing
enable to save the image without the default filename suffix, with the feature of saving caption txt file to save prompts or descriptions of the image.
Compose prompts from multiple toggleable text slots. Combine enabled entries or select one. Dynamic slot visibility with up to 10 text slots in a single node.
Prompt cleaner for Anima Base: Danbooru underscore tags -> comma separated, bracket-escaped, deduplicated, blacklist-filtered prompt text. Several text inputs are merged in order, and all toggles/blacklist live in one settings dialog (STRING out, plus a CLIP Text Encode variant).
Make any ComfyUI workflow faster, use less VRAM, or produce better quality - with AI assistance
Architecture-aware nodes for AI image generation. Auto-detects model architecture (SD1.5, SDXL, Flux, Klein, Chroma), loads correct CLIP/VAE, enhances prompts via local LLM, and samples with optimal settings.
Custom ComfyUI nodes for AllTalk TTS integration
A powerful ComfyUI extension that enables seamless integration with AWS S3 storage for reading and saving images directly within your ComfyUI workflows. This extension eliminates the need for manual file transfers by providing direct S3 connectivity for your image processing pipelines.
一个 ComfyUI 自定义节点,用于通过任意 OpenAI 兼容的图像 API 端点进行图像生成与编辑
EARLY ALPHA - experimental classroom release. Experimental realtime live creative tools for ComfyUI: code (p5js,glsl), drawing, shaders, 2d operators, web audio, MIDI, lessons and performance timelines. Expect bugs and changing interfaces; back up workflows.
The extension opens a window containing the Anima Style Explorer website by ThetaCursed
Convert video framerates and remux containers with two simple FFmpeg-powered nodes. Drag-and-drop, no pip dependencies, chains via the VIDEO type.
MiniMax H3 prompt authoring for ComfyUI
Build filename_prefix from live workflow variables, or pick from saved presets.
Drag one output onto another to take over its downstream links — reroute a connection's source without hunting for the target input.
Full-screen touch editor for multiline prompt widgets: weight steppers, embedding/LoRA insert palette, and a token counter.
Magnifier loupe under your finger for accurate node-connection dragging on touch devices.
Touch-friendly keypad, steppers, and slider modal for seed and INT/FLOAT widgets — seed history, keypad lock, and control_after_generate folded in.
Small stopgap mobile/touch QOL fixes for upstream ComfyUI frontend bugs — each shim toggleable and deletable once fixed upstream.
Long-press tooltips for ComfyUI widgets and sockets on touch devices.
Production-ready custom nodes for advanced diffusion workflows. Modular implementation grounded in proven techniques from leading ComfyUI repositories.
Stepwise asymmetric quantization for flow matching models. Use low-bit models for early denoising steps and high-bit models for later steps — with no quality loss.
Gemini image generation for ComfyUI using Google's official Python SDK with support for text-to-image and reference-image workflows.
A set of ComfyUI custom nodes allowing the use of the Mureka API in your workflows
Helps explore different parameters quickly
Useful to me. Might be to you.
TRELLIS.2 image-to-3D for ComfyUI using the MLX backend — runs natively on Apple Silicon (M1–M4).
A custom node for ComfyUI that extracts the main color palette from an uploaded image and calculates the percentage of each color.
Essential nodes to use dictionaries in ComfyUI: for smart prompt-formatting, general organization (passing a single connection instead of spaghetti), or anything else.
A ComfyUI custom node that creates an animated stroke reveal effect, where a brush/pen follows a mask edge to reveal an image with feathered blending and 8 direction support.
Manage and download models with Hanakura inside ComfyUI.
customized comfyui integration of Spectrum for LTXV by DL
A ComfyUI node to crop images and masks using pixel values. Supports cropping from top, bottom, left, and right, with optional rotation. No resizing is applied — the output keeps the exact cropped size.
String random picker for ComfyUI
MoCapAnything V2 for ComfyUI (Windows portable): video -> motion (FBX, GLB or BVH) on your own rigged character. RMBG-1.4 replaced with MIT BiRefNet.
LingKe AI (灵客AI) API nodes: call cloud image / video / audio / chat models directly from ComfyUI workflows, no GPU required.
ComfyUI nodes for scraping and working with images from public VSCO profiles.
Unified RegionEdit toolkit for region selection, mask editing, cropping, tiling, color and detail harmonization, and strict compositing in ComfyUI. Includes example workflows for smart removal and object replacement.
QG 工具包:提示词规则/翻译/优化、图生文反推、Qwen Image 2.1 提示词优化与文本编码、BiRefNet 抠图、证件照/排版/拼版、人脸矫正与面部对齐、取色器/颜色转换/自定义颜色/主色提取、水印等 32 个节点的中文友好合集(GPL-3.0)
Custom Node to implement iSeeBetter upscaling method.
Workflow protection plugin for ComfyUI that packages selected nodes as encrypted black-box nodes with offline licensing, expiration control, and tamper protection. (Description by CC)
NODES: AnimaArtistCrossAttnMix
Connect ComfyUI to your local llama.cpp server with full support for text, vision, and reasoning extraction.
Krea 2 scanline attention: every DiT block head uses a scanline block mask (global text prefix, image tokens attend +/- window//2 latent rows) via flex_attention. Pair with the scanline-distilled Krea 2 LoRA.
Flexible pipe nodes for bundling and unbundling multiple connections of any type in ComfyUI
ComfyUI custom nodes for creating and applying multiple find/replace pairs to text
ComfyUI nodes for Inkvec, the LogoLabs raster-to-SVG tracer: trace to SVG (with native transparency), save SVG, and JPEG/WebP denoise and x4 upscale for logos before tracing.
ComfyUI Nodes to use [a/SSR Encoder:Encoding Selective Subject Representation for Subject-Driven Generation](https://github.com/Xiaojiu-z/SSR_Encoder).
A node for including random artists usernames to a prompt.
Nodes:Init Layer Info Array, Added Layer Info To Array, Layer Info Array Fuse, Layer Image Seleted, Layer Images IPAdapter Advanced, Enhanced Random Light Source
Nodes: LLM Prompt Tags, LLM Prompt Conditioning. Generates positive and negative prompts with an OpenAI-compatible chat/completions API and can encode them directly into CLIP conditioning.
This package provides two ComfyUI nodes: 🗣️ Dia2 TTS Generator for text-to-speech using Dia2-2B, and 💬 Dia2 Captions Generator to convert TTS timestamps into SRT/SSA/VTT subtitles. Includes example workflow and voice samples. GPU users require CUDA 12.8+.
Video post-processing for ComfyUI: Lanczos upscaling, optional RIFE frame interpolation (Fill Nodes required), adjustable film grain, automatic join tone matching, audio passthrough, configurable H.264 presets/profiles and CRF or lossless encoding, with selectable frame batching. Accepts video or image batches; outputs video, FPS and timing statistics.
ComfyUI custom nodes for streamlined LoRA dataset generation — atomic image+caption saving and built-in variety wildcards.
Replace the first and last frames of an image sequence in ComfyUI. Auto-resizes frames to match video dimensions, handles batches, and adjusts parameters if needed.
Ultra-lightweight text-to-speech model with just 15 million parameters
A professional extension to automatically arrange nodes with Grid and Flow layouts, fully respecting groups.