ComfyUI Custom Extensions
Browse 9,446 ComfyUI extensions and 73,215 custom nodes.
A ComfyUI node that integrates DINO-X API for object detection and segmentation. This node allows you to detect and segment objects in images using text prompts.
This is a node for using cohere (Command R+) from ComfyUI. You need to edit the startup .bat file of ComfyUI and describe the API key obtained from Cohere as follows.
This is a ComfyUI node that rotates an image by specifying an angle or selects one image from multiple images.
Extended ComfyUI node with WD14 Tagger-based image analysis and automatic character identification to apply optimized prompts dynamically. (Description by CC)
Custom nodes to make ComfyUI sampling closer to flow-matching behavior for Flux, Qwen-Image, and Z-Image.
Extract color palettes from reference images and package them as Krea 2 image_style_references.
This ComfyUI extension provides custom nodes for working with Google Gemini and OpenAI ChatGPT.
ComfyUI custom node to support 3D GS rendering
Nodes:BatchImageFromList, Load Face Occlusion Model, Create Face Mask, Simple FaceAlign, Cacul FaceAlign, Gen Blurbord, Face Align, Face Rotate, ImageAdd, LoadImageList, SaveImage Plus, RegTextFind
Restart the ComfyUI server as a queued job (after the current queue finishes), with a top-bar button.
Native Ming 3.0 Tiny speech synthesis, voice cloning and Confucius ASR for ComfyUI.
MOSS-Image Base and SFT with the official pipeline and independent DiT, text encoder and VAE loaders.
Batch image processing and Vision-Language Model (VLM) interrogation nodes for ComfyUI, supporting Qwen-VL, JoyCaption, and WD14 Tagger.
A custom node for ComfyUI designed to manage and automate optimal settings for different Checkpoints. It allows you to save and recall specific parameters—including steps, CFG, samplers, schedulers, and resolutions—linked directly to the model name.
Custom node for ComfyUI that analyzes songs using NVIDIA's Music Flamingo 2601 model with optimized 4-bit quantization.
TTS Audio support without cosyvoice ASR. (Description by CC)
A zero nonsense ComfyUI node that resizes every incoming image (or batch) so the final resolution hovers around one million pixels—while respecting the original aspect ratio and snapping both dimensions to multiples of 64. Perfect for pipelines that need a predictable VRAM footprint without destroying composition.
MiniMax H3 utility nodes for ComfyUI — text-to-video, image-to-video, video post-processing (upscale/interpolate/export), and MiniMax music generation/post-processing, wrapping the local 33.1B joint video+audio model merged into ComfyUI core.
ComfyUI Wan 2.1 toolkit with visual multi-ref node, positioned reference gallery, and VACE multi-ref to video nodes.
Custom nodes for loading and previewing media from URLs in ComfyUI.
Training-free CharaConsist algorithm implementation for ComfyUI - Generate consistent subjects across multiple generations with enhanced mask generation and multi-model support.
Works on AMD AND NVIDIA - the name is where it started, not what it supports. VRAM monitoring, crash early-warning and per-run analytics for ComfyUI. NVML-based monitors show no VRAM at all on an AMD card; this reads ComfyUI's own data instead, so it works everywhere, and the warnings adapt to the backend it finds. Everyone gets: time-by-node breakdown showing exactly where a run's minutes went, run history with model, LoRAs, settings, outputs and peak usage, per-run logs written to disk as the run happens so they survive a crash, plain-English explanations of ComfyUI's cryptic errors, progress with ETA, swap, CPU and output-disk monitoring, and optional AI analysis over your own Ollama or any OpenAI-compatible endpoint. AMD users additionally get VRAM figures and partial-load crash warnings that nothing else provides.
ComfyUI custom nodes to reshape/project Conditioning tensors and ensure SDXL conditioning metadata.
Vision-aware text conditioning for MOSS-Image (Qwen3.5-9B text encoder)
ComfyUI custom node for Venice.ai text generation with E2EE support (v3 schema). Includes ECDH secp256k1 + AES-256-GCM end-to-end encryption for e2ee- models.
Pack of 6 nodes: MI Lora (several LoRAs with remembered trigger words and on/off per row), MI Unet Rate (UNET loader where you rate and comment each model), MI Embedding (embedding picker with strength), MI Compare 3 Images (switch, wipe and save between three images), MI Save Image (preview first, save to output on demand) and MI Reroute Sound (a minimal reroute that plays a WAV sound when the flow passes).
Nodes for Jasper's LBM models
A custom node for ComfyUI that allows you to send images directly to [a/Eagle](https://jp.eagle.cool/).
A custom node for ComfyUI that classifies images into NSFW (Not Safe For Work) categories.
ComfyUI nodes for MiMo-V2.5-TTS preset voices, voice design, and voice cloning.
DINO-guided upscaler for ComfyUI with semantic-aware image enhancement
ComfyUI custom nodes for creating polka dot collages
A few enhanced versions of vanilla nodes, mainly Qwen/ZImage controlnet nodes & scheduled CFG guider, and some utility nodes.
Use ollama to generate prompts based on reference text in comfyui.
Share your local ComfyUI via a Cloudflare Quick Tunnel with an optional password gate
Load tonera vitoom-nunchaku SVDQuant DiT checkpoints in ComfyUI.Support flux-3 klein, Qwen-Image-2.1...
把 ComfyUI 输出文件里内嵌的提示词与工作流汇总成一份可搜索、可归类、可预览的档案。
Extracts a single line from multiline string. Useful for batch processing
A collection of custom nodes for ComfyUI by Ha Manh Tuyen
Architecture-aware FLUX.2 Klein LoRA loader for ComfyUI with identity-preserving edit controls and drift-risk analysis.
This is a custom node for ComfyUI that allows you to use the Compshare API directly in ComfyUI. Compshare is a high-performance AI inference platform providing powerful image generation models including popular text-to-image and image editing models without local GPU setup. [a/compshare docs](https://www.compshare.cn?ytag=comfyplug) 优云智算的ComfyUI的自定义节点,可直接在ComfyUI中使用优云智算的生图API。优云智算是一个高性能的AI推理平台,提供强大的图像生成模型(文生图、图生图、图像编辑)和行业领先的推理速度,无需本地GPU设置即可使用。[a/优云智算官网](https://www.compshare.cn?ytag=comfyplug)
Adapts Qwen Multiangle Camera prompts and camera coordinates for multi-angle MiniMax H3 image editing.
Visual timeline for MiniMax H3 video generation with shot editing, reference images, audio, and project saving. Powered by the H3 Continuum generation engine.
Experimental beta: settings-controlled MOSS-Image prompt weights for ordinary ComfyUI text inputs. Actual ComfyUI browser UI and MOSS GPU generation remain unverified.
A compact Empty Latent Image replacement that also outputs WIDTH and HEIGHT values for ComfyUI workflows.
Lightweight custom nodes that complement ComfyUI's native Hunyuan3D-2 support, including image loading from URL, mesh conversion, UV unwrapping, and mesh export.
A package of utility nodes. Master Batcher, Folder File Counter, Incrementor Plus, and Remember Me (Environment Info)
A collection of custom nodes for ComfyUI, focusing on improving workflow efficiency and adding new functionality.(work in progress!!!)
ntage-DreamOmni2 brings Qwen2.5-VL’s multimodal intelligence to ComfyUI, enabling unified text–image understanding for both generation and editing. Create context-aware prompts, perform reference-based visual edits, and build adaptive image grids — all with high consistency, abstraction, and creative control.
A better SaveImage than SaveImage-Plus.
Real-time per-node execution timing, live log, and workflow reconstruction for ComfyUI
A collection of ComfyUI nodes including an all-in-one wildcard processor with LoRA loading and aspect ratio node for image sizing.
ComfyUI Nodes for Remote VAE Decoding.
A ComfyUI custom node package for generating lip-sync videos from audio using WAN 2.1 models with automatic frame extension for long audio.
A ComfyUI custom node that enables direct loading of musubi-tuner based LoRAs (Z-Image Turbo, Hunyuan Video 1.5) with automatic in-memory conversion and caching.
Automatic multi-segment SCAIL-2 character replacement sampling for ComfyUI.
Daily-use ComfyUI utilities for workflow repair, subgraph previews, clean switching, and better run feedback.
ComfyUI nodes to create and integrate with Voxta
A ComfyUI custom node that replicates the function of the -Free model and node cache- toolbar button, usable INLINE in a workflow, so you can force a VRAM/RAM cleanup automatically after every generation instead of clicking a button manually.
NSFW content detection node for ComfyUI - blocks inappropriate images automatically using YOLOv9 model.
Dots TTS — 2B-parameter fully continuous, end-to-end autoregressive text-to-speech system with zero-shot voice cloning support.
Multilingual Translation for Text & Images
Simple ComfyUI RetroArch Web Player embed
Agnes.ai nodes and example workflows for text-to-image and text-to-video generation in ComfyUI. (Description by CC)
To enhance the usability of ComfyUI, optimizations and integrations have been implemented for several commonly used nodes.
Provides a structured model and helper nodes for producing long video sequences on ComfyUI using low vram machines.
Provide custom nodes to let you write python code snips directly in ComfyUI web ui.
NODES: PromptStyler, PromptLatent, PromptCLIPEncode, PromptSelector
A series of custom, novelty nodes for ComfyUI by Lord Winnougan
Utility nodes built to work inside subgraphs: buttons, live results and previews stay usable on the subgraph node. A seed with random/fixed controls, a seeded random number, an aspect-ratio resolution selector, plain-text previews with concat/switch, a math expression node, an image folder loader with file names and filtering, a minimal reroute, a resolution reroute with a RESOLUTION type, a slider with a range you set, and a group bypasser and muter that work in the Vue nodes view.
This Module provides nodes to allow the development of 3D Generative AI workflows that use the MasterpieceX Python SDK.
A visual LoRA selector node that displays a LoRA gallery in the node, allowing direct browsing, searching, and selection with thumbnail uploads and built-in prompt storage. (Description by CC)
Anima Edit ReferenceLatent compatibility patch for ComfyUI. Lets built-in ReferenceLatent workflows drive single or multiple Anima Edit reference latents.
Use this ComfyUI from Fuobe with one install: no start flag, no extra file. Trusts only the Fuobe studios you name, checks every workflow they send against the node list it ships, keeps ComfyUI's own check for every other website, and can pair with your Fuobe account for Safari, phones and PCs with no screen.
Custom node providing direct support for 36000+ anime characters from Animadex with dropdown selection and automatic trigger words and prompt tags output. (Description by CC)
Windows-only MOSS BBox scene-scale and composition control nodes for ComfyUI.
A ComfyUI workflow manager plugin with auto-save, version history, sidebar integration and more.
结构化标签库节点: 双模式随机引擎 + 反冲突系统 + .json 库文件进出 + 管理页, 输出 STRING 直连 CLIPTextEncode。
NODES: WWL_StableDelight
Nodes:Create N-Token String
This node enables integration between ComfyUI and external services by adding callback capabilities to the image saving process. When an image is saved, the node automatically call your webhook with your specified URL with custom data.
A very useful automatic cropping tool! It can realize cropping, expansion and rotation functions in the form of a slider.
ECHO: reference-guided color consistency for fashion SKU image sets
ComfyUI nodes for quick and easy style selection for the Anima model
Two simple nodes that load a public Google Sheet and let you pick text from any cell by row and column index.
Nodes to fast load, navigate, create, modify and save JSON data inside ComfyUI workflows.
This ComfyUI custom node package mirrors the async flow used by the local plugin: submit a video task, query task status, download the next completed video.
ComfyUI prompt toolkit: random draw, danbooru search, tag translate, prompt enhance
comfyUI_LLM is the integration of a large language model into ComfyUI
Real-time image transfer between client and server Base64 image encoding/decoding support Supports PNG image format Includes a floating preview window for received images Preview window has minimize/maximize functionality
给 ComfyUI 的资源加载节点加媒体浏览器:文件夹浏览、瀑布流、收藏、提示词提取、工作流载入、预览遮蔽与截图。
ComfyUI-NIUNIUAPI is a professional API call node designed to facilitate the rapid construction and efficient integration of ComfyUI API workflows.
ComfyUI 0.37.x launcher-compatibility shim: runtime-disables the comfy_aimdo memory/compiler graph, equivalent to launching with --disable-comfy-compiler, preventing aimdo memory compile error crashes during the first sampling step.
Floating multi-hardware real-time monitor HUD for the ComfyUI canvas - a draggable overlay panel with five gauge dials showing real-time utilization, temperature, and memory levels of RTX GPU / Intel iGPU XPU / NPU / CPU / RAM.
Native 2K Joint Video-Audio Generation with Block Sparse Attention — ComfyUI Plugin
Unifies removal of the PyTorch per-process VRAM memory fraction (set_per_process_memory_fraction) limits imposed by ComfyUI custom nodes.
Intel AI Boost (NPU) face inference service for ComfyUI. Uses an NPU -> GPU -> CPU fallback for face detection, landmark and gender/age tasks. Face recognition (face-recog-r50) runs on the CPU because the NPU backend produces NaN outputs.
BSAI video merge node - concatenates multiple video clips of different resolutions and frame rates in order into one video, automatically handling size differences and audio alignment.
A ComfyUI custom node implementing CFG-Ctrl / SMC-CFG as a GUIDER.