ComfyUI Custom Extensions
Browse 9,420 ComfyUI extensions and 73,215 custom nodes.
Copy the shot, not the actors: reference video → depth map, OpenPose skeleton (DWPose, tracked and smoothed) or canny line video for Seedance 2.0/2.5 reference video, MiniMax H3 Fun ControlNet (depth / pose / canny) and Wan VACE. Six nodes: Depth / Pose / Canny as IMAGE→IMAGE and VIDEO→VIDEO with the model presets (fps, frame-size multiples) applied. Apache-2.0.
A wild collection of custom nodes for ComfyUI including noise schedulers, samplers, audio preview, latent visualizers, and more — built for maximal creative chaos.
A collection of utility nodes for ComfyUI that enhance workflow efficiency and flexibility.
Save a png or jpeg and option to save prompt/workflow in a text or json file for each image in Comfy + Workflow loading
Cached alternatives to ComfyUI's native MiniMax H3 FL2VA and Ref2VA conditioning nodes that avoid repeatedly loading the large Qwen3-VL text/vision encoder when the same H3 conditioning was already computed.
Structured, seeded, rule-aware prompt composition for ComfyUI from drop-in JSON content packs.
A ComfyUI custom-node pack that enables pixel-position-perfect editing with Qwen-Image-Edit-2511.
Standalone tool to extract generation info from PNG, JPEG, WebP images, works for A1111/Forge, NovelAI, InvokeAI, Midjourney and (most) ComfyUI types of metadata.
Extract generation info from PNG and JPEG images, supports both A1111 and (some) ComfyUI metadata
Windows-first advanced suite for Pollinations, LM Studio, Copilot CLI, and OpenAI-compatible multi-engine workflows.
Multi-LoRA loader with optional LTX layer-specific strength control, drag-and-drop file import, and a companion cycle node for batch-testing LoRA training checkpoints at multiple strengths. Outputs modified model plus JSON metadata of applied LoRAs.
ComfyUI custom node for managing trigger words with 50+ editable presets optimized for SDXL Illustrious
Rui's workflow-specific custom node, written using GPT.
ComfyUI nodes for Foundation-1 model that generate tempo-synced, key-aware music loops with deep control over instrumentation, timbre, effects, and structure. (Description by CC)
A custom node for ComfyUI that enables smooth, keyframe-based animations for image generation. Create dynamic sequences with control over motion, zoom, rotation, and easing effects. Ideal for AI-assisted animation and video content creation.
ComfyUI node integration for the Camie tagger v2 model from Hugging Face for image tagging. (Description by CC)
[a/CSD_MT](https://github.com/Snowfallingplum/CSD-MT) is a method about 'Content-Style Decoupling for Unsupervised Makeup Transfer without Generating Pseudo Ground Truth', you can use it in comfyUI.
NVIDIA PiD (Pixel Diffusion Decoder) as a drop-in replacement for VAE Decode on Anima / Qwen-Image latents.
Welcome to Star Beta Nodes - a collection of experimental custom nodes for ComfyUI designed for beta testing and feedback. These nodes provide enhanced functionality for image processing, video handling, and workflow automation.
A custom node extension for ComfyUI that enables batch processing of prompts from text files to generate multiple images.
Wrapper for EvTexture Video Upscaler: [a/https://github.com/DachunKai/EvTexture](https://github.com/DachunKai/EvTexture)
It allows you to control complex graphs by unbypass/unmute and bypass/mute entire groups of nodes, just by adding tags in the title of your nodes, no connection required.
A slightly different Resolution Selector node, allowing to freely change base resolution and aspect ratio, with options to maintain the pixel count or use the base resolution as the highest or lowest dimension.
This extension provides two nodes to use with my experimental [a/ip-adapter finetune](https://civitai.com/models/1233692?modelVersionId=1390253) for NoobAI-XL style transfer. [a/Here](https://github.com/vahlok-alunmid/reForge-preprocessor_bigG_448) is the counterpart extension for Reforge WebUI.
LM Studio calls with image support and easy task directions.
Kandinsky 5 DiT. High-quality T2V with advanced text conditioning.
ComfyUI is a collection of custom nodes focused on the Doubao ecosystem of the Volcano Engine, providing comprehensive AI content creation functions such as image generation, video generation, text generation, and comic strip creation..
This repo provides a ComfyUI Custom Node implementation of [a/YOLO-NAS-POSE](https://github.com/Deci-AI/super-gradients), powered by TensorRT for ultra fast pose estimation. It has been adapted to work with openpose controlnet (experimental)
Wan2.2 prompt preset plugin for automatically generating high-quality video generation prompts
Fixed WD14 / PixAI tagger nodes for ComfyUI: cached ONNX sessions, visible device reporting, dropdown model selection
Intelligent FFMPEG agent node for ComfyUI with natural language video editing
Simple node to patch ComfyUI generation with the SDXL Flux2VAE arch support.
A very barebones mostly-copypaste implementation of [a/https://github.com/xie-lab-ml/Golden-Noise-for-Diffusion-Models](https://github.com/xie-lab-ml/Golden-Noise-for-Diffusion-Models)
Turn creative ideas into Music 3 caption + lyrics via native VLM generation
Scaled Uniform Noise for Ancestral and Stochastic samplers
A powerful custom node for ComfyUI that transforms fake or blurry AI-generated pixel art into true, crisp, mathematically perfect pixel art.
ComfyUI 3d engine
Qwen Image 2.1: Turbo Prediction and Strix Halo packed attention for text-to-image and image editing. Start edits with full evaluations; edit prediction is opt-in.
Improved Pose & Face Detection for Wan Video 2.2 Animate based on GitHub Issue #1410 (WanVideoPreProcessor) and #10 (WanAnimatePreProcessor). (Special thanks: @kijai and @steven850)
Flux Prompt Generator is a custom node set for ComfyUI that enhances prompt generation and image captioning capabilities. It integrates advanced language models and image captioning techniques to provide versatile and powerful prompt manipulation tools for your AI image generation workflows. NOTE:PORT OF [a/https://huggingface.co/Aitrepreneur/FLUX-Prompt-Generator](https://huggingface.co/Aitrepreneur/FLUX-Prompt-Generator) for COMFYUI
ComfyUI [a/movingforward100/Shadow_R](https://github.com/movingforward100/Shadow_R) Wrapper
A ComfyUI custom node for video analysis and prompt generation using Qwen3-VL-235B vision-language model via OpenRouter API.
A one-click collection of alternate connection styles for ComfyUI.
ComfyUI-Lumina-Next-SFT-DiffusersWrapper is a custom node for ComfyUI that integrates the advanced Lumina-Next-SFT model. It offers high-quality image generation with features like time-aware scaling, optional ODE sampling, and support for high-resolution outputs. This node brings the power of the Lumina text-to-image pipeline directly into ComfyUI workflows, allowing for flexible and powerful image generation capabilities.
Ollama and Llava / vision integration for ComfyUI
This is a node to detect the similarity between two faces
Nodes:Neural Neighbor, CAST, EFDM, MicroAST, Coral Color Transfer.
ComfyUI node for zero-shot singing voice synthesis supporting melody and score conditioning with timbre cloning across multiple languages.
Powerful tools for image and text processing, including cross-platform workflow fixes, optimized resolution, dynamic text/image combos, and batch processing. Unlock seamless AI art creation and boost your productivity!
A multimodal API workflow toolkit for ComfyUI with image, video, audio, text, batch generation, asset cards, and an integrated chat assistant.
Interactive 3D pose editor inside ComfyUI. Load VRM/GLB/GLTF models, drag bones to pose, control shape keys/expressions, and capture as IMAGE. All Three.js dependencies are bundled locally (no CDN required).
This repository provides the ComfyUI plugin for DFloat11 models. DFloat11 reduces model size by more than 30% while producing bit-for-bit identical outputs to the original. Unlike quantization techniques which trade quality for size, DFloat11 is a lossless compression method, preserving model output quality fully while supporting efficient inference.
ComfyUI nodes to use [a/Flux-version-LayerDiffuse](https://github.com/RedAIGC/Flux-version-LayerDiffuse)
Embeds an invisible watermark in the input image
A stylized node with simple operation. The effect is achieved by I2I and lora. The clay style is currently implemented.Comes with watermark function.
PromptStyler is a custom ComfyUI node that helps you keep your prompts consistent by applying a selected style template to your positive prompt, then outputting CONDITIONING for KSampler.
A custom node for ComfyUI where you can paste/type any python code and it will get executed when you run the workflow.
A studio rig for ComfyUI: a tabbed workspace panel, painting with auto masks, wireless control panels, LoRA management, grading and prompt tooling, plus Krea 2 moodboard and identity editing.
Makes the prompts for generating anime images more detailed by upsampling Danbooru tags.
The ultimate text processing suite for ComfyUI. Now features Batch Cleaning, LLM Parsing, Dynamic Wildcards, Image and Logic Tools for sophisticated automation.
Returns the duration of an audio source, which can be provided either as a filesystem path or as raw audio data (samples and sample rate) from an upstream node. Duration calculation primarily uses ffprobe (from FFmpeg) for paths, or calculates the duration directly from samples/SR if no path is available.
A custom node for ComfyUI that integrates DeepSeek's powerful chat and instruction API, enabling seamless AI interactions within your ComfyUI workflows.
The ComfyUI_pixtral_vision is a powerful ComfyUI node designed to integrate seamlessly with the Mistral Pixtral API. It facilitates the analysis of images through deep learning models, interpreting and describing the visual content. Users can input an image directly and provide prompts for context, utilizing an API key for authentication.
Reason-then-encode for Krea 2. The encoder's own Qwen3-VL rewrites your prompt into a richer one — or reads a reference image (describe → filter → combine) — before conditioning: a built-in prompt enhancer and language-level image reference, running on the exact model that conditions.
A collection of nodes and improvements created for general ease and lora management. These are just nodes I made and found useful, they should work with most other nodes. Most nodes that take in a prompt are made with booru tags in mind and might not work as expected with other prompts.
Phi-3.5-vision-instruct fast talk with image ! Fast , Fast ,Fast! 1.Phi-3.5-vision-instruct
NODES:CXH_Gemini2_TX, CXH_Gemini2_Vision, CXH_Local_Prompt
NODES:StoryMakerSinglePortraitNode, StoryMakerTwoPortraitNode, StoryMakerSwapClothNode.
This project implements a set of custom nodes for ComfyUI, integrating some of the API interfaces provided by [a/TAMS](https://tams.tensor.art/).
Connects ComfyUI to the PhotoCraft image editor: the open document, its selection and a prompt from its notes go into a workflow, the result comes back as a layer. Needs PhotoCraft started with its control channel; the console shows how.
LockPixel is a ComfyUI custom node pack for Qwen-Image-Edit workflows where pixel drift, small shifts, or subtle zoom-like deformation must be avoided.
A toolset for Hunyuan Video (mainly), with some additional nodes applicable to Framepack Video and WAN Video
A collection of utility nodes primarily for interacting with comfy via automated systems
Nodes:Combine HDMasks, Cover HDMasks, HD FaceIndex, HD SmoothEdge, HD GetMaskArea, HD Image Levels, HD Ultimate SD Upscale
I'm SD-VJ. (share SD-generating-process in realtime by gpu)
A ComfyUI custom node pack implementing Diff-Aid-inspired inference-time text-conditioning patches for Flux and SDXL models.
Custom ComfyUI nodes: XMath, XResolution, XImageSave, XLatentLoad, XLatentSave, XVideoSave, XAudioSave, XStringGroup, XWorkflowSave, XMetadataWorkflow(Extension), XDateTimeString, XFitView(Extension) ...and more
Native ComfyUI nodes for EverAnimate-style chunked WanAnimate workflows, including master/chunk settings, chunk calculation, image carry-over, and boundary color correction.
Compatible with alimama's SD3-ControlNet Demo on ComfyUI.
A powerful OCR node for ComfyUI that integrates the DeepSeek-OCR model from Hugging Face.
Custom nodes for ComfyUI by Comflowy.
Nodes:abyz22_Padding Image, abyz22_ImpactWildcardEncode, abyz22_setimageinfo, abyz22_SaveImage, abyz22_ImpactWildcardEncode_GetPrompt, abyz22_SetQueue, abyz22_drawmask, abyz22_FirstNonNull, abyz22_blendimages, abyz22_blend_onecolor. Please check workflow in [a/https://github.com/abyz22/image_control](https://github.com/abyz22/image_control)
a custom node for [a/DiffMorpher](https://github.com/Kevin-thu/DiffMorpher),you can find base workflow in [a/doc](https://github.com/AIFSH/DiffMorpher-ComfyUI/blob/main/doc)
A custom node for ComfyUI that provides the ability to clear the cache directly from your workflow.
Huggingface Api Serverless request
Intel XPU dynamic model unloader for ComfyUI, reimplements NVIDIA CUDA VRAM management for Arc B series GPUs. (Description by CC)
Nodes:Asymmetric_Tiling_KSampler.
Manual tapered context noising for MiniMax H3 Motion Context (character-swap continuation)
Unofficial: run HyperFlow H3 Flash's own INT8 (convrot) checkpoints, fl2va and ref2va, on ComfyUI's native INT8 kernels, and use regular MiniMax H3 LoRAs on them. Needs the official ComfyUI-HyperFlow-Flash pack.
Tile the latent, not the pixels: latent-native tiled decoding for NVIDIA PiD. Seam-free masters up to 244MP; size-picker dropdowns with render-tested presets for Krea2/qwen AND FLUX.2 Klein 4B/9B; avoids the past-envelope color collapse of single-shot decoding. No VAE re-encode round-trip.
Comprehensive ComfyUI nodes for 3D mesh processing, PBR texture baking, caching, character consistency, and prompt iteration
Simple node for setting the sigma values directly. Note, for a full denoise the last sigma should be zero.
Generate high-fidelity, synchronized foley audio for any video directly within ComfyUI, powered by Tencent's HunyuanVideo-Foley model.
ComfyUI workflow preset switching plugin allowing quick toggle of multiple presets in a single workflow via index value. (Description by CC)
ComfyUI workflow preset switching plugin that allows quick toggling between multiple presets in a single workflow using an index value.
Investigating the Effectiveness of Cross Attention to Unlock Zero-Shot Editing of Text-to-Video Diffusion Models
Unofficial implementation of [a/deepseek-ai/Janus](https://github.com/deepseek-ai/Janus) in ComfyUI.
Generate audio-reactive SCAIL pose sequences for character animation without requiring input video tracking. Now supports Multi-Character Choreography. (Description by CC)
This extension integrates ebsynth_utility into comfyui, and I've written some of my own nodes for secondary use.