ComfyUI Custom Extensions
Browse 9,420 ComfyUI extensions and 73,215 custom nodes.
A ComfyUI custom node set to offload CLIP encoding to a remote machine.
CSV Loader for prompt building within ComfyUI interface. Allows access to positive/negative prompts associated with a name. Selections are being pulled from CSV files.
Node to compare 2 videos with slider.
A collection of custom nodes for ComfyUI. Includes a quick canny edge detection node with unconventional settings, simple LoRA stack nodes for workflow efficiency, and a customizable aspect ratio node.
Custom nodes for connecting 3D scenes and ComfyUI workflows.
Custom nodes for connecting 3D scenes and ComfyUI workflows.
DiffusionLight node implmented to ComfyUI
Panoramic, fisheye, VR180, and VR360 image/video preparation and projection nodes for ComfyUI.
This is an imaginative project that allows for one-click deployment, providing both an online page and a ComfyUI cloud node.[w/This custom node is vulnerable because it can dynamically download and execute nodes.]
Custom ComfyUI node pack by rogala
Background remove and/or replacement. Can be applied to images and videos. Low dependencies, not heavy packages needed. Supports a lot of models including the populars: BiRefNet, BEN1/2 (MVANet), InSPyReNet Some older models: U-2-Net and IS-Net And some modern approaches: PDFNet, MODNet and DiffDIS. Also supports the models trained by BRIA. Support for Approximate Fast Foreground Colour Estimation for smooth background replacement.
LR-Anchored Multi-Reward Preference Optimization for Generative Real-World Super-Resolution
You can use [a/Popspaper](https://popspaper.github.io/pOps/) method in comfyUI
Unofficial Luma API-ComfyUI version.[w/WARN: This project is for learning purpose only!]
Rolling-overlap continuation nodes for ComfyUI's native MiniMax H3 AV latents
Nodes: Psd2Png.
Animated optical illusions in ComfyUI
Exact activation-only multi-GPU attention head sharding for MiniMax H3 in ComfyUI.
Advanced GRAG (Guided Region-Adaptive Guidance) implementation for ComfyUI. Features: Simple Controller (3 parameters) for beginners, Unified Controller (25+ parameters) for experts, Advanced Sampler with v2.2.1 contamination fix, Preset Manager with 54 curated presets, per-layer control, adaptive schedules, and multi-resolution support.
Integrates GLSL shader support.
This extension allows users to load styles from a CSV file, supporting Chinese. 在ComfyUI中加载中添加以下节点之一: - `Load Styles CSV` (加载风格CSV文件) - `Multi Styles CSV` (多风格合并节点) - `Preview Text` (预览文本节点)
Implementation of the Relay-VSR and LightX2V Relay-VSR-2x as comfyui nodes.
ComfyUI node to beautify your image generation prompt
This custom node for ComfyUI will add a simple and elegant resource monitor.
Turn natural language into validated Danbooru tags using a local LLM
Custom nodes for combining multiple LoRAs without overexposure on distilled models like Z-Image Turbo
An AI movie studio inside ComfyUI: a browser-based studio with timeline, per-scene prompts and one-click montage generation, plus a self-improving engine that learns from your ratings. Every feature is a module, so a broken one is simply absent.
Industry leading face manipulation platform
Nodes:Render SD3 Attention, SD3 Attention To Image, SD3 Image Into Attention.
Unofficial implementation of [a/OpenVoice](https://github.com/myshell-ai/OpenVoice) for ComfyUI
长视频导演台,面向 MiniMax H3。一键启动 WebUI 在 timeline 上创作!
A versatile and highly customizable node for ComfyUI to add camera-style watermarks and frames to your images. Whether you want to emulate the classic Leica look, add EXIF data, or create professional-looking framed images, this plugin has you covered.
ComfyUI node for DramaBox — expressive TTS with voice cloning (ResembleAI / LTX-2.3)
Workflow management and generation UI plugin. Browse, organize, and execute workflows with auto-generated parameter editing, AI prompt assistant (Ollama), JSON syntax highlighting, Eagle integration, and multi-language support (EN/JA/ZH).
A curated collection of reusable ComfyUI nodes developed by TGB. These sidecodes encapsulate key breakthroughs in model sampling, noise scheduling, and image refinement for enhanced stable diffusion workflows.
Official companion node for Image MetaHub. Saves A1111/Civitai-compatible parameters plus extended Image MetaHub metadata (workflow JSON, SHA256 model hashes, VRAM peak/total, GPU device, generation time, and steps/sec) for reproducibility and benchmarking.
Modular prompt engineering with wildcards, variations preview, auto-sort, and presets. Two modes (Simple/Extended), LoRA Manager integration, and 7 themes shared with FlowPath.
Qwen2.5-VL in ComfyUI
ComfyUI Node to work with Hugging Face repositories
Simple ComfyUI nodes to set the actual amount of noise in img2img, ensuring consistent behavior when changing schedulers.
Olm Liquify is a ComfyUI custom node for interactive image warping, inspired by Photoshop's Liquify tool.
Generate long MiniMax H3 videos from one prompt and a total duration. Segments are sampled with motion and audio continuation and joined into one video, with optional LLM prompt splitting, a Ref2VA prompt builder and the clip-chaining nodes. Based on NikoDemon80/ComfyUI-H3-Motion-Context.
This repository provides an unofficial ComfyUI custom node implementation of the paper TKG-DM: Training-free Chroma Key Content Generation Diffusion Model.
MOSS-Image edit LoRAs (chinoll's Moss-Image-Edit and MOSS-Image-Edit-HQ) on ComfyUI's native MOSS-Image model: edit one to four images with a text instruction, with ComfyUI's loaders, LoRA loader, samplers and memory management.
Performs various processing on images containing NSFW content within ComfyUI. The model used for detecting NSFW content can be selected.
This node enables better performance for Stable Diffusion models, by leveraging AMD MIGraphX, on Navi3 and Navi4 GPUs.
Warning, uses experimental package `comfy-env` to attempt a one click isolated install. Will download and use pixi package manager. ComfyUI nodes for Pixal3D (SIGGRAPH 2026, TencentARC) — pixel-aligned image-to-3D generation. Single image in, textured GLB out. Models autodownload from HuggingFace.
ComfyUI custom node for image captioning using Gemma 4 GGUF models via llama-cpp-python. (Description by CC)
A ComfyUI custom node that integrates with the sm4ll-VTON API for virtual try-on functionality.
[a/BizyAirPlus](https://github.com/siliconflow/BizyAirPlus) Run your workflow on cloud.
DEPRECATED. Install ComfyUI-UtilsCollection instead from https://github.com/silveroxides/ComfyUI-UtilsCollection
Drum pad node for ComfyUI with configurable 16/64 pad layouts.
NTSC/VHS analog tape effect for ComfyUI, with a live preview inside the node.
A collection of small utilities for ComfyUI: Auto Dark Mode, Detailed Job Status timer, Save Compressed Weppy (WebP with embedded prompt/workflow metadata), Skutils Resolution Calculator, Skutils Scale Image to Total Pixels (the built-in resize node with an optional image input), Skutils Text Preview (text persisted in the saved workflow JSON), and Skutils Style Selector (a bundled 434-style visual picker with positive/negative prompt output, combine/batch modes, and custom styles).
Fun-CineForge: A Unified Dataset Pipeline and Model for Zero-Shot Movie Dubbingin Diverse Cinematic Scenes
Training-Free Video Matting and Compositing via Latent Diffusion Models
Basic support for StyleGAN2 and StyleGAN3 models.
comfy extension for lumina2 TeaCache
Support both CosyVoice1.0 and CosyVoice2.0. Referenced [a/CosyVoice-ComfyUI](https://github.com/AIFSH/CosyVoice-ComfyUI), the following modifications have been made: Add support for CosyVoice2.0, Add whether to use stream processing options, Use speed control by CosyVoice, Add model path check to avoid duplicate downloads, Provide two ways of use
ComfyUI custom nodes for Gaussian viewer interaction and camera pose prompt control.
A custom node collection for ComfyUI that provides advanced face detection, alignment, and transformation capabilities using MediaPipe Face Mesh.
SAM-Audio (Segment Anything Model for Audio) node for ComfyUI - Separate any sound using text, visual, or span prompts
ToriiGate-0.5 image captioning nodes with local GGUF support for ComfyUI.
Most durable and flexible VAE decoder for ComfyUI designed for long video sequences with crash-proof dynamic batching, temporal scale detection, and aggressive out-of-memory recovery. (Description by CC)
Character-sheet prompt and layout designer for Qwen Image 2.1 with seven selectable views, per-part instructions, style selection, output dimensions, and a framed layout reference image. Uses character image references and prompting; no character-sheet LoRA required.
NovelAI V5 generation, refinement, references, Director Tools and tiled upscale nodes for ComfyUI
ymc 's nodes for comfyui. This extension is composed of nodes that provide various utility features such as text, region, and I/O.
Nodes: Image URL to Text, Image to Text.
ComfyUI nodes for MOSS-Image: storyboard grids (3/4/6/8/9), visual layout editor, face swap, and an LLM prompt node (caption / expand / edit / assemble) with up to 9 reference images and MiniMax-H3 output; also targets Flux, SDXL and Qwen-Image. No extra pip dependencies for the core nodes, MIT.
为 ComfyUI 引入 Google NanoBanana Gemini 文生图/图生图能力的自定义节点,提供批量生成、并发调度与余额监控等高级特性。
Auto translate all languages to english
HyperFlow: 8-step LoRA + two-time (t, r) conditioning for MiniMax-H3
The Text Queue Processor can split text into groups line by line for batch processing (unfinished, with noticeable bugs still being fixed). Currently, on the first run, it may only process one item. Once that is completed, running it again usually works fine. After each run, the `start_index` needs to be manually reset. Occasionally, strange issues occur, such as multiple queues being added at once, while other times it works perfectly fine. The potential causes are still under investigation.
Prompt-only Krea moodboard search and style nodes for ComfyUI.
The plug-in is designed to automatically save the association between the LoRA model and Trigger words to a Local JSON file so that when the LoRA model is loaded, the associated trigger words can be automatically loaded via the node 'LoRA Trigger Local' without manual input.
Using IPAdapter for style consistency, the node accepts a story structured as text {prompt} text {prompt} etc. and generates a comic, saving it to /output. It also adds LLM API Request node, providing an openai compatible LLM API for generating the stories.
Nodes to create small variations on noise, to shape noise, and to control noise in batches. Replaces the old 'variation-seed' nodes.
Load images in order(All other nodes are in the wrong order)! Using LLM and Joy tag pipeline to tag your image(s folder), it's suitable for train FLUX LoRA and also sdxl.
Using LLM and Joy tag pipeline to tag your image(s folder), it's suitable for train FLUX LoRA and also sdxl. Load images in order!
A custom node for ComfyUI to FramePack.
Fixes image batch mixing and missing filenames when using large multi-channel KSampler workflows.
CLIP text encoder with BREAK formatting like A1111 (uses chained ComfyUI conditioning concat).
Alt+H spotlight for ComfyUI: hide or dim every link that doesn't touch the selected node. Frontend-only, zero nodes, zero dependencies.
A set of custom nodes enabling LoRA support for LTX Video
ComfyUI Easy Padding is a simple custom ComfyUI node that helps you to add padding to images on ComfyUI.
Caching optimization for Chroma
Fix Nov. 2024, Davemane42's Custom Node for ComfyUI
This node improves the quality of the image mask. more suitable for image composite matting
Unofficial implementation of [a/LatentSync](https://github.com/bytedance/LatentSync) for ComfyUI
Advanced tiling of various shapes for ComfyUI
Extract LoRA from the original Fine-Tuned model.
This extension provides the ability to build prompts using wildcards for each region of a split image.
Concat_strings atm - celebrating first_node
A set of nodes to track objects through videos using YOLO and other processors.
This is an attempt to recreate the wildcards plugin for Automatic1111 but for ComfyUI.
Nodes:CLIPTextEncodeWithWildcards. This wildcard node is a wildcard node that operates based on the seed.
ScenarioPrompt is a custom node that helps you understand what you're prompting for each property as you build your prompts
WanVideo Integrated KSampler - Github: https://github.com/luguoli - K-sampler for video generation with integrated high/low noise stages,supports generating videos with start/end frames,automatic VRAM/RAM management, sage attention, block swapping, SD3 sampling
Memory-conscious video and image tiling nodes for ComfyUI, including disk-buffered processing and sampler timing.