ComfyUI Custom Extensions
Browse 9,446 ComfyUI extensions and 73,215 custom nodes.
Advanced RoPE modifications for Chroma/FLUX models including DyPE, YaRN, and other RoPE extension methods.
A ComfyUI Custom Node that enables arrow key canvas navigation with a pan speed setting.
A ComfyUI node that improves inpainting results by extending mask boundaries with geometric patterns, helping create smoother transitions and better context for AI-driven image completion.
Generate LTX-Video 2.3 optimized prompts using a local Ollama LLM. Drop-in replacement with 100% local processing, no cloud API keys needed.
This allows for mathematical operations on input images and precise manipulation of channels through NumPy formulas, making it suitable for ComfyUI users with programming experience.
Offline keyword-based SFW/NSFW prompt classification for ComfyUI.
Prompt workbench for ComfyUI: read original prompts, reverse prompts from images, OC edits, optional TIPO or LLM expansion, and adapt the result for Anima, Krea2 or Qwen.
A couple nodes to rewrite prompts using your local LLM (or other api I guess) because copy/paste is a lot of work. Currently only for MiniMax H3 Text to Video and Ref to Video.
Create a fully autonomous Telegram bot inside a single ComfyUI workflow. Simply copy your bot_token, hit run and that's it! The pack features a unified user message receiver (uses Long Polling) and a rich collection of media senders. The nodes are designed for continuous operation of the workflow under the Run (Instant) queueing mode.
Custom node that previews generated images on the node and pauses workflow until user confirmation.
focused on variation exploration with loading and saving within a project structure . load random files by string , multiply or randomize lora string strength , load and save to an external folderpath with filename safe outputs . load random checkpoint or load checkpoint from string matching . work in progress.
A visual pill-based prompt composer node for ComfyUI. Features 4 prompt panels (Quality, Main, Aesthetic, Negative), tag browser, drag-to-reorder pills, weight control, presets, Lucky prompt generator, manual override mode, and wildcard (.yaml) file loader.
A ComfyUI custom node package for Wan 2.2 Image-to-Video generation workflows.
Example of a Combo widget changing Node appearance
Post images and video to Discord. Nodes to facilitate communication using REST.
A ComfyUI extension for local administration, custom node management, backup and restore, and workflow metadata repair.
Analyze ComfyUI image, audio, and video lists with Ollama, llama.cpp GGUF, or native generative CLIP backends
GPU-aware queue autopilot (NVIDIA) plus a cross-GPU pause/cooldown/notify node for ComfyUI
This is a set of custom nodes for ComfyUI that provide anime-style image segmentation using efficient pre-trained models.
Saves images directly to OneDrive using Microsoft's free API service.
A custom node for ComfyUI designed to automatically clean and format natural language captions (generated by VLMs like Florence-2, CogVLM, or JoyTag) for LoRA training datasets.
Standalone Three.js adult character editor with immutable ComfyUI PNG snapshots
ComfyUI-native mixed-precision checkpoint converter with exact per-layer metadata.
Custom ComfyUI node for background removal in images. Built with ONNX models and Pillow (PIL).
Load Qwen-VL and Qwen3-VL models locally and apply PEFT LoRA adapters for enhanced image captioning. Supports 4-bit and 8-bit quantization (BitsAndBytes), FP8 pre-quantized models, Flash Attention 2, SageAttention, and torch.compile. Includes a LoRA strength slider and configurable captioning prompts.
This custom node allows you to convert trajectories from JSON files directly into WanMove TRACKS.
A ComfyUI node that converts images to pixel art using the neural network from the SIGGRAPH Asia paper Make Your Own Sprites.
Custom nodes for managing [a/Ollama](https://ollama.com/) models in ComfyUI workflows. Load and unload models on-demand to optimize memory usage in constrained environments.
Pin widget controls from any node to a persistent sidebar panel
ComfyUI node that lets you set the aspect ratio with the prompt. Converts megapixels + aspect ratio into width/height, with prompt-based ratio detection and randomization.
Nodes:SmartMask, ResizeCropFit, Percent Padding, SoftErosion, StringScheduleHelper, DStack, DavchaConditioningConcat, DavchaModelMergeSimple, DavchaCLIPMergeSimple, DavchaModelMergeSD1, DavchaModelMergeSDXL, ConditioningCompress... Some personal QoL and experimental nodes
vLLM nodes for comfyui using llama-cpp-python
Custom node for ComfyUI integrating audio-separator library that separates audio stems (vocals vs instrumental), removes noise, and fixes audio artifacts using GPU acceleration. (Description by CC)
Essential custom nodes for ComfyUI workflows
ascii art preprocessors in ComfyUI
Interactive JavaScript nodes: Load Video with real-time preview, Video Maker (image batch to MP4), plus upcoming queue management and dynamic UI tools. Enhances workflow flexibility.
A custom ComfyUI node for generating consistent character images using Ideogram API v3's character reference feature. Part of the Desert Pixel (DP) node collection.
A dynamic theme-based prompt generator for ComfyUI that creates versatile, random prompts optimized for face swap workflows.
A custom ComfyUI node that splits sampling steps between a main sampler and a refiner sampler, with independent denoise control for each phase, with upscalers now
Custom Node for ComfyUI that allows you set a weighted curve to your ControlNet giving you more control over the weight of the model over the course of generation.
Public-alpha ComfyUI node for Wan I2V/V2V cascade continuation. R178 adds the Tail 5 Continuation Gate: pause at cascade boundaries, inspect the source plus five tail-frame candidates, choose the best continuation frame, resume the same run, save one stitched video, and write a detailed report. Defaults stay conservative with Observe Only math, Report Only prompt analysis, no prompt text injection, and experimental research modes kept out of the public default path.
This is a node to convert workflows to cyuai api available nodes.
ComfyUI custom node package for bridging NAIA 2.0 prompts into ComfyUI workflows via its Remote Web API.
A set of custom nodes enabling Text-to-Video support for FramePack-F1
Validated MiniMax H3 prequantized INT8 ConvRot loader for AMD ROCm, preserving H3 mixed-precision boundaries.
Professional audio processing nodes for ComfyUI
Prometheus metrics exporter for ComfyUI with minimal dependencies and data gathered
Easy functional programming and flow control within ComfyUI
A collection of custom nodes for ComfyUI
Run SeedVR2 video upscalers on Tesla V100 (Volta) — fixes the fully-black-output fp16 bug with an FP16Safe-style txt-residual patch.
ComfyUI custom nodes to covert string/image/file to md5 hash
ComfyUI-Cleaner 文件目录清理插件,主要应用linux服务自动清理生产物文件
VFX-grade 2D tracking for ComfyUI: planar tracking, stabilize/destabilize round-trips, corner-pin compositing, mask propagation, track export
Interactive 3D camera-angle control for Fibo 2.0 Edit in ComfyUI: drag the camera around the subject and get a ready-to-use camera instruction.
A ComfyUI custom node that automatically extracts KSampler metadata and stamps it onto generated images.
Aggressive RAM and VRAM cleanup nodes for ComfyUI on Windows. Clears model caches, CLIP, checkpoints, VAE, and forces Windows to release physical RAM.
Compact MiniMax H3 directors with OpenRouter, Ollama, API and local LLM providers.
Extract, track and retarget character motion from video, entirely inside ComfyUI: video of a person -> cleaned side-view skeleton track (DWPose with feet, planted feet, one ground line) -> fitted to your character -> z-buffered cylinder pose frames for SCAIL-Preview.
Lots of randomizers, a simple oobabooga adapter with json options to pass, and other useful nodes.
Quality-of-life tools: open model folders from nodes, one-click download of a workflow's missing models, and unpack subgraphs. Works with Nodes 1 and Nodes 2.
ComfyUI custom node for image privacy protection using RGB channel independent differential bit-shifting encryption. (Description by CC)
Control different parts of your image with separate prompts using visual box drawing.
Minimal SAM3D body reconstruction toolkit for calibrated camera-shot workflows in ComfyUI.
RIFE video frame interpolation node for ComfyUI using real-time intermediate flow estimation to generate smooth intermediate frames between video frames with automatic model download and multiple precision support.
A monorepo for ERPK's custom ComfyUI nodes, extending ComfyUI's functionality through integrations with various AI services and APIs.
Text-to-image prompt enhancer node for ComfyUI using an OpenAI-compatible endpoint
Semantic image, video, and document search for ComfyUI using Qwen3-VL multimodal embeddings
Generate, Browse, Organize, Share and Re-Use ControlNet Maps
A clean, lightweight ComfyUI plugin to transmit data wirelessly between nodes using a global key-value store.
ComfyUI nodes for dynamic JSON loading and string/path utility operations
Companion add-on for the main SeedVR2 ComfyUI nodes, providing a native file-backed video path
Custom nodes for sharp frame selection and analysis. (Description by CC)
Use The F***ing Core Nodes — suggests core (or otherwise-available) equivalents for custom nodes, and replaces them across a workflow after a preview.
Automatically snapshots workflow state with a sidebar to browse and restore previous versions.
Genera custom nodes and extensions
Bring your ArcEnCiel models straight into ComfyUI with one click. Includes Link Key support, remote worker control, inventory sync, and sidecar generation.
ComfyUI custom nodes for LTX2 to R2V video generation. (Description by CC)
Nodes to manipulate HTML.
RelayVSR 4x real-world video super-resolution for ComfyUI, with automatic checkpoint download
NVIDIA RTX Video Super Resolution, denoise, and deblur for ComfyUI images and video frame batches
A suite of nodes for ComfyUI that helps making ComfyUI more accesible for artists
Panorama equirectangular to perspective rectify node with interactive preview for ComfyUI.
ComfyUI nodes that write MiniMax H3 T2VA and I2VA prompts from filmmaker fields and stills, via local Ollama or the xAI Grok API. Every node outputs h3_prompt.
Enhanced Load Image/Video nodes for ComfyUI with visual thumbnail browser, hover preview, draggable/resizable/maximizable modal
Image browser extension for ComfyUI with search, favorites, metadata view, compare mode, and media support.
A simple seed generator based on special number sequences: Fibonacci, Prime, Padovan, Triangular, Catalan, Pell, Lucas
ComfyUI-Fossiel-QoL-Nodes is (what will hopefully become) a suite of custom nodes for ComfyUI to improve quality of life. These nodes have been developed for personal use but maybe you’ll find them useful as well.
Suite of nodes providing VAE tiling functionality for WAN conditioners to reduce VRAM usage while maintaining generation quality. (Description by CC)
A ComfyUI custom node for automated face verification, designed to check if a person is clearly visible and suitable for passport-style photos. This node performs comprehensive facial analysis to ensure photo quality meets identification document standards.
A bunch of nodes that can be useful to manipulate primitive types (numbers, text, ...) Also some helpers to generate text and timestamps.
A set of ComfyUI nodes for calling the Nano-Banana API , while also supporting the Google API and frankAI API.
A Custom Node for ComfyUi to add an interactive Radar graph to visually control weights.
Transparent ComfyUI patch to load Diffusers-format LoRAs for Anima models.
Huggingface and Civitai model downloader nodes. Supports civarchive too.
A simple custom node for ComfyUI to load LoRAs and videos directly from a URL. Ideal for users hosting files on a server with publicly accessible URLs.
ComfyUI custom node wrapper for SenseNova-SI VLM inference.
Elemental (per-UNet-key) model merge and LoRA apply nodes for ComfyUI: synth-style knob matrix with sub-element tabs, MBW-style block sliders, text recipes, per-key LoRA strength control, and shareable merge presets (drag & drop JSON). Independent implementation inspired by SuperMerger's block/elemental merge concept.
Creates lora from two checkpoints by extracting the difference.
Various custom nodes; guidance, latents, sampling, tokenization, etc.
Unified prompt enhancement node: heretic VL model catalog (HF + GGUF), preset catalog, native ComfyUI CLIP engine, thinking separation and automatic model unloading.