ComfyUI Custom Extensions
Browse 9,446 ComfyUI extensions and 73,215 custom nodes.
A custom ComfyUI node that generates comparison grids in a single node.
A ComfyUI custom node plugin for assembling prompts. It allows you to generate positive and negative prompts by selecting different options.
A ComfyUI custom node that uses the Gemini API to rewrite a raw prompt into one optimised for a specific image generation model and workflow.
Custom nodes for ComfyUI to automatically send generated images to Discord channels.
Custom ComfyUI node providing isolated Python script execution within persistent Docker containers with security isolation and the sidecar pattern for performance.
A visual prompt builder node for ComfyUI. Build your prompts by clicking categorized word chips, set token weights, save presets — all from an interactive modal.
Fold negative prompts into positive prompts for use with Turbo models. (Description by CC)
ComfyUI-ParamNodes is a lightweight, dependency-free collection of custom nodes for ComfyUI, designed to parameterize your workflows for API-driven generation. It provides a clean set of input nodes to control strings, numbers, booleans, and model/LoRA selections, along with a simple logic switch for conditional execution.
This will add a restart button to ComfyUI home menu.
NODES: Load Camera From File, Load 3D, Load 3D - Animation, Preview 3D, Preview 3D - Animation
ComfyUI-Classifier is a custom node for ComfyUI that uses a zero-shot classification model to classify text inputs based on a set of candidate labels. This node leverages the power of Hugging Face Transformers to provide accurate and flexible text classification.
A very basic ComfyUI node for ScenemaAI that requires a local docker container instance available to ComfyUI.
SEGA anima is implementation of "SEGA: Spectral-Energy Guided Attention for Resolution Extrapolation in Diffusion Transformers" (arXiv:2605.22668) for Anima model in ComfyUI.
Run ComfyUI's text encoder on a second PC (or a second GPU) and send the conditioning over the network, so the GPU that renders never loads the encoder.
A powerful tool to modify the CFG
ComfyUI nodes for using SimpleTuner-trained Flux 2 LoRA models. SimpleTuner's Flux 2 architecture uses fused layers, making its LoRAs incompatible with standard ComfyUI Flux nodes. (Description by CC)
This is the rgb2x and x2rgb wrapper node for ComfyUI. The required models are automatically downloaded on the first run. original project : [a/https://github.com/zheng95z/rgbx](original project : https://github.com/zheng95z/rgbx)
ComfyUI custom node implementing SMC-CFG (Sliding Mode Control Classifier-Free Guidance), a nonlinear control-theory approach to CFG that prevents instability and semantic overshooting at high guidance scales. (Description by CC)
A ComfyUI custom node for generating AVIF animations with transparency support using avifenc.
SoX and BS-RoFormer nodes for ComfyUI. SoX for sound editing, BS-RoFormer for audio stem separation.
ComfyUI precision flight instruments for latent space navigation, vector steering, and structural stabilization.
Nodes for lama+refiner inpainting with ComfyUI.
Spatially tiled sampler for MiniMax H3 audio-video generation.
ComfyUI custom node pack for videorouter.sh — hosted image/video/3D/audio generation nodes + Partner Node billing relay (Kling, MiniMax, Veo, and more)
Adds `XY Input: LoRA Weight (simple)` for Efficiency Nodes. Outputs XY subtype 'LoRA' (name, weight, weight) with a linear sweep; works with `XY Input: Checkpoint`.
Community ComfyUI nodes for Volcengine AI MediaKit
A simple mosaic blur node for ComfyUI that uses CV2 or Pillow
A comprehensive ComfyUI extension toolkit providing nodes for algorithm calculations, Base64 encoding/decoding, image processing, and workflow utilities to streamline your ComfyUI workflows.
ComfyUI String Similarity Node - Advanced text comparison with multiple algorithms
Model analysis nodes for ComfyUI (SDXL): a block x element diff heatmap between two models, an auto-merge recipe generator from that diff, and a block ablation analyzer that measures which blocks actually drive a given image. Pairs with ComfyUI-RecipeMerge for diff-driven merging.
intelliPrompt - An intelligent prompt optimizer for ComfyUI that fixes typos, balances parentheses, and enhances prompts
A powerful multi-model image captioning node that supports both local BLIP models and the cloud-based Google Gemini API, specifically designed for ComfyUI.
This is a custom node suite for ComfyUI that automates the conversion of web content into an AI-refined news script. This simplified version focuses on the core 'fetch-and-process' workflow for a fast and direct automation experience.
Helps deciding different settings depending on the input string
Lets you send a section of an image to qwen edit or kontext to help isolate areas in need of change
Custom ComfyUI nodes for running Z-Image Turbo with SDNQ quantization (q8/q4). Supports text-to-image, image-to-image, inpaint, ControlNet, and LoRA.
NODES:Image Selector (geocine), Image Scale (geocine)
A custom ComfyUI node for saving generated images directly to Cloudflare R2 (or S3-compatible) buckets using boto3, with secure random filename generation and public URL return.
These are two ComfyUI nodes used for converting input prompts into the Danbooru tag format (tag_a, tag_b, ...) and outputting the converted prompts.
These nodes identifies specific content in the input prompt (each content occupies only one line) based on txt files containing predefined content that needs to be recognized. It then adds a special prefix (as a marker) to the identified content and inserts it back into the original prompt.
A self-contained ComfyUI custom node for BiRefNet background removal. Does not depend on other third-party extensions; only requires model weights from Hugging Face.
在 ComfyUI 中调用 CYAI 中转站 (cyai.club) 的 doubao-seedance 视频生成模型(火山方舟 Ark 原生接口),含文生视频/首尾帧/多图参考,输出 VIDEO 对象
A ComfyUI custom node that estimates the Field of View (FOV) and tilt angle (horizon angle) of images using computer vision techniques.
GetFileNameFromURL is a ComfyUI custom node that extracts the filename from a URL. It can handle various URLs and is capable of handling redirects.
Nodes:Private ImageCPostprocessor
Add nodes for interacting with Ollama through the Open WebUI API.
Mask to SAM2 segmentation ComfyUI custom node
Custom nodes for ComfyUI to enable face swapping using the Roop library.
ComfyUI node for managing latent resolution and steps profiles
Nodes:NNLatentUpscale, A custom ComfyUI node designed for rapid latent upscaling using a compact neural network, eliminating the need for VAE-based decoding and encoding.
A small toolbox gadget for Appmode in ComfyUI that shows terminal information, restart button, VRAM usage and clear VRAM button.
A Countries Border distance mapper
A visual ComfyUI latent canvas dimensions selector
A Qwen 2/2.5, Moondream and SmolVLM LLM caption/prompt genrator
Misc collection of nodes. Some are just slightly improved versions of the native ones.
Floating panel that lists every group in your ComfyUI workflow — jump to any group, toggle on/off via rgthree Fast Groups Muter, reorder via drag-and-drop, configurable keyboard shortcuts. Designed for large multi-stage workflows.
A ComfyUI node that runs a [a/JMESPath](https://jmespath.org/) query against input JSON and outputs the result.
A ComfyUI node that runs a [a/jq](https://jqlang.github.io/jq/) query against input JSON and outputs the result.
A ComfyUI extension that improves panning and zooming on trackpads and with the mouse wheel.
A Pause / Resume button for the ComfyUI queue: the current job finishes, nothing else starts until you resume. No workflow nodes, no queue hijacking, no database.
Composite a masked dancing person video onto any background image with foot-anchored placement and perspective-aware scaling — make your dancer appear anywhere.
Seperation the voice to 2 track/person by voicesep, keep the office effect in Modelscope demo meanwhile accelerate the resample speed.
Comfyg Queue is a custom node for create multiple queues at same time, but with some config variations.
ComfyUI custom nodes for Nano-Banana (nano-banana-pro / nano-banana-2) and GPT-Image-2 image generation and editing, via the Nerapi API gateway. Text-to-image and image-to-image with up to 8 reference images.
Nodes:Abracadabra Summary, Abracadabra
Procedural image generation toolkit designed as input to image edit workflows.
This node improves the merging of LoRA for movements and physical resemblance when adding multiple LoRA to a model.
Transcribe videos or audio with Whisper, either locally or via OpenAI's API
Add a button that clears the console
ComfyUI custom nodes for uploading and loading images from S3 buckets
Utility nodes for ComfyUI featuring a dynamic Text Preset Switch with workflow-safe persistence and Node 2.0 support.
only show dropdown of loras in a given subdirectory
Just a Sampler and another settings pipe for comfyui
Nodes:Download Dreambooth Checkpoint, Get Random Value From List, Load Canny Pose Face, Transparent to White Background, Download Flux Lora.
Review-gated CivitAI publishing for ComfyUI images and native video
Measure where a LoRA stores its learned change, per transformer block, then apply only the blocks you choose. Read-only: your .safetensors is never rewritten.
A custom ComfyUI node for managing image generation prompts using Pinecone vector database with OpenAI embeddings. Features interactive approval with image preview for quality control.
ComfyUI custom nodes for image-to-video and portrait workflows: synced remote values, chainable prompt switch, private image previews, temporal color match and skin retouch.
Extract CLIP and VAE models from a loaded checkpoint in ComfyUI.
Media input and library, reference images, resolution tools, local prompt assistants, MiniMax H3 adapters and audio mixing for ComfyUI.
A universal ComfyUI timer with real-time duration tracking and dynamic unit mapping for any generation pipeline.
TrackedPoseAndFaceDetection - PoseAndFaceDetection con selezione manuale dei punti di tracking e optical flow (Lucas-Kanade). Richiede ComfyUI-WanAnimatePreprocess.
Pack labeled text through one clean ComfyUI connection and automatically rename unpacked outputs to match.
Add ASS/SSA subtitle to video using ffmpeg.
Unofficial community integration, not affiliated with the Causal Forcing authors: runs the official Causal Forcing++ frame-wise 2-step and Causal Forcing frame-wise 4-step checkpoints (Wan2.1-T2V-1.3B based) in a separate WSL2/Linux runtime, with real VIDEO outputs and step/timing reports.
SemBridge speech, singing and mixed speech+singing generation for ComfyUI: the official ASLP-lab model, run in-process, 44.1 kHz AUDIO output, verified sample-for-sample against the official code.
Unofficial ComfyUI nodes for Xiaomi-Lens-0 (4B, Apache-2.0): masks and boxes for the objects a text describes, including reasoning requests ("what would you use to stir the drink?"). Same output as the official inference code, bit for bit. Masks at the input image's size for inpainting and cut-outs.
A ComfyUI custom-node pack for abstract glitch art with three nodes that perturb diffusion sampling: sigma-schedule jitter, attention-map value dropout, and temporal frame interleaving for video latents.
Record Chrome tab or system playback audio and output standard ComfyUI AUDIO, with saved recordings and rolling snapshots.
Save Image/Latent to Shared Memory
Run a Photoshop droplet on your renders from ComfyUI, bridging WSL2 → native Windows Photoshop (no Wine).
Compatibility fixes and reusable workflow utility nodes for ComfyUI
XPU/CUDA VRAM cache clean nodes (soft/hard) for ComfyUI, inspired by Qwen3TTS-XPU cache_clean. Fixes VRAM_Purge's dead torch.cuda branch on Intel Arc. Prevents UR_RESULT_ERROR_OUT_OF_RESOURCES (level_zero error 40) OOM.
ComfyUI custom node for adding watermarks to videos using FFmpeg, with configurable position, opacity, scale, and hardware acceleration support.
A FastGAN Node for ComfyUI
A ComfyUI custom node for downloading and managing LoRA models directly within the UI.
A ComfyUI custom node for downloading and managing upscaler/super-resolution models directly within the UI.
Text-to-sound-effect nodes for MOSS-SoundEffect v2. Picks a local model and runs inference in a compatible Python, creating one beside the node when needed.
ComfyUI nodes for turning any picture into a full PBR material.