ComfyUI Custom Extensions

ComfyUI-FastVLM-7B Apple A custom node for Apple’s FastVLM-7B vision-language model. This node lets you pass an image + instruction and returns a generated text response.

This custom node for ComfyUI allows you to scrape and download images and videos from the Midjourney showcase pages. It uses undetected_chromedriver to bypass anti-scraping measures, but requires session cookies from a logged-in browser session to function correctly.

Native ComfyUI implementation of ByteDance SeedVR2 using ComfyUI's own model stack and core graph nodes.

A net tool node for comfyui, rewrite from [comfyui-tooling-nodes](https://github.com/Acly/comfyui-tooling-nodes) but support more big data sending.

ComfyUI prompt word template management plug-in. Instead of adding any nodes to the canvas, a "Prompt Word Assistant" button is provided in the operation bar to open a full-screen modal box to manage the prompt word template with variables. After filling in the variables, copy the merged results with one click.

a simple resize image node(s) in comfyui. This repository is not specific to the length and width values of an image, but focuses more on adjusting the total pixel count, side length specifications, and size ratios of the image.

A collection of custom ComfyUI nodes and utilities for generating AI image prompts representing the diverse attire, cultures, regions, and appearances of the world. This project is designed for easy extension to new countries, cultures, and body parts, using a modular JSON-based data structure and dynamic node generation.

Scene Viewer of SpatialGen

Utility nodes including ImageBlend, Math, Strings, Color tools, Wildcard Processor, Latent Noise Mask, and WAN Three Frame To Video for keyframe-based video generation

Node for ComfyUI designed for more neatly switching between tiled and default VAE Decode Nodes.

Bboxes (Bounding Boxes) handling and visualization

This project makes local LLMs easy to use for prompt enhancement and image captioning –No API keys. No external tools. No headache.

Automatically detect and download missing LoRAs for ComfyUI workflows

Assorted custom nodes with a focus on simplicity and usability including watermark node and others focused on customizing my comfy experience.

This custom node allows you to overlay a watermark image onto an existing image within ComfyUI.

The ultimate tool for creating complex iterative workflows and frame-by-frame video processing in ComfyUI, designed to handle huge tasks without GPU crashes using intelligent sequential looping.

Run ComfyUI workflows on RunPod Serverless

ComfyUI's Smart Memory Management efficiently manages RAM, GPU memory, and garbage collection. This feature keeps frequently used models in memory to increase inference speed, and dynamically releases less important models when memory is low to optimize resources. However, not all ComfyUI custom node developers create nodes that are compatible with Smart memory management. This includes several impressive models. Mem-safe-wrapper is a custom node that wraps these model nodes to enable ComfyUI's Smart memory management capabilities.

Describe image or create caption files using Janus Pro Model

ComfyUI custom node Pushover Notification

Send generated image to PushOver API webhook with optional parameters such as prompt-id and metadata payload.

This is a tool that lets you digitally sign your images to prove you created them. Think of it like putting an invisible, tamper-proof seal on your artwork that says 'I made this.'

An OCR node for detect text in image and returns covering mask.

This node can easily switch between models and prompts by saving presets.

Simple ComfyUI Custom Node to enable saving checkpoints with V-Pred ZTSNR tensors and metadata.

Simple node to convert lora_stack output into string to generate a1111-style lora strength output text, like <lora:name:0.69>. Used to include your Lora names from Lora Stack in text input-output nodes and so on.

The FLUX model API from DashScope, developed by Black Forest Labs, offers superior image generation capabilities with optimized support for Chinese prompts, achieving a commendable tradeoff between performance and the quality of generated images compared to other open-source models.

These nodes allow you to use a distilled Qwen3-4B model as your text encoder in Stable Diffusion XL.

ComfyUI node that makes AI-generated images look like iPhone 15 Pro photos by emulating Apple's ISP color science pipeline, including tone mapping, Display P3 color, Smart HDR, and sensor effects. (Description by CC)

Help comfy deisgner develope custom nodes by foreground GUI without any coding knowledge, complementing the workflow what you design with LLM automatically

A small node suite for ComfyUI

Deterministic Prompt Enhancement via Procedural Augmentation for ComfyUI

Drift Johnson's XTTS nodes

AI dance choreography from music. Generate full-body dance motion (BVH/FBX/JSON) from any audio track using MVNT's diffusion model trained with 100+ professional choreographers. Create rigged 3D characters from images, preview motion as stick-figure animations, and export photorealistic AI music videos — all inside ComfyUI. Nodes: MVNT Generate Dance, MVNT Generate Character, MVNT Export Video, MVNT Preview BVH, MVNT List Styles, MVNT Estimate Cost, MVNT Load Motion.

Based off of: [a/Birch-san/diffusers-play/approx_vae](https://github.com/Birch-san/diffusers-play/tree/main/approx_vae). This ComfyUI node allows you to quickly preview SDXL 1.0 latents.

Advanced custom samplers and schedulers for ComfyUI, ported from the Stable Diffusion WebUI reForge extension.

A custom node set for ComfyUI that provides nodes for encoding and decoding images using Tiny AutoEncoders for Stable Diffusion (TAESD) models.

Russian localization of ComfyUI, ComafyUI-Manager & more...

Some additional audio utilites for use on top of Sample Diffusion ComfyUI Extension

A ComfyUI integration for [a/ChatTTS](https://github.com/2noise/ChatTTS), enabling high-quality, controllable text-to-speech generation directly in your ComfyUI workflows.

This repository adds ComfyUI custom nodes that wrap the Ovis-U1 multimodal model, exposing three primary workflows inside the ComfyUI editor.

ComfyUI node wrapping qwen3-tts-triton for 7-mode Qwen3-TTS inference (Triton kernel fusion + TurboQuant KV cache)

Reusable tagged image and audio references for MiniMax H3 video workflows in ComfyUI. Add as many references you want and save them in the directory. When you tag them in your prompt, the node auto injects them to use as references and changes the tag to your saved reference information.

Automation toolkit for ComfyUI featuring dynamic tag combinations, LoRA parsing, and workflow metadata reading.

A custom node for ComfyUI that generates creative and detailed prompts using OpenAI's GPT models.

Collection of nodes for ComfyUI for automation

Pixel Anchored Remaster for ComfyUI.

Concept Steering for ComfyUI — steer image generation with learned concept directions via SAE + DPO

JoyAI-Image-Edit (16B MMDiT + Qwen3-VL-8B) instruction image editing for ComfyUI.

MiniMax H3 for ComfyUI: one drag-and-drop input node covering t2va/fl2va/ref2va, plus step caching, partial decode and profiling.

Multi GPU parallel queuing for ComfyUI. Keep all GPUs busy from one ComfyUI session by automatically dispatching workflows across every CUDA device.

A simple and easy to use Hugging Face model loader.

S3-backed input/output nodes for ComfyUI to browse, load images/videos from S3 buckets and push outputs back to S3 automatically.

A ComfyUI custom node that parses prompt text for LoRA tags and visualizes their metadata, including trigger words, strength values, thumbnail previews, and example images.

Composable JSON-building and schema-validation nodes for ComfyUI

This is a node to run regex for strings.

GigaAM custom nodes for ComfyUI by OreX

A custom node for ComfyUI that allows you to cache an image and its mask to avoid recalculating upstream steps in your workflow.

Nodes: BinaryImageMask, ImagePadding, LoadLastCreatedImage, RandomMask, TransparentImage.

Cut and and drag that allows you to cut and drag multiple images on a path

Custom node for ComfyUI using Google's Gemini 2.5 Flash Image and Gemini 3 Pro Image Preview APIs to generate images with resolution scaling for 1K, 2K, 4K, 8K. (Description by CC)

ComfyUI wrapper nodes for OmniVoice multilingual zero-shot TTS.

Reproject an image or clip through its MoGe geometry along a camera path authored in an interactive 3D editor.

Reference-image prompt engineering workflow for still images and MiniMax H3 video, with built-in aspect-ratio detection.

Quicker connections, Circuit board connections

Custom nodes for ComfyUI QWen3 8b running based on llama.cpp, which only support the CUDA framework and do not support MPS.

An enhanced Vision-Language Model (VLM) conditioning node for ComfyUI designed for the Qwen-Image series, supporting up to 5 reference images with customizable system prompts and token tracking.

Warning, uses experimental package `comfy-env` to attempt a one click isolated install. Will download and use pixi package manager. ComfyUI nodes for MoGe-2: monocular metric geometry estimation (depth, points, normals, intrinsics) from Microsoft Research (CVPR'25 Oral).

EA collection of ComfyUI custom nodes for formatting and debugging string data with the intention of collecting generation data to be processed by a custom node pack like comfy-image-saver, as well as miscellaneous extra nodes to experiment with.

PainterImageFromBatch is an enhanced ComfyUI custom node for extracting contiguous frames from image batches.

ComfyUI node for saving, previewing, and passing images downstream while supporting mask editing and intermediate node execution. (Description by CC)

Automatically scales your input video to specified dimensions and encodes it into latent space for sampling.

This node is mainly based on [a/OmDet](https://github.com/om-ai-lab/OmDet) for object detection, and it outputs related images, masks, and Labelme JSON information.

Utilities to make it easy to develop advanced Workflows without having to use a lot of nodes for simple stuff.

A package designed to enable multi-regional prompting for architectural rendering, integrated with the Rhino Pseudorandom plugin.

A collection of ComfyUI custom nodes designed for image batch processing, per-index image operations, and AWS integration using EventBridge.

Simple latent-passthrough node for running a full VRAM cleanup between workflow stages.

A ComfyUI custom node for DreamOmni2 GGUF multimodal models — powered directly by llama-cpp-python.

Nodes:LFO_Triangle, LFO_Sine, SawtoothNode, SquareNode, PulseNode. ComfyUI custom nodes to create Low Frequency Oscillators.

NVIDIA-first hardware telemetry chip + popup for ComfyUI. Linux/Docker-friendly with explicit accounting for VRAM used by other containers/host.

A custom node for ComfyUI that leverages the Google Cloud Vertex AI Imagen API to generate and edit images.

A custom node for ComfyUI that provides seamless integration with the Wan models from Alibaba Cloud Model Studio. This solution delivers cutting-edge image and video generation capabilities directly within ComfyUI, supporting both international and Mainland China regions.

Lightweight random selection node for rapid prompt variations with independent multiple-node support and wildcard processing capabilities. (Description by CC)

this node contains a lot of small little helpers like switches, passers and selectors that i use a lot to build my workflows.

A centralized wrapper of all MediaPipe vision tasks for ComfyUI.

A set of ComfyUI nodes for animating a foreground layer over a background, with unified timing and high-quality composition. It includes effectors for rotation, position, path-following position, scale, opacity, distortion, and masking, plus a core node that composes frames and exports APNG/WEBP and a frame sequence.

A custom node package for image processing in ComfyUI, for automating image processing within the ComfyUI workflow.

Text rendering and styling nodes for ComfyUI. This extension provides a basic text renderer, multiple font loaders, and a style node that adds stroke, shadow, gradient fill, and opacity control.

ComfyUI Workflow to run audioldm-l-full pipeline [a/https://huggingface.co/cvssp/audioldm-l-full](https://huggingface.co/cvssp/audioldm-l-full)

A ComfyUI custom node that provides quick access to all image resolutions for the Z Image Turbo model, sourced from its official Hugging Face Space.

This node adds improvement over some already existing nodes as well as some custom nodes.

Utility node for converting a .webp format image into sequential png images in ComfyUI

Spotlight: An extensible, magical, and essential search bar for ComfyUI

[a/openai Consistency Decoder](https://github.com/openai/consistencydecoder). After downloading the [a/OpenAI VAE model](https://openaipublic.azureedge.net/diff-vae/c9cebd3132dd9c42936d803e33424145a748843c8f716c0814838bdc8a2fe7cb/decoder.pt), place it in the `model/vae` directory for use.

Nodes:VAE Encode Inpaint, VAE Decode Inpaint, ColorCorrection Inpaint, ImagePreprocess Inpaint, ImagePostprocess Inpaint, Load Model Inpaint, Inpainting (using Model)

ComfyUI custom node for experimental Anima/Cosmos flow-slider LoRA training.

A Google Moogle is a Google Translator node for ComfyUI that provides easy-to-use text translation capabilities directly within your ComfyUI workflows.

Custom ComfyUI nodes designed to interface with a separate LM Studio instance for language model operations.

A collection of custom nodes for ComfyUI designed to replicate certain visual elements and effects reminiscent of Snap and early digital aesthetics. These nodes are ideal for image processing tasks and are fully compatible with ComfyUI.

Training-free Chebyshev feature forecasting for MiniMax H3 in ComfyUI