ComfyUI Custom Extensions
Browse 9,446 ComfyUI extensions and 73,215 custom nodes.
Isolate speakers in audio using Pyannote mode.
Camera viewfinder & REC overlay generator for images and videos with retro UI effects in ComfyUI.
NODES:Object Mask. NOTE:push [a/yolov8x-seg.pt](https://github.com/ultralytics/assets/releases/download/v8.2.0/yolov8x-seg.pt) in models/yolo
NODES:ScaleToTargetMegapixels.
A custom node for ComfyUI that enables automatic checkpoint rotation during batch image generation. This allows you to generate images using multiple different model checkpoints in a single workflow, automatically switching between them based on your specified interval.
Scatter stamp/sticker shapes over a mask instead of mosaic. Built-in white or black-on-white presets can be tinted; custom PNGs keep their own alpha.
A better node for using Google Gemini models, supporting the latest API endpoints with a v3 ComfyUI node interface.
ComfyUI Taylor attention custom node
A custom node for ComfyUI which simplifies the process of masking multiple prompts, making it easier to manage scenes with multiple distinct characters.
The face restore node for ComfyUI, based on RestoreFormer
Two simple nodes that help you author sequence-friendly prompts from a single script that contains multiple clips. (Description by CC)
A complementary tiling implementation for SeedVR2 that provides advanced VRAM-aware image tiling, seamless stitching, perceptually accurate color matching, and artifact-free sharpening for ComfyUI workflows.
A suite of custom nodes for ComfyUI by Ha Manh Tuyen
A collection of custom nodes to help with saving images, providing generation parameters, static literal nodes, and other useful nodes.
Image to textured 3D asset on AMD Radeon and Instinct GPUs (ROCm), Windows and Linux: TripoSG image-to-mesh and Hunyuan3D-2.1 PBR texturing as ComfyUI nodes, with no CUDA or C++ extensions to build. Also paints meshes from ComfyUI's own Hunyuan3D-2.1 nodes. Setup: install here and restart (Manager installs the Python packages from requirements.txt; nothing pins torch, so your ROCm PyTorch stays), then open Templates > Extensions > comfyui-3d-rocm. TripoSG and the painter download their models (18 GB) on the first run.
This is a custom node for ComfyUI that automatically saves your AI-generated images specifically to [a/www.txt2any.com](http://www.txt2any.com/).
Tools for scaling images and latents appropriate to SD3 in ComfyUI.
Fetch top tags from Safebooru and Danbooru for ComfyUI.
VRAM-safe VAE decoder for video latents in ComfyUI. Decodes 5D video latents in temporal batches to prevent GPU memory explosions.
S3 Storage Utilities that do not rely on environments to interact with S3 instances. Work with your buckets directly from within the workflow.
String Chain: Reconnect your prompts
A Wireless, Nodes 2.0 Ready, and Aesthetic Toolkit for ComfyUI.
Compose region-aware prompts from a 3x3 grid with an in-app visual editor, for rule-of-thirds spatial composition in ComfyUI.
Standalone ComfyUI custom nodes for stacking multiple LoRAs in a single, dynamic node — including an orthogonal stacking mode that reduces interference between LoRAs.
Custom ComfyUI node supporting wildcard text processing from files with YAML tags, randomization, latent ratio selection and preset aspect ratios.
A set of custom nodes developed by SunxAI for ComfyUI, including image loop processing and more.
Audio analysis nodes for ComfyUI. Detects BPM and musical key from audio files.
AST-based prompt parser with inline negative routing, smart percentage wildcards, and visual debugging
A set of tinkering nodes as well as QoL nodes like video scrub, image crop, rotate, sort loras by recent, show lora triggers
Get Mask Dimensions
Nodes:Image Dimension Resizer, Image Sizer, Random Ratio, Show Text, Random Title Character, Random Wildcard Tag Picker, Random Show Atm Loc Outfit, Contains Word, Elements Concatenator, ...
Velvet Vice custom nodes and interface integration for Krea 2 ComfyUI workflows.
Word-level forced alignment for the Synthy/iRead custom TTS provider that aligns generated speech against known input text and returns per-word timings.
Universal grid image splitter for ComfyUI that automatically detects grid layout by analyzing brightness/contrast and splits any grid image into individual frames.
ComfyUI custom node that acts as a mini video editing studio, automatically assembling multiple video clips from a folder with intelligent resolution resizing, aspect ratio handling, and frame rate management.
OpenAI-compatible LLM client nodes for ComfyUI (any provider: llama.cpp, vLLM, NInfer, Strata, OpenAI)
An advanced multi-ratio Bayer matrix dithering node for ComfyUI with edge protection.
A ComfyUI custom node for normalizing depth maps to a specific range (0–230 by default).
ComfyUI V3 custom nodes for [SoPro](https://github.com/samuel-vitorino/sopro), a lightweight multilingual text-to-speech model with zero-shot voice cloning.
Splittable canvas grid frames with drop-to-fit node stacks for ComfyUI
ComfyUI extension for parsing Civitai PNG workflows and automatically downloading missing models
A collection of custom nodes for ComfyUI that enable progressive blending and color matching effects across image batches/video frames.
VRGDG ComfyUI nodes for face repair, image and video utilities, local LLMs, long video, LoRA training, LTX and MiniMax tools, music, audio, and workflow UIs.
Cultural art evaluation nodes — L1-L5 scoring across 13 traditions, Layered Generation + Inpainting, Brief-driven creative workflows. Evaluate AI-generated art for cultural accuracy and get actionable improvement suggestions. Requires: pip install vulca
Cultural art evaluation nodes — L1-L5 scoring across 13 traditions, Layered Generation + Inpainting, Brief-driven creative workflows. Evaluate AI-generated art for cultural accuracy and get actionable improvement suggestions. Requires: pip install vulca
Nodes:color2RGB
Advanced LoRA management and prompt building tools for WanVideo. Features: Wakawave LoRA Loader (unlimited LoRAs, save/load presets, drag-to-reorder, per-LoRA strength control), Wakawave Prompt Builder (dual positive/negative prompts, weight control, segment mode, visual preset browser with import/export).
ComfyUI node for [a/Abstract Image Generation](https://github.com/wakattac/ComfyUI-AbstractImaGen/). This node is designed to create unique abstract base images on the fly within your ComfyUI workflows, which can then be used as input for VAE encoding, image-to-image generation, or other creative processes.
A ComfyUI custom node collection for image processing.
A custom node allowing to save images with CIVITAI readable datas
Local prompt enhancement for ComfyUI powered by llama.cpp, run LLM models natively to refine & generate image prompts.
Ultra-Efficient Flow Matching sampler ported from [a/https://github.com/eddyhhlure1Eddy/ode-t1](https://github.com/eddyhhlure1Eddy/ode-t1), suitable for native ComfyUI workflows.
RVC (Retrieval-based Voice Conversion) nodes for ComfyUI (AI voice changing via external RVC portable environment)
Interactive 3D camera control node for ComfyUI that precisely controls horizontal, vertical angles and zoom, with output of professional cinematography prompts. (Description by CC)
Mode-aware Chinese director-draft Context-IR compiler and hard auditor for MiniMax H3 ComfyUI workflows
WFA Node Suite (WaifuArchitect) for ComfyUI: High-performance utility and refiner nodes including Batch Prompt Loader & Interactive Label Filter UI, Loop Iterator & Auto-Shutdown, 100-input lazy Index Switch, Find & Replace, and Full-Image Refiners.
A simple and efficient custom node that generates random numbers between two user-selectable values, perfect for seed variation and parameter randomization in ComfyUI workflows
This custom node allows you to generate personalized video messages (PVM) using audio, video, and multilingual text prompts inside ComfyUI.
This custom node allows you to perform audio-video lip synchronization inside ComfyUI using a simple interface.
This custom node is based on the [a/rembg-comfyui-node](https://github.com/Jcd1230/rembg-comfyui-node) but provides additional functionality to select ONNX models.
Quickly add watermarks to pictures with batch processing, greatly improving work efficiency for bulk watermarking tasks. (Description by CC)
A lightweight ComfyUI custom-node pack for the Kepri Background Removal V2 pipeline containing mask merge and image finalization nodes.
Nodes: InvisibleWatermarkEncode
Nodes:Ignore Braces, Settings Switch.
A video editing studio for MiniMax H3 in one ComfyUI node: write clips in plain language, get format-valid prompts, render picture and sound, carry motion from clip to clip, join a film.
This project is a ComfyUI version of [a/https://github.com/cozheyuanzhangde/Invariant-TemplateMatching](https://github.com/cozheyuanzhangde/Invariant-TemplateMatching).
Stop hand-crafting prompts. Pipe your ideas through an LLM right inside ComfyUI — works with local backends (Ollama, vLLM, no API cost, full privacy) or cloud APIs (OpenAI, Claude via OpenRouter, Gemini), and ships text-to-image, multi-reference vision-to-prompt, pipeline composition, and AI/manual switching nodes.
A inversed euler sampler to revert image to noisy latent. Can be used to improve content consistency when perform I2I.
Simple nodes to help clean up your workflow, mostly focussed on text operations.
A simple node for detecting vrediction, outputs a boolean value, used for logical switching (for example, automatically adding rescalecfg when vrediction model is detected), recommended to be used with if else and other nodes.
Use the point editor to perform image composition editing.
Native custom nodes to initialize and run original Latent Diffusion Model (LDM) F8 Large architecture from CompVis with a BERT text encoder.
Fetch booru tags inside ComfyUI and inject them into your prompts
Run Wiro AI models from ComfyUI: one node for any of the 1,000+ image, video, audio and LLM models in the Wiro AI catalog, plus nodes for Nano Banana Pro, GPT Image 2, Seedream 4.5, Veo 3.1, Kling 3, Seedance and ElevenLabs speech, music and transcription. Results come back as IMAGE, VIDEO, AUDIO and text. Cloud API: needs your own Wiro AI API key and runs are billed to your Wiro AI account; no local model or GPU. No extra dependencies.
Adds a button to the ComfyUI menu to toggle full screen. That's literally it.
A collection of mostly random nodes of random usefulness.
A comprehensive outfit generation system for ComfyUI with AI-powered prompt enhancement and dynamic outfit composition.
ChatterBox Turbo TTS node for ComfyUI — fast 350M-param text-to-speech with emotion tags
ComfyUI 采样加速日志:每次运行自动汇报实际注意力后端、调用次数与来源(纯补丁,无节点,删文件夹即移除)| Reports the actual attention backend, call count and source after every sampling run
ComfyUI 顶栏一键切换全局注意力加速后端(Sage/Kitchen/xFormers/PyTorch/SubQuad/Split)+ 顶栏缩放,即时生效、免重启、不改工作流(纯补丁,无节点)| Switch the global attention backend from the top bar instantly, no restart, no workflow changes
A small utility custom nodes.
Krea2 LoRA functional-area matrix routing, post adjustment, isolation testing, and XYZ comparison nodes for ComfyUI.
Minimal ComfyUI custom node that outputs an environment variable value by key.
Comprehensive ComfyUI theme plugin with cyberpunk aesthetics, dynamic visual effects, multilingual support, and extensive customization features for workflows. (Description by CC)
Nodes:SimpleAspectRatio
Wylon image and video nodes for ComfyUI
A fully modular and high-performance implementation of SAM2 (Segment Anything Model 2) for ComfyUI with dual-prompt support, negative point sampling, and upper body segmentation.
Krea 2 image editing for ComfyUI. Prepare grounded prompts and model conditioning for text-to-image, single-image editing, and two-reference scene and identity editing. Update: v0.3.1 is available on GitHub, with per-image grounding and pixels options, a fix for edge streaks at sizes not divisible by 16, and a built-in quick reference. ComfyUI registry approval is pending.
Regional Prompting for ComfyUI - assign different prompts to different image areas.
A powerful ComfyUI custom node for image preview and monitoring
ComfyUI V3 nodes wrapping deepghs/imgutils for anime image analysis and processing.
Sidebar organizer for ComfyUI: folders for workflows and node blueprints with drag & drop, drag-to-canvas insertion, and an auto-backup area.
随机提示词固定人物:通过 LLM 融合固定人物提示词与随机提示词,支持多 API 适配
Twenty continuity, directing, quality, runtime, collaboration, export, and reliability nodes with a bilingual ComfyUI dashboard.
Qwen-Image 2.1 on Intel Arc Battlemage (B390/B580/B570/B770): 4x faster T2I, 158s -> 27s with 3 reference images, runs 2.0MP where stock OOMs.
Auto-filters MiniMax-H3 acceleration LoRAs against the loaded base: trunk fingerprint (final_layer.video_out), build detection (use_adaln_curves), safetensors-header-only file classification, live dropdown filtering, and per-family sampling recipes. NOTE: PDD file selection requires the companion BSAI-ComfyUI-MiniMax-H3-PDD-Acc node (https://github.com/xm6018924/BSAI-ComfyUI-MiniMax-H3-PDD-Acc), which must be installed separately; without it, PDD files cannot be selected.
Multi-hardware orchestration plugin for ComfyUI - registers task leases before sampling/VAE steps, automatically routes VAE decode offload (CUDA / XPU / CPU) based on GPU1 VRAM headroom, and lets other BSAI plugins integrate via a single-file SDK.