ComfyUI Custom Extensions
Browse 9,446 ComfyUI extensions and 73,215 custom nodes.
Stable-Layers / Qwen-Image-Layered RGBA layer decomposition nodes for ComfyUI, plus a PEFT-format LoRA loader that actually applies.
This repository contains a ComfyUI node for blending images using various blending modes. Can be used to watermark images, create overlays, or apply effects to images in a ComfyUI workflow.
This custom node for ComfyUI allows users to input multiline text and select a specific line by its number. The node processes the input and returns the selected line along with its index.
OllamaKiller is a cross-platform utility node for ComfyUI that automatically manages Ollama processes (`ollama.exe` on Windows, `ollama` on macOS/Linux). It helps free up VRAM by terminating Ollama processes after model execution, improving workflow performance and memory management. Features include graceful process termination, detailed status reporting, and seamless text passthrough.
Real-time node execution timing panel for ComfyUI. Displays per-node duration, progress bars, heat-map coloring, and total execution time in the sidebar.
ComfyUI nodes for LLM Structured Outputs with integration for prompting
A ComfyUI custom node and workflow for architectural visualization with Krea 2 Turbo, featuring configurable presets for lighting, atmosphere, architectural style, composition, camera angles, and rendering detail. Designed to help create consistent, high-quality architectural images.
ComfyUI custom nodes for line art centerline extraction and SVG manipulation with advanced editing and visualization capabilities.
This repository provides a custom node for ComfyUI that allows managing positive and negative prompts in a structured JSON format. The node supports adding, updating, and logging prompts, ensuring seamless integration into your workflow.
This is a custom node to convert image to ascii art string.
ComfyUI nodes for generating and editing images via Nano Banana 2 through Google Vertex AI. (Description by CC)
Simulates authentic film quartz date stamps with physical optical burn-in, multi-layer glow, and halation in ComfyUI.
Advanced Block LoRA merging node for ComfyUI (allows selective LoRA block merging)
Advanced LoRA merging node for ComfyUI (additive, average, sequential)
Create Bounding Boxes editor (same behaviour as ComfyUI's built-in node) that works in the classic LiteGraph renderer.
This is a custom node for ComfyUI that allows you to use the UCloud API directly in ComfyUI. UCloud is a high-performance AI computing platform providing powerful image generation models and industry-leading inference speeds. It offers a wide range of AI model options, including popular text-to-image and image editing models, ready to use without local GPU setup.
Three sigma schedules for ComfyUI: MOSS Shift, MOSS Curve, and MOSS Sigma, available in KSampler and a dedicated SIGMAS node. Developed for MOSS-Image workflows.
Translate prompts with PLaMo 3 Translate 0.8B in ComfyUI or through LM Studio, with Live translation and 53 language choices.
FLUXNATION is a fused FP8 CUDA kernel for FLUX.1 that replaces the entire SingleStreamBlock forward pass with a single torch.ops call covering modulation dual GEMMs QKV projection RoPE attention gating and residual all in one shot. On top of that it ships with neuromorphic spike attention which scores every attention block by real dot product similarity and kills the ones that dont matter after image structure has formed keeping only the top 45 percent. Step caching then replays the cached output on alternating spike steps for zero attention compute every other step. The result is 30 percent faster generation than stock ComfyUI on an RTX 4090 with no quality loss. Works with every sampler ComfyUI supports at any step count. FP16 ports for 10 series 20 series and 30 series GPUs are included. A must have for anyone running FLUX.1.
FaceDetailer variant for ComfyUI Impact Pack with a separate denoise value per cycle pass and optional color matching.
ComfyUI custom node pack for SDNQ quantized models with a modular split-node architecture. Forked from EnragedAntelope/comfyui-sdnq and redesigned into dedicated nodes. Provides 6 dedicated nodes: SDNQ Model Loader, SDNQ LoRA Loader, SDNQ VAE Encode, SDNQ Sampler V2 (general models), Flux2 SDNQ Sampler V2 (Flux2-optimized), and Flux2 SDNQ TorchCompile (performance optimization). Enables 50-75% VRAM savings while maintaining quality. Note: Currently focused on FLUX.2 development; other models not tested.
Custom nodes for ComfyUI: interactive bounding boxes, a multi-image gallery, hierarchical workflow controls, switches, and loops.
Connect cloud image and video generation APIs to composable ComfyUI workflows.
Velvet Vice custom nodes shared by the LTX 2.3 and LTX 2.5 ComfyUI workflows.
Velvet Vice MiniMax H3 image-to-video custom nodes with Native/GGUF routing, Quality Refine second sampling, persistent prompting, live preview and finishing tools.
Pose-aware decal placement, projection, masking, displacement, and compositing tools for ComfyUI.
Cinematic AI prompt node for LTX video. Three-pass pipeline: Choreography pre-pass, main generation, lint + auto-repair. Supports Windows, Linux, macOS via llama.cpp (Gemma 4 26B).
Implementation of architectural related graph algorithm in ComfyUI.
The ComfyUI-Visionatrix nodes are designed for convenient ComfyUI to [a/Visionatrix](https://github.com/Visionatrix/Visionatrix) workflow support migration, in particular to extract prompt input params (input, textarea, checkbox, select, range, file) to be used in simplified Visionatrix UI.
Nodes:Inspect, Any String, Model From URL
An experimental STAR Video Upscale Wrapper.
A ComfyUI custom node that converts prompts between different AI image generation models.
A collection of utility nodes designed to bring dynamic logic and flexible selection to your ComfyUI workflows.
TTM (Time-to-Move) node for ComfyUI enabling motion-controlled video generation with Wan2.2 models using dual-clock denoising for independent background and object animation control.
Replicates Forge Neo (neo branch) sampling, RNG and prompt encoding inside ComfyUI - same parameters, same results.
Free MiniMax H3 (Hailuo) text-to-video and image-to-video through the AMD gateway. Register for an API key; no local GPU required.
Simple ComfyUI Game Boy [Advance] Player based on gbaONLiNE and IodineGBA
Custom node to handle text. Save and load text files. Remove comments from text.
Load a .txt file of prompts and cycle through them automatically in ComfyUI — one image per prompt, hands-free.
ComfyUI node for browsing Pixiv directly within ComfyUI, supporting recommendations, rankings, bookmarks, artist browsing, and batch image export. (Description by CC)
A ComfyUI custom node for enhancing prompts using OpenAI-compatible APIs (like DeepSeek, OpenAI, etc.).
comfyui_imagetostl is a simple node for ComfyUI that converts grayscale images to STL.
BSAI Asset Library Auto List - auto-extract @图N asset descriptions from BSAI storyboard scripts into an asset library for film production
BSAI ComfyUI Nodes - Video/Image/Audio processing, LLM inference, AI generation
Smart VAE decode/encode routing node - automatically selects the decode device by GPU1 VRAM headroom: CUDA (GPU1) / XPU (8190 iGPU worker) / CPU (fallback), offloading VAE decode from the discrete GPU to the Intel iGPU to free VRAM for sampling.
Chain-load multiple LoRAs in a single node for Qwen-Image-2.1 (and any diffusion model) workflows.
A custom node that enables MonarchAttention for self-attention in ComfyUI model branches by setting model-level override, affecting only the connected model branch rather than patching globally.
A ComfyUI custom node for safely blending large IMAGE tensors in chunks with explicit resize and compute-device controls.
A ComfyUI custom node pack implementing Scale-Locked Residual Diffusion for high-resolution composition and anatomy stability.
A collection of artistic glitch and image manipulation nodes for ComfyUI, featuring advanced noise effects, color manipulations, distortions, and more.
Generate 24/30/40 reference images of one person with varied expressions, head angles, framings, outfits and backgrounds - for character sheets, pose references and LoRA datasets (PNG + TXT captions). Qwen Image Edit 2511 / Qwen Image 2.1.
Load COF (Character Open File) characters in ComfyUI: primary image, reference views, prompt, LoRA, pre-encoded references (RefMods), voice, persona text, memories, psyche, poses, motion clips and 3D files — with the character's rights and age rules enforced.
ComfyUI node for background removal, implementing InSPyReNet
ComfyUI custom node - universal model quantizer. Converts diffusion models (safetensors) to NVFP4 / FP8 (e4m3fn) / GGUF in place, with a per-architecture blacklist that protects structure-sensitive layers so the result loads and renders correctly. Auto (Universal) mode works on any architecture.
ComfyUI custom node - bilingual (English/Chinese) all-in-one Ultimate Sampler V2.3 EN. Whole checkpoint or split stack (diffusion + dual CLIP + VAE), Flux/FLUX3 reference editing/MOSS-Image/SD3/Wan/Qwen-Image/Krea2/Z-Image/LTXV support, LoRA list, fp8/nvfp4 weight precision. Same node also available as the Chinese-only edition.
Local metadata backend for ComfyUI Prompt Workbench: offline safetensors trigger scanning, editable character/outfit groups and output organization. Optional explicit online scans query Civitai using SHA-256 hashes. No models or personal database bundled; no standalone panel.
Sidebar for prompt composition, field templates, LoRA character/outfit selections, saved batch presets and pause/resume submission. Chinese/English UI and local backups. Install ComfyUI LoRA Trigger Helper for metadata scanning/editing and character output organization.
Composite a real cut-out product PNG into an AI-generated scene: geometry-adaptive placement, grounding shadows, product-led lighting, and a GPT-scene swap backend — the product pixels stay 100% real.
ComfyUI nodes for PhotoDoodle model.
ComfyUI-SoundHub is a collection of audio processing nodes designed for ComfyUI, enabling seamless audio processing and generation within your ComfyUI workflows.
Nodes:Upscale Video Tensorrt
A custom node for ComfyUI that performs color correction on clothing in face-swapped images. This node helps maintain the original clothing color when using face swap tools, addressing common color shifts that occur during the face swap process.
A ComfyUI Custom Node to backup workflows and their required models locally.
EmbeddingsNameLoader, EmbendingList
A set of nodes for rapidfiring the half backed latents, cleaning up obvious bad generations and automatically queueing the rest to fully generate.
ComfyUI node for creating some Turtle Graphic demos.
A ComfyUI node extending Flux/Krea2 model image generation beyond native resolution using YaRN and NTK-aware frequency scaling to prevent repeating-pattern artifacts. (Description by CC)
ComfyUI custom nodes for integrating multiple external APIs including Gemini, OpenAI, Replicate, and ElevenLabs directly into local workflows. (Description by CC)
ComfyUI custom nodes for converting prompt text between common anime image-generation formats.
Nodes:ZC DrawShape Node
Universal ComfyUI model loader for checkpoints, diffusers, diffusion models, UNet, GGUF, dual CLIP outputs and dual VAE outputs with dynamic UI, Registry Node Pack Info previews, selected-model metadata cleanup, and auto CLIP/VAE detection for latest model families including Qwen Image 2.1/QWN2 and MiniMax Music3/MM3.
ComfyUI nodes for video normal-map estimation with NormalCrafter and motion-aware temporal stabilization.
Provide a simple interface to simplify prompt writing.
Single-node image quality gate for LoRA dataset filtering. Analyzes blur, sharpness, noise, clipping, and entropy with three-tier verdict (PASS/SO-SO/FAIL), center-weighted blur detection, and shot-type presets.
A ComfyUI custom node that patches the attention mechanism inside a VAE's encoder and decoder to use a specific implementation, regardless of what ComfyUI auto-selected at startup.
A LLM API key management solution for ComfyUI with automatic load balancing support.
You'll get a new node Latent Size Selector, you can pick the x and y sizes from a list.
DeepSeek-powered Illustrious prompt generator nodes for ComfyUI.
This project allows you to adjust SDXL's two text encoder's strengths individually for clip_g (ViT-bigG) and clip_l (CLIP-ViT-L) within ComfyUI. (And other adjustments)
ComfyUI custom nodes in a single package: Arena legacy + AutoCache + Updater
NODES: Image Loader, Get Num Frames, Get Resolution Node, Video Loader, Film Grain
A simple ComfyUI nodes that integrates [a/PiAPI faceswap](https://piapi.ai/faceswap-api) service into ComfyUI. This can be helpful if you're trying to create a workflow that includes faceswap for commercial usage.
ComfyUI custom node suite integrating multiple AI image generation APIs (Gemini, Flux.1 Kontext Pro) with unified authentication via ComfyUI API Key and real-time cost tracking. (Description by CC)
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。
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。
ComfyUI custom nodes for Hunyuan3D - Complete 3D generation suite
Pause ComfyUI workflows to inspect and manually select images from a batch.
Expand short prompts into MiniMax-H3 structured prompts with a local GGUF Qwen VLM and the official h3-prompt-writing rules.
ComfyUI nodes for video -> audio -> JSON transcript pipeline. Designed to pair with ComfyUI-MiMoASR for the ASR step.
ComfyUI nodes for integrating Ollama local LLM models with image or text generation workflows.
This extension can send HTTP Requests. You can request image generation to StableDiffusion3 and post images to X (Twitter) and Discord.
RAFT-based optical flow nodes for ComfyUI with robust batch/index handling, flow save/load, and inpaint crop/stitch support.
Nodes:Blender viewport color, Blender Viewport depth
基于 Qwen3 的本地翻译节点
A native dependency-free port of Microsoft VibeVoice long-form multi-speaker TTS for ComfyUI.
Warp Unfake.py for ComfyUI
A ComfyUI custom node for generating dynamic scrolling lyrics videos with smooth animations. Supports audio-to-text transcription using OpenAI Whisper or importing existing SRT subtitles.