<h1 id="comfyui-tj_node-v2.2.0">🚀 ComfyUI TJ_NODE v2.2.0</h1><h1 id="revolutionize-your-workflows-with-wireless-routing-smart-batching-and-bypass-safe-data!">Revolutionize Your Workflows with Wireless Routing, Smart Batching &amp; Bypass-Safe Data!</h1>Are you tired of looking at a canvas that resembles a bowl of spaghetti? Do your complex multi-channel KSampler workflows break or lose filename data when you bypass a node?Meet TJ_NODE — the ultimate custom node pack designed to bring absolute order, efficiency, and safety to your ComfyUI workspace.🔗 <a target="_blank" rel="ugc" href="https://github.com/designloves2/ComfyUI-TJ_NODE/tree/main">Get it on GitHub Here!</a> (Currently pending approval on ComfyUI Manager! Be an early adopter and install manually—instructions below!)<h1><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/12b63f61-5a0b-4f77-9827-b4f0918e3299/original=true/12b63f61-5a0b-4f77-9827-b4f0918e3299.jpeg" /></h1># 🚀 What's New in v2.2.0## ✨ New: Model Set Loader (TJ)Model / Clip / VAE를 각각 개별 드롭다운으로 선택하여 CheckpointLoaderSimple처럼 한 노드에서 MODEL + CLIP + VAE를 한 번에 출력하는 신규 노드입니다.Select Model, Clip, and VAE independently with individual dropdowns and output all three in a single node — just like CheckpointLoaderSimple.- Model: <code>diffusion_models / unet / GGUF</code> 자동 감지 및 로더 선택- Clip: <code>text_encoders / clip</code> + GGUF CLIP 자동 지원- VAE: <code>vae</code> 폴더 목록- <code>model_dtype</code> / <code>clip_dtype</code> 위젯으로 fp8/fp16/bf16 직접 선택 가능- 각 슬롯 <code>[none]</code> 선택 시 해당 출력만 None 반환 (부분 연결 지원)- Auto Set 내장 — MODEL ▶ / CLIP ▶ / VAE ▶ 출력이 Wireless Provider로 자동 등록- 카테고리: <code>✨ TJ_Node/Loaders</code>---## ✨ TJ_Node/Loaders 카테고리 신설<code>TJ_MultiModelSelecter</code>와 <code>TJ_ModelSetLoader</code> 두 노드를 기존 Utility에서 분리하여 전용 Loaders 카테고리로 이동했습니다.---## 🔧 Bug Fixes v2.2.0| 파일 | 수정 내용 || ------------------------- | ------------------------------------------------------------ || <code>ltx2_sampler.py</code> | <code>_sampler_names()</code> 내 <code>nodes</code> NameError 수정 (Critical) || <code>flux2_klein.py</code> | <code>ImageScaleToTotalPixels</code>에 존재하지 않는 <code>resolution_steps</code> 파라미터 제거 || <code>flux2_klein.py</code> | BasicGuider fallback 시 콘솔 경고 출력 추가 (negative conditioning 무시됨 알림) || <code>scene_maker.py</code> | CLIPLoader type 하드코딩<code>stable_diffusion</code>) 제거 → 실제 타입 목록 동적 탐색으로 교체 || <code>save_primary.py</code> | 파일명 카운터 파싱 <code>ValueError</code> 방지 try/except 추가 || <code>z_image_turbo.py</code> | LoRA 이름 목록 중복 항목 제거 || <code>go_stop_tj.py</code> | <code>timeout_sec</code> 파라미터 추가 — 사용자 설정 가능 타임아웃 (0=무제한, 기본값) || <code>multi_model_selecter.py</code> | <code>_connected_output_indices()</code> 최적화 활성화 — 연결된 슬롯만 로드, PROMPT 감지 실패 시 load_all fallback |<h2 id="tj_node-v2.0.1-update!!"># ✨ TJ_NODE v2.0.1 UPDATE!!</h2><h2 id="kill-your-spaghetti-wires-in-comfyui">Kill Your Spaghetti Wires in ComfyUI</h2>## Advanced Wireless Workflow Architecture for ComfyUIYour workflow is not complicated.Your wires are.TJ_NODE is a next-generation ComfyUI node package designed for creators who are tired of spaghetti-wire chaos.Instead of dragging endless wires across giant workflows, TJ_NODE introduces a powerful Wireless Fake Wire Architecture that keeps your graph:* Cleaner* Modular* Scalable* Easier to debug* Actually readable---# 🚀 Why TJ_NODE ExistsAs ComfyUI workflows become larger and more advanced, wire management becomes a nightmare.You know the feeling:* giant wire jungles* impossible reroutes* debugging hell* wires crossing the entire canvas* nodes becoming unreadableTJ_NODE was built specifically to solve this problem.# ⚡ Core Concept## ✨ Invisible Fake Wire SystemTraditional Workflow:```text id="we11d"Load Image ─────────────────────────→ KSampler```TJ_NODE Workflow:```text id="w11ad"Load Image→ Set→ Get→ KSampler```The actual graph connection still exists internally.But visually?The wire disappears.Your workflow instantly becomes dramatically cleaner.# ✨ Main Features## 🧠 Wireless Fake Wire Architecture* Invisible workflow routing* Clean large-scale pipelines* Real internal graph linking* Modular workflow design* Reduced wire clutter---## ⚡ Embedded Get SystemGet functionality can be embedded directly into normal nodes.No need to place endless standalone Get nodes everywhere.Supported Nodes:* Smart Show (TJ)* Save &amp; Preview Image (TJ)* Prompt Text (TJ)* Batch to Multi Image Output (TJ)## 👀 Realtime Hover Wire PreviewInvisible wires can be previewed dynamically on hover.Clean while working.Visible while debugging.```text id="sff22"TJ Node→ Realtime Wires View Mode```## 🌌 Eclipse CompatibilityTJ_NODE fully supports Eclipse SetNodes.Eclipse providers can be received directly from TJ Get systems.Perfect for:* Eclipse workflows* massive routing systems* advanced production pipelines## 🛠 Advanced Workflow UtilitiesIncluded Systems:* Set Node (TJ)* Get Node (TJ)* Multi Get Node (TJ)* Multi Router (TJ)* Dynamic Image Batch (TJ)* Multi Image Loader (TJ)* Smart Show (TJ)* Save &amp; Preview Image (TJ)* Prompt Text (TJ)* Text Concatenate (TJ)* Batch to Multi Image Output (TJ)# 🔥 Built For Large WorkflowsTJ_NODE is not just another utility node pack.It is a workflow architecture toolkit designed for:* Large ComfyUI pipelines* Professional workflow organization* Advanced AI creators* Eclipse users* Massive image systems* Debug-heavy workflows* Clean UI lovers# ✨ What Makes TJ_NODE DifferentMost node packs add more nodes.TJ_NODE changes how workflows are structured.This is not just:&gt; "more utility nodes"This is:&gt; workflow infrastructure---# ⚡ Key Advantages✅ Cleaner workflows&lt;br&gt;✅ Reduced spaghetti wires&lt;br&gt;✅ Faster debugging&lt;br&gt;✅ Better scalability&lt;br&gt;✅ Embedded wireless systems&lt;br&gt;✅ Eclipse compatibility&lt;br&gt;✅ Dynamic provider architecture&lt;br&gt;✅ Ghost wire cleanup&lt;br&gt;✅ Hidden wire rebuilding&lt;br&gt;✅ Professional workflow organization&lt;br&gt;# 🧩 Workflow PhilosophyTJ_NODE treats ComfyUI like a real modular architecture system.The goal is simple:&gt; Build workflows that stay clean even after they become massive.# ✨ Created by TJFeedback, ideas, and bug reports are always welcome.Let’s evolve ComfyUI workflows together.<hr /><h2 id="what's-new-in-v0.22-the-&quot;wireless&quot;-revolution!">🔥 What's New in v0.22? The "Wireless" Revolution!</h2>We've completely overhauled how you connect nodes. Say goodbye to messy cables crisscrossing your screen.<ul><li>🪄 Invisible Wires (Set / Get / Multi Get Nodes): We've introduced a wireless paradigm! Use a <code>Set Node</code> to broadcast your data, and use a <code>Get Node</code> (or <code>Multi Get Node</code>) anywhere on your canvas to receive it via a simple dropdown menu. The connecting wires become invisible, giving you a beautiful, clean canvas.</li><li>🎛️ Multi Router with "Auto Set": Our new Multi Router can handle up to 20 inputs/outputs dynamically. Turn on the <code>Auto Set</code> toggle, and every single output port automatically becomes an independent Set Node broadcasting its data!</li><li>🖱️ Smart Global Context Menu: Right-click anywhere for pure magic!<ul><li>Instantly <code>Show/Hide Wires</code> for specific connections.</li><li>Click <code>Go to Source</code> on any Get node to instantly jump across the canvas to where the data is coming from.</li><li>Use the Convert ALL Inputs/Outputs to Set/Get button to instantly untangle any giant node with one click.</li></ul></li><li>🎨 Fixed UI &amp; Top-Tier Sorting: TJ Nodes now exclusively feature a sleek Purple &amp; Black brand color and are forcibly sorted to the very top of your Add Node menu (<code>0. 🟣 TJ Nodes</code>) so you never have to hunt for them again.</li></ul><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/72248c91-527f-4bd2-9450-40361b7d36b8/original=true/72248c91-527f-4bd2-9450-40361b7d36b8.jpeg" /><a target="_blank" rel="ugc" href="https://github.com/user-attachments/assets/e084d5c4-3edc-44fb-b78d-e1d57fdef77e">https://github.com/user-attachments/assets/e084d5c4-3edc-44fb-b78d-e1d57fdef77e</a>)<hr /><h2 id="core-features-that-will-save-your-life">🌟 Core Features That Will Save Your Life</h2><h3 id="bypass-safe-architecture">🛡️ Bypass-Safe Architecture</h3>Ever bypassed an intermediate node only to have your whole pipeline output errors or break? TJ_Nodes are built to be bypass-safe. Even if a node passes a <code>None</code> value, the workflow won't stop. It intelligently collects only valid data. If everything is bypassed, it safely outputs a black screen instead of an error.<h3 id="dynamic-expanding-inputs">➕ Dynamic Expanding Inputs</h3>No more fixed input slots. When using our Dynamic Batch nodes, connecting a wire automatically spawns the next input pin (<code>image_2</code>, <code>files_2</code>, etc.). Disconnect a wire, and the UI cleans itself up automatically. It even supports wildcard (<code>*</code>) matching to prevent connection errors!<h3 id="advanced-save-and-path-tracking-(eclipse-integrated)">📂 Advanced Save &amp; Path Tracking (Eclipse Integrated)</h3>If you use <code>ComfyUI_Eclipse</code>, you know it generates complex regex-based filenames and date folders. TJ_Nodes perfectly trace these back in memory!<ul><li>Custom Paths: Leave the path blank to save in the original folder, or use relative paths like <code>..\</code> or <code>\upscale</code> to dynamically create sub/parent folders on the fly.</li><li>100% Quality Output: Choose to keep original extensions or convert to PNG/JPG/WebP at guaranteed maximum lossless quality.</li></ul><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/cb474709-c169-4031-896c-569d2de9e628/original=true/cb474709-c169-4031-896c-569d2de9e628.jpeg" /><a target="_blank" rel="ugc" href="https://github.com/user-attachments/assets/ce20c91c-9129-402c-b910-4939250c6da2">https://github.com/user-attachments/assets/ce20c91c-9129-402c-b910-4939250c6da2</a>)<hr /><h2 id="the-ultimate-utility-nodes">🛠️ The Ultimate Utility Nodes</h2><h3 id="multi-image-loader-(tj)">🖼️ Multi Image Loader (TJ)</h3>An all-in-one image manager directly inside your node.<ul><li>URL Download: Paste a link and download directly to your ComfyUI input folder!</li><li>Built-in File Browser: Switch between Input/Output/Download folders, multi-select, and Drag &amp; Drop files from your PC directly into the node.</li><li>Auto-Resizing: Load images of different sizes? No problem. It automatically resizes them to match the first image (or by Megapixel budget) and outputs a perfect batch tensor.</li></ul>&lt;div style="display:flex; gap:10px;"&gt; &lt;img width="48%" src="https://github.com/user-attachments/assets/51311568-aa6d-471e-82e2-de01b8939aaf" /&gt; &lt;img width="48%" src="https://github.com/user-attachments/assets/100c923b-5d12-4407-b3f1-b1f2faf8f7fe" /&gt; &lt;/div&gt;<h3 id="batch-to-multi-image-output-(tj)">✂️ Batch to Multi Image Output (TJ)</h3>Need to process images individually after a batch load? This node splits a batch tensor into up to 64 individual IMAGE slots. Perfect for routing different images to different KSamplers simultaneously!<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/18692788-d8c0-4852-adf9-44cdb1a8cb7e/original=true/18692788-d8c0-4852-adf9-44cdb1a8cb7e.jpeg" /><a target="_blank" rel="ugc" href="https://github.com/user-attachments/assets/1aa30408-18b0-4943-bbd0-99bc2d2c233b">https://github.com/user-attachments/assets/1aa30408-18b0-4943-bbd0-99bc2d2c233b</a><hr /><h2 id="how-to-install-(early-access!)">⏳ How to Install (Early Access!)</h2>We are currently sitting in the queue waiting for approval on the official ComfyUI Manager. 😭 But you don't have to wait! Getting it manually is super easy:Method 1: Using Git (Recommended)<ol><li>Open your terminal/command prompt.</li><li>Navigate to your ComfyUI custom nodes folder: <code>cd ComfyUI/custom_nodes/</code></li><li>Clone the repo: <code>git clone https://github.com/YOUR_GITHUB_USERNAME/comfyui-TJ_Node.git</code> (Make sure to replace with your actual repo link!)</li><li>Restart ComfyUI.</li></ol>Method 2: Manual ZIP Download<ol><li>Go to the <a target="_blank" rel="ugc" href="https://www.google.com/url?sa=E&amp;q=https%3A%2F%2Fgithub.com%2FTJ-Kim%2Fcomfyui-TJ_Node">GitHub Repository</a>.</li><li>Click the green <code>&lt;&gt; Code</code> button and select Download ZIP.</li><li>Extract the ZIP file.</li><li>Rename the extracted folder to <code>comfyui-TJ_Node</code> and place it inside your <code>ComfyUI/custom_nodes/</code> directory.</li><li>Restart ComfyUI and look for <code>✨ TJ Nodes</code> in your menu!</li></ol>

Kill the Spaghetti! 🍝 TJ_NODE v2.0 (Update!!!): Wireless Routing & Smart Batching for ComfyUI

🚀 What's New in v2.2.0✨ New: Model Set Loader (TJ)Model / Clip / VAE를 각각 개별 드롭다운으로 선택하여 CheckpointLoaderSimple처럼 한 노드에서 MODEL + CLIP + VAE를 한 번에 출력하는 신규 노드입니다.Select Model, Clip, and VAE independently with individual dropdowns and output all three in a single node — just like CheckpointLoaderSimple.<ul><li>Model: <code>diffusion_models / unet / GGUF</code> 자동 감지 및 로더 선택</li><li>Clip: <code>text_encoders / clip</code> + GGUF CLIP 자동 지원</li><li>VAE: <code>vae</code> 폴더 목록</li><li><code>model_dtype</code> / <code>clip_dtype</code> 위젯으로 fp8/fp16/bf16 직접 선택 가능</li><li>각 슬롯 <code>[none]</code> 선택 시 해당 출력만 None 반환 (부분 연결 지원)</li><li>Auto Set 내장 — MODEL ▶ / CLIP ▶ / VAE ▶ 출력이 Wireless Provider로 자동 등록</li><li>카테고리: <code>✨ TJ_Node/Loaders</code></li></ul>✨ TJ_Node/Loaders 카테고리 신설<code>TJ_MultiModelSelecter</code>와 <code>TJ_ModelSetLoader</code> 두 노드를 기존 Utility에서 분리하여 전용 Loaders 카테고리로 이동했습니다.🔧 Bug Fixes v2.2.0파일수정 내용<code>ltx2_sampler.py_sampler_names()</code> 내 <code>nodes</code> NameError 수정 (Critical)<code>flux2_klein.pyImageScaleToTotalPixels</code>에 존재하지 않는 <code>resolution_steps</code> 파라미터 제거<code>flux2_klein.py</code>BasicGuider fallback 시 콘솔 경고 출력 추가 (negative conditioning 무시됨 알림)<code>scene_maker.py</code>CLIPLoader type 하드코딩(<code>stable_diffusion</code>) 제거 → 실제 타입 목록 동적 탐색으로 교체<code>save_primary.py</code>파일명 카운터 파싱 <code>ValueError</code> 방지 try/except 추가<code>z_image_turbo.py</code>LoRA 이름 목록 중복 항목 제거<code>go_stop_tj.pytimeout_sec</code> 파라미터 추가 — 사용자 설정 가능 타임아웃 (0=무제한, 기본값)<code>multi_model_selecter.py_connected_output_indices()</code> 최적화 활성화 — 연결된 슬롯만 로드, PROMPT 감지 실패 시 load_all fallback

Files with SHA256: 0eefd2f1439fca2ab351ab5dcf76da7447f21984ac54ae0f6e413ead816e0bd5