Extensions/ComfyUI-JNK-Tiny-Nodes
ComfyUI Extension

ComfyUI-JNK-Tiny-Nodes

A collection of useful custom nodes for ComfyUI - image processing, text manipulation, and workflow automation.

By Aljnk·Created about a year ago·Updated 12 months ago· 3
Aljnk/ComfyUI-JNK-Tiny-Nodes
Nodes33
On cloudLocal install
Category🔧 JNK, image
Stars3
Updated12 months ago

Nodes (33)

Add Layer Overlay JNK

Stack layers with position, rotation and fade — watch the opacity slider

🔧 JNK
Ask Google Gemini JNK

Free text-and-image API calls, throttle included

🔧 JNK
Bridge All JNK

The boring pass-through node that tidies up your graph

🔧 JNK
Create Folder JNK

Make sure the output directory exists before something tries to write

🔧 JNK
Create RGBA Image JNK

A transparent canvas when your workflow needs one

🔧 JNK
Get All Alpha Layers JNK

Pull every significant cutout out of a mask

🔧 JNK
Get Gemini Keys JNK

Keep your API keys out of the workflow file

🔧 JNK
Get Gemini Models JNK

Set your model IDs and RPM budgets in one place

🔧 JNK
Get Models JNK

Cycle through checkpoints and UNets for testing, then pin one for work

🔧 JNK
Get One Alpha Layer JNK

The single biggest cutout, noise and all the rest dropped

🔧 JNK
Get Substring JNK

Slice long names down to filename-friendly size

🔧 JNK
Get Text From List by Index JNK

Pull one item out of a list of strings

🔧 JNK
Get Timestamp JNK

A fresh Unix timestamp every run for unique filenames

🔧 JNK
Image Filter Loader JNK

Batch-load every matching image in a folder, in order

🔧 JNK
Join Strings JNK

Glue up to five strings together with a delimiter

🔧 JNK
Load Checkpoint Model with Name JNK

The 'with Name' node that actually loads

🔧 JNK
Load Image if Exist JNK

Load from a path and get a boolean you can branch on

🔧 JNK
Load LoRA with Name JNK

It doesn't load a LoRA, and that's the point

🔧 JNK
Load UNet Model with Name JNK

A name node in a loader's costume

🔧 JNK
Prepare Image for AI JNK

Shrink images before an LLM eats your token budget

🔧 JNK
Queue Stop JNK

A kill switch for your queue that can also play you a little tune

🔧 JNK
Save Frame JNK

Grab a single frame out of a video batch without the ceremony

🔧 JNK
Save Static Image JNK

Save images to any path you want, metadata included

🔧 JNK
Save Video Images JNK

Spill a video batch onto disk as numbered PNGs

image
Split String JNK

Split text into a real list, with skipping and limits built in

🔧 JNK
Stroke RGBA Image JNK

Outline anything with a stroke ring, alpha intact

🔧 JNK
Switch Index JNK

A little ON/OFF toggle that outputs 2 or 1

🔧 JNK
Switch Integer JNK

Turn an ON/OFF toggle into a 1 or a 0

🔧 JNK
Text Saver JNK

Write generated text to a file without stomping your last good copy

🔧 JNK
Text to Key JNK

Turn a long model filename into a short, stable key

🔧 JNK
Text to MD5 JNK

A stable, tiny ID for any text

🔧 JNK
ToonOut Remove BG JNK

A BiRefNet fine-tune that cuts out cartoon art

🔧 JNK
Topaz Photo Upscaler (Autopilot) JNK

Hand your image to Topaz Photo AI, then take the result back

🔧 JNK
Readme

ComfyUI-JNK-Tiny-Nodes

A collection of useful custom nodes for ComfyUI - image processing, text manipulation, and workflow automation.

📦 Installation

You can access it directly from the ComfyUI Manager. Search for "JNK," click "Install," and then restart.

OR

Method 1: Git Clone (Recommended)

cd ComfyUI/custom_nodes/
git clone https://github.com/Aljnk/ComfyUI-JNK-Tiny-Nodes.git

Method 2: Manual Download

  1. Download the repository as ZIP
  2. Extract to ComfyUI/custom_nodes/ComfyUI-JNK-Tiny-Nodes/

🔧 Dependencies

# Required Python packages
pip install opencv-python
pip install pygame
pip install google-genai

🔄 Changelog

v1.1.4 (September 10, 2025)

  • Added "ToonOut Remove BG" node - advanced background removal using fine-tuned BiRefNet model.
  • Model file: birefnet_finetuned_toonout.pth
  • Download: HuggingFace - joelseytre/toonout
  • Installation: Place model file in ComfyUI/models/RMBG/ToonOut/ folder
  • Example workflow included in example_workflows/01_ToonOut.json

v1.1.3 (July 14, 2025)

  • Added "Split String" node for advanced text splitting. Fixed minor bugs.

v1.1.2 (July 12, 2025)

  • Added 2 utility nodes to enhance "Ask Google Gemini" workflow.
<details> <summary>Show older updates</summary>

v1.1.1 (July 10, 2025)

  • Added "Prepare Image for AI" node for optimizing images before sending to AI models (token saving).

v1.1.0 (July 8, 2025)

  • Added "Ask Google Gemini" node for text and image AI processing.

v1.0.0 (June 4, 2025)

  • Initial release.
  • 25+ custom nodes.
</details>

📋 Nodes

Models

  • Load UNet Model with Name - Return UNet model and model name. The name can be used when saving files to understand which model was used in creation.
  • Load Checkpoint Model with Name - Return components of a checkpoint model and its name. The name can be used when saving files to understand which model was used in creation.
  • Load LoRA with Name - Return LoRA model and LoRA name. The name can be used when saving files to understand which LoRA was used in creation.
<div align="center"><img src="images/models.jpg" alt="Model Nodes" width="830" /></div>

Video

  • Save Frame - Save individual frames with sequential numbering. Useful for extracting specific frames from video for processing.
  • Save Video Images - Batch save multiple images as numbered sequence. Allows decomposing video into images, modifying some frames, then reassembling into video.
<div align="center"><img src="images/videos.jpg" alt="Video Nodes" width="450" /></div>

Image

  • Save Static Image - Save images with quality control, metadata, and an option to skip existing files. Allows saving images using filesystem paths.
  • Load Image if Exist - Load images with existence check, returning a boolean status. Loads image by filesystem path and returns boolean to check if image was loaded.
  • Image Filter Loader - Batch load images from folders with filtering + returns number of found images and their paths. Loads images from a folder whose names contain the filter keyword.
  • Prepare Image for AI - Optimize images (reduce image size) to save tokens when using with "Ask Google Gemini" node for computer vision tasks.
  • Stroke RGBA Image - Add customizable stroke effects to RGBA images or by mask. Preserves alpha channel after stroking.
  • Create RGBA Image - Generate transparent or colored RGBA image. Useful for compositing images from different layers.
  • Add Layer Overlay - Composite layers with positioning, rotation and opacity control.
  • Get One Alpha Layer - Extract the largest alpha region from images.
  • Get All Alpha Layers - Extract all significant alpha regions while preserving smaller details.
  • ToonOut Remove BG - Advanced background removal using fine-tuned BiRefNet model.
<div align="center"><img src="images/images.jpg" alt="Image Nodes" width="830" /></div>

Upscale

  • Topaz Photo Upscaler - Integrate with Topaz Photo AI for professional upscaling with autopilot settings. Compression parameter controls PNG compression after upscaling. tpai_exe specifies path to folder containing tpai.exe file. For custom settings, configure AutoPilot mode in Topaz Photo AI preferences.
<div align="center"><img src="images/upscalers.jpg" alt="Upscale Nodes" width="370" /></div>

Text

  • Get Text From List by Index - Extract specific text items from lists or string lists by index position.
  • Text Saver - Save text to file by path. Useful for saving translations or configuration sets.
  • Get Substring - Extract substrings with start position, length and reverse options. Useful for trimming long model names when saving results.
  • Text to Key - Convert text to key. Cuts beginning and end of text to create keys from these parts.
  • Text to MD5 - Generate MD5 hashes from text for unique identifiers.
  • Join Strings - Concatenate multiple strings with custom delimiters. Text can be written directly in the node.
  • Split String - Split text by delimiter with start position, skip patterns, and count limits.
  • Get Timestamp - Generate current timestamp strings for file naming.
<div align="center"><img src="images/texts.jpg" alt="Text Nodes" width="830" /></div>

Logic

  • Switch Integer - Convert ON/OFF states to 1/0 integers for conditional logic.
  • Switch Index - Convert ON/OFF states to 2/1 indices for node switching.
  • Get Models - Get model list by search key. In test mode, uses next model from list each run (very convenient for model testing). In work mode, uses only model from model_name or first in list.
<div align="center"><img src="images/logic.jpg" alt="Logic Nodes" width="440" /></div>

System

  • Bridge All - Simple bridge with main parameters.
  • Queue Stop - Stop processing with optional sound alerts and queue clearing.
  • Create Folder - Create directory by path programmatically.
<div align="center"><img src="images/system.jpg" alt="System Nodes" width="750" /></div>

API

  • Ask Google Gemini - Send a text prompt (with an optional image) to the Google Gemini AI model. When the RPM is greater than 0, the node will pause between requests to maximize request throughput while staying within rate limits. Get a free API key. Available models. Free tier RPM limits.
  • Get Gemini Keys - Load up to 5 Google Gemini API keys from a text file. Each key should be on a separate line. Useful for rotating between multiple API keys to manage rate limits.
  • Get Gemini Models - Configure and manage different Gemini models (Pro, Flash, Light) with their respective RPM (requests per minute) settings.
<div align="center"><img src="images/api.jpg" alt="API Nodes" width="370" /></div>

⚙️ Special Requirements

ToonOut Remove BG

Requirements:

Example Workflow:

<div align="center"><img src="example_workflows/01_WF_ToonOut.png" alt="ToonOut Remove BG Workflow" width="830" /></div>

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

📞 Support

☕ Support the Author

If you find these nodes helpful, consider buying me a coffee: