ComfyUI Extension: ComfyUI-nhknodes
ComfyUI custom nodes collection with workflow utilities, image processing, and AI integration
Custom Nodes (0)
README
NHK Nodes for ComfyUI
A comprehensive collection of utility nodes for ComfyUI workflows. Organized into logical categories for better workflow management.
π Featured Node
πΌοΈ Image Loader With Previews
The standout feature of this collection - an advanced image loader that hopefully simplifies how you browse and select images in ComfyUI.
<p align="center"> <img src="./images/loader_1.png" width="45%" /> <img src="./images/loader_2.png" width="45%" /> </p>- Browse from any folder on your system
- Live image previews - see images before selecting
- Multiple sorting options: name, date modified, created
π¦ Installation
Via ComfyUI Manager (Recommended)
- Open ComfyUI Manager
- Search for "NHK Nodes"
- Click Install
- Restart ComfyUI
Manual Installation
- Clone or download this repository to
ComfyUI/custom_nodes/nhknodes - Restart ComfyUI
- Nodes will appear in organized categories under
nhk
ποΈ Node catalog
π€ Text Processing (nhk/text)
- π Simple Text Input β Minimal text entry node that just forwards its value.
- π Text Display β Shows any incoming string inside the UI while keeping the data flowing.
- π§© Text Combiner β Unlimited text inputs with automatic sockets and configurable separator.
- π§· Text Template β Lightweight templating (
The [text_1] walks in the [text_2]) with numbered placeholders.
πΌοΈ Image Processing (nhk/image)
- πΌοΈ Image Loader With Previews β The featured browser with searchable folders, previews, and sorting.
- πΈ Load Image Series β Auto-incrementing sequence loader with reset/progress tracking.
- π¦ Image Grid Batch β Stacks arbitrary images into a batch tensor for downstream samplers.
- π― Image Grid Composite β Creates presentation grids with gutters, padding, and background control.
- π Visual Resizer β Drops any image onto a custom canvas size with precise offsets.
- π§ Add Headroom β Shrinks the subject within the original canvas to create breathing room up top.
- π¨ Edit with Krita β Sends a frame to Krita, waits for edits, and re-imports it into the workflow.
βοΈ Workflow Utilities (nhk/utility)
- π Cycling Switch β Rotates through unlimited inputs, staying on each for a configurable number of runs.
- πͺ Interval Gate β Turns a branch on/off every N executions (perfect for βevery 5th imageβ flows).
- π Double Switch (In/Out) β Paired A/B switches that route image/text tuples together.
- β±οΈ Execution Counter β Counts queue runs, stops when a limit is reached, and shows progress.
- π Size Picker β Flux/SDXL/Qwen-optimized resolution presets with handy metadata.
- π Play Sound β Small notification node that plays an audio file when a queue finishes.
π€ AI & Media (nhk/ai)
- π¦ Ollama API β Local chat/vision models with hidden thinking output and optional image prompts.
- π€ OpenAI API β GPTβ4/GPTβ5 chat with optional vision input (requires
OPENAI_API_KEY). - π¬ Sora Video Generation β REST client for OpenAI Sora 2 / Sora 2 Pro text-to-video jobs (API key required).
ποΈ Video Output (nhk/video)
- πΎ Save Video (Short) β Variant of Comfyβs save node that uses a 3βdigit counter (
000β999) for tidier filenames.
π Key Features
- Dynamic Inputs - Many nodes support unlimited inputs that auto-expand as you connect
- Conditional Branching - Interval Gate enables periodic workflow paths (upscale every 5th image, etc.)
- Configurable Cycling - Cycling Switch with stay duration control for precise input testing
- Smart UI - Hover tooltips and emoji icons for easy identification
- Clean Organization - Logical categories make nodes easy to find
- Professional Quality - Consistent design and error handling throughout
π Requirements
For OpenAI API & Sora nodes
- OpenAI API key in
.envfile:OPENAI_API_KEY=your_api_key_here - GPT nodes support gpt-4o, gpt-4o-mini, chatgpt-4o-latest, plus gpt-5 / gpt-5-mini / gpt-5-nano.
- Sora node currently targets
sora-2andsora-2-prowith optional image reference input.
For Ollama API Node
- Ollama server running locally (
ollama serve) - Pull at least one supported model (
ollama pull qwen3-vl:8b, etc.) - Optional vision input works automatically when the selected model supports it
For Edit with Krita Node
- Krita available either as AppImage, system install, or Flatpak
- Writable directory for round-tripping frames (default
/home/nhk/workspace/editing)
π License
MIT License - Feel free to use, modify, and distribute.
π€ Contributing
Contributions welcome! Please follow the existing code style and add appropriate documentation.
Created by NHK | GitHub | ComfyUI Registry