# ComfyUI-oshtz-nodes Custom nodes for ComfyUI created for some of my workflows. ### LLM All-in-One Node Easy GPT/Claude/OpenRouter integration in ComfyUI: - OpenAI & Anthropic models - Live OpenRouter model list fetched directly from the registry - Image-to-text capabilities
### GPT Image 1 Node Generate images via OpenAI's vision model: - Direct OpenAI API integration - Image editing capabilities with mask support - Quality and size customization - Transparent background option ### Easy Aspect Ratio Node Simplify your workflow with preset aspect ratios: - Large selection of common ratios (1:1, 16:9, 21:9, etc.) - Returns width and height values ### String Splitter Node Split text into multiple outputs: - Up to 10 separate outputs - Customizable separator ### LoRA Switcher (Dynamic) Dynamic LoRA selection powered by a single configurable node: - Drive LoRA choice via API-friendly widget data - Unlimited rows with custom strengths managed from the UI - Keeps workflows simpler than juggling fixed 10/20/40 variants ### Image Overlay Node (Beta 🚧) Combine images with precision: **Current Features** - Basic image compositing - Position control (X/Y) - Transparency adjustment - Auto size matching - Alpha channel support **Coming Soon** - Advanced positioning - Transform controls - Multiple blend modes - Layer management - Interactive preview ## Installation First navigate to your ComfyUI installation custom nodes directory. 1. Clone this repository: ```bash git clone https://github.com/oshtz/ComfyUI-oshtz-nodes.git ``` 2. Install the required dependencies: ```bash cd ComfyUI-oshtz-nodes pip install -r requirements.txt ``` ## Requirements - requests - torch - pydantic - Pillow - boto3 (>= 1.34.101) ## License This project is open-source and available under the MIT License.