Projects & Concepts
I work with AI tools, such as ComfyUI, and develop custom implementations for generative AI, LLMs and Agents, and more to accelerate workflows.
WAS Node Suite for ComfyUI
Enhancing ComfyUI with Over 210+ Custom Nodes
The WAS Node Suite is an expansive collection of powerful and modular nodes designed to supercharge ComfyUI workflows with advanced image processing, text operations, and AI model integration.
Key Features:
Headless & Service-First – Designed for businesses to utilize for services, nodes do not rely on JavaScript targeting the UI to operate.
Image & Video Processing - Apply filters, transformations, and generate animations.
AI-Powered Nodes - Integrates BLIP (Bootstrapping Language-Image Pre-training) and SAM (Segment Anything Model) for image analysis and segmentation.
Text Manipulation & Prompt Engineering - Modify, analyze, and optimize text for AI generation.
Workflow Enhancement - Tools for masking, depth mapping, face detection, grid composition, and more.
Model Management - Model loaders with name outputs for automated workflows.
Power Noise Suite
Advanced Latent Noise for Enhanced Image Generation
The Power Noise Suite is a specialized collection of nodes for ComfyUI, focusing on latent noise manipulation to achieve more consistent and detailed outputs in AI-generated images. This suite offers advanced tools for noise input tweaking, diffusion processes, and latent space adjustments, providing users with greater control over the generative process.
Key Features
Power-Law Noise - Experience a revamped noise system for more accurate noise representation in outputs.
Latent Noise Customization - Fine-tune noise parameters to influence image generation outcomes.
Latent Space Adjustments - Modify latent noise with various parameter controls, such as contrast, brightness, hue, etc.
Tonal Guide Integration - Utilize a tonal guide image to influence the tonal characteristics of generated images, allowing for thematic consistency. Together with ControlNet it was an early form of IPAdapter-like transfer of concepts from a image into a generation
ComfyGallery
A Lightweight, browser-based gallery to compliment ComfyUI
ComfyGallery was developed as a research project to explore purpose-built applications for image management for generations. This tool enables users to seamlessly browse, search, and manage images from ComfyUI input/output directories, as well as user defined locations. User can easily copy their workflows to load into ComfyUI.
Key Features
Directory Browsing: Navigate through ComfyUI's input and output directories effortlessly.
Custom Directory Integration: Add and manage additional image directories as needed.
Image Deletion: Remove unwanted images directly from the interface.
Advanced Search: Locate images by filename or embedded workflow data within PNG files.
