2 posts tagged with "stable-diffusion"

Over the last couple of months I contributed to the open-source Invoke AI project. Invoke is a canvas-based generative AI tool that feels closer to Photoshop or GIMP than a single prompt box, which makes fast iteration and visual control a lot easier. It combines a layered canvas with node-based workflows, so you can mix prompts, masks, control images, and iterative edits without leaving the app.

My contributions focused on general improvements across the stack:

• UI and workflow polish to make the canvas experience smoother.
• TypeScript improvements on the frontend.
• Python-side tweaks for stability and performance.

If you want to check it out, the code is on GitHub: Invoke AI and the Invoke AI Launcher.

Recently, I have been experimenting with Stable Diffusion and often found myself needing a quick way to view image metadata, whether it was to check the prompt, get the pixel size (dimensions), or other details.

This led me to create a simple web app that allows you to quickly extract and display media file metadata. The app runs entirely on the client side, meaning no data is uploaded or shared. It supports all major browsers, although Chrome provides the best experience.

Key Features:

• Read Media File Metadata: Easily extract details from image and video files.
• Read EXIF Image Metadata: See embedded data like camera settings or location info.
• Read Safetensor File Metadata: Extract tensor details from model files.
• Drag and Drop: Conveniently drag and drop your files into the app.
• Paste Files: A Chrome-specific feature that allows you to paste files directly.
• Client-Side Processing: Private and fast, even when processing large files.

You can try it here: https://meta.rikublock.dev