I spent the entire day today working on Excalidraw, and I made two exciting improvements:
- I improved caching for nested images and rendered PDF pages. The result is much faster load times for large scenes, such as my Idea Integration Boards.
- I fixed several long-standing bugs with AI features like “Diagram to Code” and the ExcaliAI script. I also finally implemented support for different AI endpoints, not just OpenAI.
For now, these changes are only available as a beta release. I’m looking for testers who have Claude, Gemini, or xAI subscriptions, since I was only able to test with OpenAI myself.
You can find more details about the release in the Beta channel: Excalidraw 2.23.0-beta-1
Since canceling my Midjourney subscription earlier this year, I’ve really missed image editing features. ExcaliAI with gpt-image-1-mini is performing surprisingly well. Here’s an example where I took an image of a monkey, added a mask around the monkey, then submitted both the image and the mask to gpt-image-1-mini using ExcaliAI “Edit Image” option, with the prompt: “Monkey looking at a flying saucer in the sky”.
I think the end result turned out pretty nice:
