I've been working on JSAR, a new open-source spatial browser engine. Our latest work was to render a stereo 3D photo using just a simple HTML tag (<img spatial="stereo">), without needing WebGL or shaders.

The interesting part is how I built it. I paired with GitHub's Coding Agent for the entire process. It was fascinating to see it suggest entire C/C++ functions for parsing attributes and calculating UV coordinates for the renderer. The whole implementation is in a single commit: https://github.com/M-CreativeLab/jsar-runtime/commit/ff8e291...

I wrote this post detailing the technical breakdown. The goal is to make spatial web development as easy as traditional web dev.

Happy to answer questions about the engine's architecture, C/C++, or what it's like to pair-program with an AI.