4D AI Sandbox Engine 🪐

hey everyone, making my first post here.

i’ve been working on a 4d ai sandbox prototype that takes a prompt, uses an llm to compile glsl code on the fly, and injects it straight into a live webgl runtime loop.

What is running under the hood?

• GPGPU pipeline: all particle calculations are offloaded entirely to custom fragment shaders using texture buffers so the cpu doesn’t choke.
• Live rendering: currently pushing a 128x128 grid (16,384 nodes) completely smooth on localhost.
• Terminal stream: built a custom execution stream log on the bottom left to track engine boots and live mapping errors.

What is the plan for v1.0?

I want to turn this prototype into a practical tester for 3d assets. Next steps:

1. Map the gpgpu coordinates into actual solid surfaces using three.js InstancedMesh.
2. Build an exporter node so users can download the generated shapes as standard .gltf files.
3. Finish the sidebar chart to track structural/physics stress metrics over time.

P.S. — attached a screenshot of how the dashboard looks right now. let me know if anyone wants to geek out over gpu textures or shader pipelines!