3D Thunder is a solo experiment by Enes K, testing how far current AI can go at turning a plain description into an organic, figurative 3D mesh — the kind you could actually hold and print, not just render on a screen.
The name says it: 3D for the form, Thunder for the speed. The best creative tool gets out of your way and lets imagination win. 3D Thunder pairs an AI assistant (Thunder Bot) with an in-browser studio so you can describe, generate, refine, and export a model in one place.
This is a test bed for 3D agents, not a product roadmap. A deliberate goal was to separate the 3D agent — the part that holds all the know-how of 3D mesh generation, AI, and post-processing — from the web tool itself. That separation keeps each side cleaner, safer, and easier to iterate on without breaking the other.
Under the hood, the agent combines GPT-driven image generation with a modern 3D generation model to produce the meshes. It runs as its own Python service — the FastAPI server the frontend knows as Thunder Bot — and after generation it applies post-processing adjustments to the mesh to encourage a stable, dependable result.
3D Thunder is in active testing. Results can vary, and every generation helps make the experience more reliable. Guest access lets you try it instantly — no account required.