Descriptions:
Hasan Aboul Hasan demonstrates how to build fully functional 3D browser games using AI prompting alone — no manual coding required — by mastering five specific technical vocabulary terms that unlock dramatically different results from the same underlying prompt.
The five “unlock words” are: visual style descriptors (low-poly, pixel art, voxel, cell-shaded, isometric), Three.js as the JavaScript rendering engine, screenshot-based prompting for cloning existing games, game genre keywords like “endless runner” that convey complex mechanics in two words, and physics libraries — Matter.js, Cannon.js, and Ammo.js — for realistic collision and gravity. Each concept is illustrated with a working game: Lumber Valley (a math education game), a Lumosity brain-game clone built from a single screenshot, Birds Runner (a Subway Surfers-style game), and an FPS scene rendering over two million instanced grass blades in a live browser without lag. The instancing example is the most technically instructive: simply asking for “2 million grass blades” would crash the browser, but knowing to write “InstancedBufferGeometry” triggers Three.js’s optimized batch-rendering path.
The video links to two free interactive guides — one covering Three.js concepts end-to-end, one covering JavaScript physics and animation libraries — along with a reusable prompt library. It is well-suited for educators, hobbyists, and indie developers who want to build polished game prototypes by treating AI as a code generator rather than a search engine.
📺 Source: Hasan Aboul Hasan · Published May 02, 2026
🏷️ Format: Tutorial Demo
