Liked this Three.js demo? The good news: it was built using Google Gemini 2.5 Pro - no coding skills required. The catch? A base code (mine, but different) was used to guide the initial setup. Let’s walk through the process. A thread 🧵👇
Step 1: I started with a CodePen scene I’d already built using Three.js. CodePen🔗: https://codepen.io/VoXelo/full... This acted as the foundation for what I wanted to create next.
Step 2: I fed that CodePen into Google Gemini 2.5 Pro, asking it to iterate - changing shapes, colors, shaders etc. You can guide it similarly: - Swap geometries - Try different materials - Experiment with lighting or shaders
Sometimes the results aren’t perfect on the first try - that’s normal. 💡 Tip: Iterate patiently. Small, specific prompts work best.
Why use my code instead of starting from scratch if you are a non-coder? Because it nudges the AI into setting up your scene with better practices and newer Three.js versions - unlike the outdated defaults AI often picks.
Want to try it yourself? My CodePen 🔗: https://codepen.io/VoXelo Browse, fork, or remix any of the scenes I’ve shared there. Pick one, feed it into Gemini with your creative changes. If you like my work and find it useful, please don’t forget to like and repost the main post.