mshumer
Featured
Claude-of-Duty
A browser-based first-person shooter built entirely with procedural generation and orchestrated AI agents, featuring 55k lines of Three.js/WebGL2 code and no art assets.
AI repository intelligence
FR-AI / ANALYSISWhy it is worth attention
Demonstrates rapid creation of a polished 3D interactive showcase using AI coding and image generation, with full scroll, drag, and touch controls.
Who it is for
- Three.js developers
- Creative coders
- Portfolio builders
- AI-assisted prototyping enthusiasts
Use cases
- Interactive product display for beverages
- Learning scroll-driven 3D transitions
- Exploring AI-generated texture workflows
- Building a quick interactive demo for portfolio
Strengths
- Built in 20–30 minutes with AI assistance
- Supports mouse, touch, keyboard, and scroll controls
- Responsive camera framing for hero can
- Includes draggable horizontal lineup and spotlight effects
Considerations
- Textures are AI-generated and not official brand artwork
- Demo uses trademarked brand names without permission (for demonstration only)
- May lack extensive cross-browser or performance optimization due to rapid development
README quick start
CAN/DO
A scroll-driven Three.js soda-can showcase with a draggable lineup, a dark studio scene, and a top-lit hero can.
Built in 20–30 minutes
This demonstration was built by GPT-5.6 Sol Ultra in roughly 20–30 minutes, with Fatih directing the visual and interaction changes. The process started from a short reference video and moved through a handful of prompts: build the Three.js scroll choreography, fix the full-height framing, add horizontal can browsing, simplify the typography, generate recognizable label textures, and finish with a dark spotlighted stage.
What is in the demo
- Seven textured cans arranged as a draggable horizontal lineup.
- Scroll-driven transitions between lineup, hero, burst, and finale states.
- A warm spotlight that follows the highlighted Sprite can from above.
- Responsive camera fitting so the hero can remains fully visible.
- Mouse, trackpad, touch, button, and keyboard controls.
Assets and texture generation
- 3D can: “Soda Can” by Jeremy, downloaded from Poly Pizza and used under CC BY 3.0.
- Can textures: created with OpenAI ImageGen through the Codex imagegen skill as flat, cylindrical 2:1 label textures.
- Rendering and interaction: Three.js and Vite.
The label images are AI-generated demonstration graphics, not official packaging artwork. Coca-Cola, 7UP, Dr Pepper, Sprite, Fanta, Pepsi, and Mountain Dew are trademarks of their respective owners.
Run locally
npm install
npm run dev
Create a production build with:
npm run build
The main branch is automatically built and deployed to GitHub Pages through
the included GitHub Actions workflow.
Description
A scroll-driven Three.js soda-can showcase built by GPT-5.6 Sol Ultra in 20–30 minutes.
Related repositories
Similar projects matched by category, topics, and programming language.
gnipbao
Featured
story-to-handdrawn-video
A Remotion-based tool that converts Chinese story text or ordered hand-drawn images into vertical hand-drawn diary-comic animation with handwritten captions, left-to-right reveals, optional page-curl transitions, and silent H.264 output for post-production dubbing.
0xwilliamortiz
Featured
ponytail-improved
Ponytail is a plugin for AI coding agents that enforces a disciplined ladder of reuse before writing code, reducing code volume by roughly 54% while preserving safety.