Project

cubAI

A 3D raycasting engine with AI-assisted asset generation for weapons and textures.

Raycasting AI Assets Hackathon
GitHub DeepWiki Podcast

TLDR

Supercell AI Hackathon

1st Prize - Winners

The Supercell AI Hackathon brought together over 300 developers, designers, artists, and creatives from 38 countries for a full weekend of building and experimenting at Hive Helsinki. The goal was to explore and experiment with how AI can shape the future of games and entertainment, from gameplay mechanics to storytelling and creative tools.

Hive Helsinki team holding the Supercell AI Hackathon trophy Read the full article

code

Code snapshot

Quick project overview + documentation.

The cubAI repository implements a 3D raycasting engine with AI-assisted asset generation capabilities. This system combines classic first-person shooter rendering techniques with modern generative AI to create a hybrid game development environment. The core engine performs real-time 3D perspective rendering from 2D grid maps using raycasting algorithms, while the AI integration enables dynamic generation of game assets such as weapons and textures through OpenAI API calls.

"Asset Loading System""OpenAI API""api.c::generate_with_gpt_image()""sprites.c::ai_worker()""movement.c::loop_hook()"User"Asset Loading System""OpenAI API""api.c::generate_with_gpt_image()""sprites.c::ai_worker()""movement.c::loop_hook()"User"Press '2' Key""Enter Gun Description""Submit Description Text""Start pthread with description""Call generate_with_gpt_image()""HTTP POST with image prompt""Return generated image data""Save image to filesystem""Load texture into ai_hands array""Set is_gun_ready = true""Display new weapon in HUD"
Read the technical overview

Podcast

Listen in

A talk with Playing with inference about our experience in the hackathon and some reflections on the build.

Open on YouTube