Azu: Generating 3D game assets using a Kinect!

Creating high-fidelity 3D assets is a very challenging expensive part of game dev. Azu attempts to solve this using outdated console hardware, enabling indie developers to rapidly prototype and develop realistic games. The project addresses the limitations of traditional 3D asset creation by enabling the capture of geometrically consistent, watertight models from real-world objects using a TSDF-based pipeline, reducing reliance on manual sculpting, retopology, and costly industrial scanners. The hardware used is a Kinect v1 from 2010, which is niche but extremely cheap today. The usecases of this project range from architectural, film/game production usecases to industrial capture.

https://github.com/Netherquark/azu/

Building with agentic programming, developing production ready optimised ML applications. GPU programming. Computer Vision & the mathematics behind it.