Animal Stack

Game concept & AR design
The idea takes the tension of physics stacking games into AR: place objects on a detected real-world surface, then check the balance from different angles before committing. The finished game is meant to stack soft 3D animals, but this prototype uses physics test objects while the interaction and stacking systems get built. I scoped it around a small table surface, so it can be played seated instead of forcing the player to walk around holding up a phone camera.

Technical approach
I set up Unity AR Foundation with Android ARCore support and tackled the highest-risk feature first: detecting a flat indoor surface and placing a fixed starting platform on it. From there I wrote the core interaction in C#: a two-stage spawner that places the platform before it spawns stack objects, touch-based preview movement, two-finger rotation, and physics drops onto the current stack. I also prototyped a center-of-mass stability system that checks the tower against the platform's support area and can trigger a topple once the stack goes unstable.

Outcome & lessons learned
The result is a running Android prototype: it detects a surface, places a starting platform, and drops physics objects onto it. The animal art, full game loop, scoring, loss state, onboarding, and polish stayed in the backlog when the sprint ended. The biggest lesson was to front-load the uncertainty. Proving AR surface detection early told me the core idea was technically viable before I spent time on content or UI.

How I used AI
I used AI tools for planning, structuring the Trello backlog, reference-image prompts, and learning Unity's C# workflow, and reviewed the output before using it. I implemented and verified the AR setup, the interaction scripts, the physics experiments, and the device testing myself.