The studio is where I test whether automation is safe to run. Generation is cheap. Promotion is not. Anything that changes meaningful state needs a contract, verification and a named owner. Game production makes those boundaries concrete: keep, change or kill from evidence.
01 / How it is built
Systems
A game is a system under load. State, timing and transactions have to stay coherent when the player does something unexpected.
Production
One person cannot build a whole world at once. The useful move is shrinking the unit of work until the real constraint is visible.
Product
Art direction is not decoration. If the shot does not read, the product decision is not finished — keep, change or kill from evidence.
02 / The game
Too much world, then a smaller unit of work, then a failed picture, then a structural correction. The next move came from what the frame actually showed.
01
The first prototype generated a large 3D world. It proved the idea, then asked one person for environment, animation, navigation, collision, lighting and testing all at once.
02
A staged 2.5D approach kept the main machine realtime 3D and used layered scenery for depth. That was the right reduction. The first framed shot still failed: machine, route and background merged into one dark mass.
The layers had to be visible as layers before the picture could carry meaning.
Shape and scale first. Finish later.
Keep, change or kill from what the frame actually shows.
The same habit applies to operational systems: shrink the work until the constraint is visible, then build around it.
03 / Continue
Current professional work lives in the Pavilion. LinkedIn is the direct route.