Building a maze that remembers where you face

The chase began with a constraint: movement should feel immediate, but never comfortable. Free movement gave the player too much certainty. We wanted each decision to feel like choosing a corridor while something followed just beyond the flashlight.

The nodes are the source of truth

Every meaningful position in the maze is represented by an authored node. The hallway meshes can change around those nodes without rewriting the topology. From each position, the maze infers the directions the player can take based on distance, alignment, and the direction the camera is currently facing.

That gives us a small set of choices—forward, left, right, and back at dead ends—without turning the scene into a conventional grid.

The maze is not asking where you want to walk. It is asking which mistake you want to make next.

Pressure without a fail screen

The sequence currently has no traditional fail state. Pressure escalates through sound, darkness, camera behavior, and the suggestion of something closing the distance. The player can continue choosing paths, but each choice becomes harder to make calmly.

The next step is tuning the transition from exploration into pursuit so that the change is felt before it is fully understood.