The first deterministic generator could produce room data from templates and seeds, but object placement alone was an insufficient long-term model. A GUTTA-X room needed to feel built by rules rather than filled by a randomizer.
The response was a dual grammar. Map grammar composes the playable problem: bands, lanes, cover cells, enemy pressure regions, boundaries, and future exits. Structure grammar describes how selected regions become barriers, walls, scanner boxes, burned-car silhouettes, pillars, rails, plates, beams, and boundary nodes. A separate visual layer remains responsible for palette, scale, accents, and clarity.
This separation made derivation inspectable and created validation points before Unity objects were spawned. Map plans, structure plans, and final room data could each be checked, while the same seed preserved a path back to the decisions that produced a result.
By the end of the recorded batch sequence, all five initial LAX-13 templates had map grammar programs and could be adapted into the common room-data model for diagnostics. The live sandbox still retained the earlier generator, keeping the migration reversible while the grammar path matured.
← Back to GUTTA-X