Beyond-Physics Simulation Engine

Physics engines are mature because Newtonian mechanics generalize cheaply and pay off immediately in gameplay. Chemistry, geology, and biology don’t get the same treatment: games either fake them with simple crafting/alchemy recipes, or in the rare cases they go deeper, that depth gets hand-built per game rather than made reusable. On the other side, research and education software gets real scientific accuracy but isn’t built for game-speed interaction or reuse. Nothing sits in the middle: accurate enough to be genuinely data-driven, fast and reusable enough to actually drop into a game.

The idea is a standalone engine where objects carry real material properties (hardness, porosity, grain, toughness) that drive how they behave under interaction (struck, thrown, submerged, weathered, burned) rather than every interaction being scripted by hand. The fidelity bar: objects that are obviously different should behave obviously differently, and objects that are subtly different should still be discoverably different to an attentive player, not just re-skinned. A first version would prove that bar on just two material pairs (granite versus limestone, ash versus elm) in a bare-bones demo environment, before ever trying to generalize further.

Aimed at two audiences at once: game developers who want drop-in material realism without hand-authoring every interaction, and researchers or hobbyists who want an accurate, moddable, game-speed alternative to slower research software.