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.