WORLD FORGE / SCIENTIFIC MATERIALISATION · 01 NOV 2026

Make the world
behave.

Equations drive trajectories, wind fields, particles and fluid motion. We can connect those calculations to cinematic worlds—but we must distinguish a numerical demonstration from a physically validated solver and from a finished film.

01 / THE FOUR COMPUTATIONAL LAYERS

The laws change.
The discipline stays.

CHAOS / RK4 NUMERICS

Lorenz, Rössler & Thomas

Real fourth-order Runge–Kutta integration of nonlinear differential equations. A tested small initial separation can grow over time at selected parameters. This is not a universal chaos claim.

Examine the attractors ↗
FLUID / EDUCATIONAL 2D APPROXIMATION

Navier-like flow fields

Local two-dimensional advection and pressure projection generate evolving fluid-like fields. The experimental “Navier Unbound” visualisation adds dramatic vortices and particle effects; it is not a validated 3D Navier–Stokes solution.

Try Navier Unbound ↗
MATTER / DISCRETE PARTICLE MODEL

Living Matter

Positive-mass particles exchange equal and opposite internal impulses while external wind and collisions influence motion. A beautiful point cloud is not a validated continuum fluid or a human model.

Explore 3D matter ↗
FLOOD / CONSERVATIVE SHALLOW WATER

Water around a taxi

A separately tested depth-averaged Saint-Venant solver uses conservative HLL fluxes and a fixed vehicle-shaped obstacle. This is actual numerical flood physics, not moving-car coupling, spray or final cinematic footage.

Inspect numerical evidence ↗

02 / MEASURED PHYSICS · ONE TEST CASE

Water is accounted for.

In a specific source-free two-second conservative shallow-water test, the calculated total water volume was unchanged. This is a numerical mass-balance check on an idealised grid, not a measurement of a real South African flood.

166.89868m³ water initially and finally
0.0 m³Reported mass-balance error in this run
1,232Grid cells, including 21 solid obstacle cells

44 × 28 cells · 0.65 m spatial steps · 144 computed time steps · 49 recorded times at 24 samples/s. These are simulation states, not movie frames. Independent source tests passed; actual field validation, moving boundaries, wetting/drying calibration and three-dimensional effects remain future work.

02A / GENESIS 001 · A GRAVITY EXPERIMENT

Let matter find its form.

Genesis 002 adds an analytic Jeans instability comparison: gas pressure supports short-wavelength waves, whereas self-gravity can amplify long-wavelength perturbations. Explore the linear Jeans benchmark. This is not yet a nonlinear gas solver.

The Genesis 001 research lab computes pairwise softened Newtonian attraction, collision merging, velocity-Verlet steps and optional non-physical damping. Two views observe one deterministic state. Event replay and exported receipts allow reproducibility and mass-budget checks. It remains a dimensionless educational experiment—not gas thermodynamics, real stellar ignition, a verified cosmological model or artificial biological life.

03 / CINEMA / WHAT EXISTS AND WHAT DOES NOT

From physical rules
to performed worlds.

Native 3D assets for The Last Taxi include a spatial master with 920 meshes, 168 materials and six cameras, plus original stadium and vehicle geometry. An animation exporter passed synthetic-fixture tests, but that does not establish a completed performance in the real owner model. Final facial acting, coupled fluid/vehicle animation, rights-cleared sound and cinema-quality footage remain production gates.

Evidence before spectacle.

This page is not a live hardware dashboard and does not claim measured private-device status, a deployed resource scheduler, verified local-model training gains, a real-time three-dimensional CFD solver or a completed two-hour film. Both feature films are not finished. Private execution authority remains separate from educational browser simulations.