Effective potential V(r) of the last probe (white line = its energy E, dot = now)
Light deflection angle vs impact parameter b: integrated rays (dots) and exact quadrature (line)
running...
The science
Lengths are in the Schwarzschild radius, the only length scale of a non-rotating hole. The geometry in these units does not depend on the mass; the mass only sets the kilometre, hertz and kelvin values in the readouts.
r_s = 2GM/c^2 = 2.953 km (M / M_sun)
photon sphere r = 1.5 r_s (unstable circular light orbit)
ISCO r = 3 r_s = 6GM/c^2
shadow b_c = (3 sqrt3 / 2) r_s = 2.598 r_s
Hawking T = hbar c^3 / (8 pi G M k_B)
Flamm funnel z(r) = 2 sqrt(r_s (r - r_s)) (spatial slice t = const, theta = pi/2)
Matter follows timelike geodesics with conserved E and L (c = 1, r_s = 1):
(dr/dtau)^2 = E^2 - (1 - 1/r)(1 + L^2/r^2), dphi/dtau = L/r^2
circular: L^2 = r^2 / (2(r - 3/2)), stable for r > 3
periapsis advance per orbit = 4 INT_0^{pi/2} dchi / sqrt(u3 - u) - 2 pi , u3 = 1 - u_a - u_p
(weak field: 6 pi G M / (c^2 a (1 - e^2)))
Light follows null geodesics: d2u/dphi2 + u = (3/2) r_s u^2, u = 1/r. In the Lensed view the page integrates this equation on the CPU (RK4) once for a table of view angles from the camera's radius, then every GPU pixel looks up where its ray ends up on the sky and where it crosses the disk plane (primary image, secondary image under the hole, photon ring). Disk colour: thin-disk temperature T proportional to r^(-3/4)(1 - sqrt(3/r))^(1/4), brightness ~ T^4, times the Doppler and gravitational shift g^4 with g = sqrt(1 - 3/(2r)) / [ sqrt(1 - r_s/r_obs) (1 + Omega b cos i) ] for circular geodesic motion, Omega = sqrt(GM/r^3).
Simplifications and honesty: only Schwarzschild (no spin, so no frame dragging; real ISCOs of Kerr holes lie between 1 and 9 GM/c^2). Probes and grains are test masses with no mass, gas, pressure, viscosity, radiation or gravitational waves. In the Lensed view the sky and the disk are ray-traced, but the 3D meshes (probes, trails, rings, horizon) are drawn at their Schwarzschild-coordinate positions without being lensed, and a probe behind the hole is only hidden by the horizon sphere, not by the full shadow. The observer is static. Disk colours are schematic: the temperature is mapped to the visible range (a real stellar-mass disk glows in X-rays), the texture structure is a Blender-baked turbulence pattern, not an MHD simulation. Probes are drawn about 0.5 r_s long for visibility; a real 10 M_sun hole has r_s = 29.5 km. Schwarzschild r is a circumferential radius, not a distance, so the Flamm funnel is an isometric picture of the space, not of spacetime curvature in time.