The scene simulates a real portrait lighting rig: a key (spot) light orbits the head at a chosen azimuth θ and elevation φ, a dimmer fill light sits on the opposite side to hold detail in the shadows, and both cast real shadow-mapped light onto a sculpted head mesh (brow ridge, nose, cheekbones) — every shadow you see is rendered by an actual depth-based shadow map, not painted on.
Key position: x = R·sinθ·cosφ, y = R·sinφ, z = R·cosθ·cosφ
Key:Fill ratio (stops) = log2( I_key / I_fill )
- Rembrandt (θ≈45°, φ≈45°) — the nose's shadow merges with the cheek's shadow, trapping a small illuminated triangle of light under the far eye.
- Split (θ≈90°, φ≈low) — the key sits at 90° to the camera axis, so the face splits into one fully lit half and one fully shadowed half.
- Loop (θ≈35°, φ≈30°) — the nose casts a small shadow "loop" onto the cheek that does not touch the cheek's own shadow.
- Butterfly (θ≈0°, φ≈65°) — the key sits high and centered on the camera axis, dropping a small symmetric shadow directly under the nose (a classic beauty-lighting pattern).
- Broad / Short — with the head turned, "broad" lights the side of the face turned toward the camera (looks wider), "short" lights the side turned away (looks slimmer).
- Softbox size — a larger virtual softbox widens the spotlight's penumbra and softens shadow edges, mimicking how a bigger light source relative to the subject produces softer falloff (the inverse-square law still governs intensity).