FogExp2 transmittance at axial camera depth in metres:
T(d) = exp(-(density * d)^2). Consumers blend subject * T with fog.color *
(1 - T), following the same atmosphere law selected by viewer fog setup.
Depth is camera-forward distance rather than radial range. The squared law
is even in depth, so a behind-camera point has the same attenuation as its
positive-depth counterpart; consumers own visibility separately. Binary64
evaluation is a prediction of the shader law, not exact pixel agreement
with GPU precision, lighting, compositing or tone mapping.
Null or absent fog returns one. Scene-admitted density and finite depth are
caller premises; nonfinite direct inputs follow ordinary exponential arithmetic.
FogExp2 transmittance at axial camera depth in metres: T(d) = exp(-(density * d)^2). Consumers blend subject * T with fog.color * (1 - T), following the same atmosphere law selected by viewer fog setup.
Depth is camera-forward distance rather than radial range. The squared law is even in depth, so a behind-camera point has the same attenuation as its positive-depth counterpart; consumers own visibility separately. Binary64 evaluation is a prediction of the shader law, not exact pixel agreement with GPU precision, lighting, compositing or tone mapping.
Null or absent fog returns one. Scene-admitted density and finite depth are caller premises; nonfinite direct inputs follow ordinary exponential arithmetic.