3.2SimulatingInfraredSensorEffects 49
pheric transmittance models for this purpose such as LOWTRAN
1
and MOD-
TRAN,
2
but we can get by with a simpler approximation. Atmospheric transmit-
tance from water vapor for a wavelength
in micrometers is approximated
[Bonjean et al. 2006] by
20mω
λωλ
ττ
.
Here,
is the depth of precipitable water in millimeters; this is a measure of the
air’s humidity. Typical values range from 10 to 60 mm. The value
ωλ
τ
is given
by:
2
0.0075
10
ωλ
τ
.
The value of m represents the air mass, which is a measure of how much atmos-
phere the heat source has passed through before reaching the sensor. This is a
unitless value, normalized to one for a vertical path through the atmosphere at sea
level. More sophisticated implementations for “serious games” may also take
atmospheric scattering and scattering from dust into account.
3.2SimulatingInfraredSensorEffects
The Stefan-Boltzmann law gives you the simulated radiation being emitted from
an object in your scene, but that’s not directly useful for rendering. You need to
map these radiation values into monochromatic luminance values that are dis-
playable—real sensors have a feature called auto gain control that does this.
Typically, they compute the range of radiation values in the visible scene and
map these values linearly from 0.0 to 1.0. The range being mapped can be quite
high—the night sky’s temperature is near absolute zero, while a missile’s exhaust
could be several thousand Kelvins [Thomas 2003].
The actual colors used to represent this radiation vary by the type of sensor.
Most forward-looking infrared (FLIR) sensors used by the military represent IR
in black and white, and are configurable between “black-hot” and “white-hot”
polarities (shown in Figure 3.2). Thermal imaging sensors may blend between
red for hot spots in the image and blue for cold spots. These are all easy effects to
achieve in a fragment program.
In addition to modeling gain and polarity, there are many imperfections in IR
sensors that we can also emulate. Older CRT-based devices were prone to effects
1
See http://www1.ncdc.noaa.gov/pub/software/lowtran/.
2
See http://www.modtran.org/.