Graphics Reference
In-Depth Information
Light source
Lens for collecting light
Camera
Surface
r
p
Sample point
Figure 4.7
The arrangement for measuring multiple subsurface scattering used by Henrik Wann
Jensen et al. (After [Jensen et al. 01b].)
The method is based on the assumption that the absorption in the medium is small
compared to the degree of scattering, so multiple scattering dominates.
The process works by photographing a physical sample of the surface
(
Figure 4.7
)
. The surface sample is illuminated by a light source focused into
a small beam, which simulates the incident light assumed in the derivation of the
diffusion equation. The camera is placed at a fixed angle from the light source (so
the Fresnel terms are constant), and a series of digital photographs is taken with
increasing exposure times. These are then combined into a
high dynamic range
(HDR) image (see Chapter 6), which has a much greater range of values than is
possible from a single exposure. Under the assumption that diffuse multiple scat-
tering is the dominant effect, the value of each pixel in the image is a constant
factor times
R
d
(
,where
r
is the distance from the illumination point
p
to the
point corresponding to the pixel.
This constant factor actually depends on two unknowns: the incident flux of
the light source, and how the camera converts the incoming light to pixel values.
Although there are ways to measure each of these, the authors found a way to
determine the constant factor more directly using a second image as a reference,
in which the original sample is replaced by a (nearly) ideal Lambertian standard
r
)
