Graphics Reference
In-Depth Information
ior. Arno was working on light scattering in general filaments at the University of
Bonn [Zinke and Weber 07]. The work of Marschner and colleagues led him to
become interested in hair rendering. Unlike the typical “hacks” sometimes used in
the visual effects industry, the approximations of Zinke and his colleagues come
from physically based insights. The results have visual quality comparable to the
methods described earlier, but they are much faster to compute.
The basic idea of the “dual scattering” method is to split the multiple scat-
tering into local and global components. The global component represents the
amount of light from the source that reaches a particular point in the hair volume.
It is computed by approximating the overall density of the hair volume and com-
puting the scattering along a single light path to the source, which the authors call
a
shadow path
. The global multiple-scattering component depends primarily on
forward scattered light; backward scattering is treated as part of the attenuation.
In contrast, the local multiple-scattering component comes from all multiple scat-
tering within a neighborhood of a particular hair fiber, which normally includes
several other nearby fibers. The primary constituent of local multiple scattering
is backscattering from these nearby hair fibers. The total multiple scattering, as a
function of position and incident and scattering directions, is a sum of the global
scattering function
G
and the further local scattering function
L
:
Ψ
Ψ
G
G
L
,
ω
i
,
ω
s
)=
Ψ
,
ω
i
,
ω
s
)+
Ψ
,
ω
i
,
ω
s
)
Ψ
,
ω
i
,
ω
s
)
.
Ψ
(
x
(
x
(
x
(
x
Figure 8.64
The basic concept of dual scattering method is to split the scattering into the
global scat-
tering
of light from the light source and the
local scattering
within a local cluster of hair
fibers. (From [Zinke et al. 08] c
2008 ACM, Inc. Included here by permission.) (See
Color Plate XV.)
