Graphics Reference
In-Depth Information
section, then the azimuthal dependence reduces to the difference
φ r φ i of the
azimuthal angles
becomes an expression in-
volving two-parameter functions, which can be precomputed and stored in a 2D
table (i.e., a texture map).
The longitudinal functions M were fit to observations; the azimuthal functions
N came from physical considerations. The M functions were found to be fairly
close to the Kajiya-Kay model, although the specular highlight is shifted by the
angle
φ i and
φ r . Therefore, S
( θ i , θ r , φ i , φ r )
of the cuticle scaled as described previously. In the final model, the M
functions are a Gaussian functions centered at the half-angle between the incident
and reflected directions, but shifted by a multiple of
α
according to the number
of reflection and refraction events involved. The M TT and M TRT lobes are shifted
towards the tip of the hair fiber, whereas the M R is shifted toward the root.
Marschner and his colleagues were not the first to perform direct measure-
ments of reflection in hair fibers (this had been done years earlier in the context
of cosmetic science) but they were the first to perform full hemispherical mea-
surements of hair reflection. They also recorded a special set of measurements by
placing the light source and the camera at a fixed equal angle of incidence and
reflection, then rotated the camera 360 . This results in a direct azimuthal mea-
α
10
10
10
θ i = 0º
60
θ i = 15º
60
θ i = 30º
60
30
30
30
f
f
f
1
1
1
-30
-30
-30
-60
-60
-60
10
10
10
60
60
60
θ i = 45º
θ i = 60º
θ i = 75º
30
30
30
f
f
f
1
1
1
-30
-30
-30
-60
-60
-60
Figure 8.54 Plots of azimuthal measurements. For each fixed incident direction θ i , a sensor is placed at
the specular direction and rotated around the fiber. The plots show the off-specular peaks
known as glints evolving toward the normal specular highlight as θ i increases.
(From
[Marschner et al. 03] c
2003 ACM, Inc. Included here by permission.)
Search WWH ::




Custom Search