Shading & Material Appearance

© ACM. All rights reserved. This content is excluded from our Creative Commons
license. For more information, see http://ocw.mit.edu/help/faq-fair-use/.
1
MIT EECS 6.837 – Matusik
Lighting and Material Appearance
• Input for realistic rendering
– Geometry, Lighting and Materials
• Material appearance
– Intensity and shape of highlights
– Glossiness
– Color
– Spatial variation, i.e., texture (next Tuesday)

© source unknown. All rights reserved. This content is excluded from our Creative Slide Addy Ngan
Commons license. For more information, see http://ocw.mit.edu/help/faq-fair-use/.
2
Unit Issues - Radiometry
• We will not be too formal in this class
• Issues we will not really care about
– Directional quantities vs. integrated over all directions
– Differential terms: per solid angle, per area
– Power? Intensity? Flux?

• Color
– All math here is for a single wavelength only; we will
perform computations for R, G, B separately
• Do not panic, that just means
we will perform every operation three times, that is all
3
Light Sources
• Today, we only consider point light sources
– Thus we do not need to care about solid angles
• For multiple light sources, use linearity
– We can add the solutions for two light sources
• I(a+b) = I(a) + I(b)
– We simply multiply the solution when we scale the light
a
intensity b

• I(s a) = s I(a)

Yet again, linearity

is our friend!
4
Intensity as Function of Distance
• 1/r2 fall-off for isotropic point lights
– Why? An isotropic point light
outputs constant power per solid
angle (“into all directions”) r2
r1
– Must have same power in all
concentric spheres
• Sphere’s surface area grows with r2 => energy obeys 1/r2
• … but in graphics we often cheat with or ignore this.
– Why? Ideal point lights are kind of harsh
• Intensity goes to infinity when you get close – not great!
– In particular, 1/(ar2+br+c) is popular

5
Incoming Irradiance
• The amount of light energy received by a surface
depends on incoming angle
– Bigger at normal incidence, even if distance is const.
• Similar to winter/summer difference
• How exactly? n

– Cos θ law θ

– Dot product with normal

Surface

6
Incoming Irradiance for Pointlights
• Let’s combine this with the 1/r2 fall-off:

– Iin is the irradiance (“intensity”) at n

surface point x
θ
– Ilight is the “intensity” of the light
– θ is the angle between light direction l
and surface normal n l
– r is the distance between light and x.

x Surface
7
Directional Lights
• “Pointlights that are infinitely far”
– No falloff, just one direction and one intensity

n l
– Iin is the irradiance at surface point x
from the directional light θ

– Ilight is the “intensity” of the light

– θ is the angle between light direction l
and surface normal n
• Only depends on n, not x!
x Surface
8
Spotlights
• Pointlights with non-uniform directional emission
• Usually symmetric about a central
direction d, with angular falloff
– Often two angles
• “Hotspot” angle:
No attenuation within the central cone
• “Falloff” angle: Light attenuates
from full intensity to zero intensity
between the hotspot and falloff
angles d

• Plus your favorite distance

falloff curve 9
Spotlight Geometry

(direction d)

hotspot angle

Adapted from
POVRAY documentation
10
Questions?

11
Quantifying Reflection – BRDF
• Bidirectional Reflectance
Distribution Function
• Ratio of light coming from one
direction that gets reflected in
another direction © ACM. All rights reserved. This content is excluded from
our Creative Commons license. For more information,
see http://ocw.mit.edu/help/faq-fair-use/.

– Pure reflection, assumes no

light scatters into the Incoming
material direction

Outgoing
• Focuses on angular aspects, not direction

spatial variation of the material

• How many dimensions?
12
BRDF fr

13
BRDF fr

14
BRDF fr
• Relates incident irradiance from l = light direction
every direction to outgoing light. (incoming)
v = view direction
How?
(outgoing)

15
BRDF fr
• Relates incident irradiance from l = light direction
every direction to outgoing light. (incoming)
v = view direction
How?
(outgoing)

• Let’s combine with what

we know already of pointlights:

16
2D Slice at Constant Incidence
• For a fixed incoming direction,
view dependence is a 2D
spherical function
– Here a moderate specular
component © ACM. All rights reserved. This content is excluded from
our Creative Commons license. For more information,
see http://ocw.mit.edu/help/faq-fair-use/.

highlight
incoming

incoming

Courtesy of Mitsubishi Electric Researh Laboratories, Inc. Used with permission.

Example: Plot of “PVC” BRDF at 55° incidence 17

Demo

18
Isotropic vs. Anisotropic
• When keeping l and v fixed, if rotation of surface
around the normal does not change the reflection, the
material is called isotropic
• Surfaces with strongly oriented microgeometry
elements are anisotropic
• Examples:
– brushed metals,
– hair, fur, cloth, velvet

Westin et.al 92
19
Demo

20
How do we obtain BRDFs?
• One possibility: Gonioreflectometer
– 4 degrees of freedom

Source: Greg Ward

© ACM. All rights reserved. This content is excluded from our Creative Commons
license. For more information, see http://ocw.mit.edu/help/faq-fair-use/. 21
How Do We Obtain BRDFs?
• Another possibility: Take pictures of spheres coated
with material, rotate light around a 1D arc

Matusik et al.
– This gives 3DOF => isotropic materials only

22
Parametric BRDFs
• BRDFs can be measured from real data
– But tabulated 4D data is too cumbersome for most uses
• Therefore, parametric BRDF models represent the
relationship between incident and outgoing light by
some mathematical formula
– The appearance can then be tuned by setting parameters
• “Shininess”, “anisotropy”, etc.
– Physically-based or Phenomenological
– They can model with measured data (examples later)
• Popular models: Diffuse, Blinn-Phong, Cook-
Torrance, Lafortune, Ward, Oren-Nayar, etc.
23
Questions?

24
Ideal Diffuse Reflectance
• Assume surface reflects equally in all directions.
• An ideal diffuse surface is, at the microscopic level, a
very rough surface.
– Example: chalk, clay, some paints

Surface

25
Ideal Diffuse Reflectance
• Ideal diffuse reflectors reflect light according to
Lambert’s cosine law
– The reflected light varies with cosine even if distance to
light source is kept constant

26
Ideal Diffuse Reflectance
• Ideal diffuse reflectors reflect light according to
Lambert’s cosine law
– The reflected light varies with cosine even if distance to
light source is kept constant

Remembering that incident irradiance depends on cosine,

what is the BRDF of an ideally diffuse surface?

27
Ideal Diffuse Reflectance
• The ideal diffuse BRDF is a constant fr(l, v) = const.
– What constant ρ/π, where ρ is the albedo
• Coefficient between 0 and 1 that says what fraction is reflected
– Usually just called “diffuse color” kd
– You have already implemented this by taking dot products
with the normal and multiplying by the “color”!

28
Ideal Diffuse Reflectance
• This is the simplest possible parametric BRDF
– One parameter: kd
• (One for each RGB channel)

29
Ideal Diffuse Reflectance Math
• Single Point Light Source
– kd: diffuse coefficient (color)
– n: Surface normal.
– l: Light direction.
light source
– Li: Light intensity
– r: Distance to source
– Lo: Shaded color n
θ r
l

Surface

30
Ideal Diffuse Reflectance Math
• Single Point Light Source
– kd: diffuse coefficient (color)
– n: Surface normal.
– l: Light direction.
light source
– Li: Light intensity
– r: Distance to source
– Lo: Shaded color n
We do not want  light rfrom below the
l
surface! From now on we always
Do not forget assume (on this lecture) that dot
to normalize products are clamped to zero and
Surface
your n and l! skip writing out the max().
31
Questions?

32
Ideal Specular Reflectance
• Reflection is only at mirror angle
• View dependent
– Microscopic surface elements are usually oriented in the
same direction as the surface itself.
– Examples: mirrors, highly polished metals.
light source

θ θ
r l

Surface

33
Recap: How to Get Mirror Direction
• Reflection angle = light angle
– Both R & L have to lie on one plane
• R = – L + 2 (L · N) N
N

L R
light source

θL θR L N N

L N N
-L

34
Ideal Specular BRDF
• Light only reflects to the mirror direction
• A Dirac delta multiplied by a specular coefficient ks

• Not very useful for point lights, only for reflections

of other surfaces
– Why? You cannot really see a mirror reflection of an
infinitely small light!

35
Non-ideal Reflectors
• Real glossy materials usually deviate significantly
from ideal mirror reflectors
– Highlight is blurry
• They are not ideal diffuse surfaces either …

36
Non-ideal Reflectors
• Simple Empirical Reasoning for Glossy Materials
– We expect most of the reflected light to travel in the
direction of the ideal mirror ray.
– However, because of microscopic surface variations we
might expect some of the light to be reflected just slightly
offset from the ideal reflected ray.
– As we move farther and farther, in the angular sense, from
the reflected ray, we expect to see less light reflected.

37
The Phong Specular Model
• How much light is reflected?
– Depends on the angle  between the ideal reflection
direction r and the viewer direction v.
light source

r n
l
Camera  θ θ
v
Surface

38
The Phong Specular Model

• Parameters
– ks: specular reflection coefficient
light source
– q : specular reflection exponent
r n
l
Camera  θ θ
v
Surface

39
The Phong Model
• Effect of q – the specular reflection exponent

40
Terminology: Specular Lobe
• The specular reflection distribution
is usually called a “lobe”
– For Phong, its shape is

light source

Surface

41
The Complete Phong Model
• Sum of three components:
ideal diffuse reflection +
specular reflection +
“ambient”. light source

Surface

42
Ambient Illumination
• Represents the reflection of all indirect illumination.
• This is a total hack!
• Avoids the complexity of
indirect (“global”) illumination

43
Putting It All Together
• Phong Illumination Model

© Leonard McMillan. All rights reserved. This content is excluded from our Creative
Commons license. For more information, see http://ocw.mit.edu/help/faq-fair-use/.
44
Putting It All Together
• Phong Illumination Model

• Is it physically based?
– No, does not even conserve energy,
may well reflect more energy than what goes in
– Furthermore, it does not even conform to the BRDF model
directly (we are taking the proper cosine for diffuse, but
not for specular)
– And ambient was a total hack
45
Phong Examples
• The spheres illustrate
specular reflections as
the direction of the
light source and the
exponent q (amount of
shininess) is varied.

46
Fresnel Reflection
• Increasing specularity near grazing angles.
– Most BRDF models account for this.

Source: Lafortune et al. 97

© ACM. All rights reserved. This content is excluded from our Creative Commons
license. For more information, see http://ocw.mit.edu/help/faq-fair-use/.
47
Questions?

48
Blinn-Torrance Variation of Phong
• Uses the “halfway vector” h between l and v.

Light source
h n


Camera l

Surface

49
Lobe Comparison
• Half vector lobe
– Gradually narrower when approaching grazing
• Mirror lobe
– Always circular

Half vector lobe Mirror lobe

50
Half Vector Lobe is Better
• More consistent with what is observed in
measurements (Ngan, Matusik, Durand
2005)

Courtesy of Mitsubishi Electric Researh Laboratories, Inc. Used with permission.

Example: Plot of “PVC” BRDF at 55° incidence 51

Questions?

52
Microfacet Theory
• Example
– Think of water surface as lots of tiny mirrors (microfacets)
– “Bright” pixels are
• Microfacets aligned with the vector between sun and eye
• But not the ones in shadow
• And not the ones that are occluded

Image of sunset removed due to copyright restrictions.

53
Microfacet Theory
• Model surface by tiny mirrors
[Torrance & Sparrow 1967]

54
Microfacet Theory
• Value of BRDF at (L,V) is a product of
– number of mirrors oriented halfway between L and V

55
Microfacet Theory
• Value of BRDF at (L,V) is a product of
– number of mirrors oriented halfway between L and V

56
Microfacet Theory
• Value of BRDF at (L,V) is a product of
– number of mirrors oriented halfway between L and V

57
Microfacet Theory
• Value of BRDF at (L,V) is a product of
– number of mirrors oriented halfway between L and V
– ratio of the un(shadowed/masked) mirrors

58
Microfacet Theory
• Value of BRDF at (L,V) is a product of
– number of mirrors oriented halfway between L and V
– ratio of the un(shadowed/masked) mirrors
– Fresnel coefficient

59
Shadowing and Masking
• Some facets are hidden from viewpoint
• Some are hidden from the light

hidden from viewpoint hidden from viewpoint

60
Microfacet Theory-based Models
• Develop BRDF models by imposing simplifications
[Torrance-Sparrow 67], [Blinn 77], [Cook-Torrance
81], [Ashikhmin et al. 2000]

• Model the distribution p(H) of

microfacet normals
– Also, statistical models
for shadows and masking

spherical plot of a
Gaussian-like p(H)

61
Full Cook-Torrance Lobe
• ρs is the specular coefficient (3 numbers RGB)
• D is the microfacet distribution
– δ is the angle between the half vector H and the normal N
– m defines the roughness (width of lobe)
• G is the shadowing and masking term
• Need to add a diffuse term

62
Questions?
• “Designer BRDFs” by Ashikhmin et al.

© ACM. All rights reserved. This content is excluded from our Creative Commons
license. For more information, see http://ocw.mit.edu/help/faq-fair-use/.

63
BRDF Examples from Ngan et al.
Acquired data

Lighting

Material – Dark blue paint

Courtesy of Mitsubishi Electric Researh Laboratories, Inc. Used with permission.

64
Dark Blue Paint
Acquired data Blinn-Phong

Courtesy of Mitsubishi Electric Researh Laboratories, Inc. Used with permission.

Finding the BRDF model parameters that best reproduce the real material
Material – Dark blue paint
65
Dark Blue Paint
Acquired data Cook-Torrance

Courtesy of Mitsubishi Electric Researh Laboratories, Inc. Used with permission.

Finding the BRDF model parameters that best reproduce the real material
Material – Dark blue paint
66
Observations
• Some materials impossible to represent with a
single lobe
Acquired data Cook-Torrance

Material – Red Christmas Ball

Courtesy of Mitsubishi Electric Researh Laboratories, Inc. Used with permission.
67
Adding a Second Lobe
• Some materials impossible to represent with a
single lobe
Acquired data Cook-Torrance 2 lobes

Material – Red Christmas Ball

Courtesy of Mitsubishi Electric Researh Laboratories, Inc. Used with permission.
68
Image-Based Acquisition
• A Data-Driven Reflectance Model, SIGGRAPH
2003
– The data is available
http://people.csail.mit.edu/wojciech/BRDFDatabase/

© ACM. All rights reserved. This content is excluded from our Creative Commons
license. For more information, see http://ocw.mit.edu/help/faq-fair-use/.

69
 T. Weyrich et al., Fabricating
Questions? Microgeometry for Custom Surface
Reflectance, SIGGRAPH 2009

Images of Fig. 1 and Fig. 6 in Weyrich T. et al, "Fabricating Microgeometry for Custom Surface Reflectance."
SIGGRAPH '09 ACM SIGGRAPH 2009 papers; Article No. 32 --removed due to copyright restrictions.

70
Phong Normal Interpolation (Not Phong
Shading)

• Interpolate the average vertex normals across the

face and use this in shading computations
– Again, use barycentric interpolation!

71
That’s All for Today!

Images from the movie, “The Matrix,” removed due to copyright restrictions.

72
 Spatial Variation
• All materials seen so far are the same everywhere
– In other words, we are assuming the BRDF is independent
of the surface point x
– No real reason to make that assumption
– More next time

Courtesy of Fredo Durand. Used with permission.

© ACM. All rights reserved. This content is excluded © source unknown. All rights reserved. This
from our Creative Commons license. For more content is excluded from our Creative
information, see http://ocw.mit.edu/help/faq-fair-use/. Commons license. For more information,
see http://ocw.mit.edu/help/faq-fair-use/.

73
MIT OpenCourseWare
http://ocw.mit.edu

6.837 Computer Graphics

Fall 2012

For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.