Download - Microfacet Model - Computer graphicsPage 1 CS348B Lecture 13 Pat Hanrahan, Spring 2004 Reflection Models Last lecture Phong model Microfacet models Gaussian height field on surface

Transcript

Page 1

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Reflection Models

Last lecture

Phong model

Microfacet models

Gaussian height field on surface

Self-shadowing

Today

Torrance-Sparrow model

Anisotropic reflection models

Multiple importance sampling

Microfacet Model

Page 2

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Torrance-Sparrow Model

( )( )4cos cos

hr i r

i r

Df ωω ωθ θ

→ =

Niωrωhω

( )( ) ( ) cosss

h hD Dω α α ω= = = •N

iθ rθ

α

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Torrance-Sparrow Theory

( )( ) ( ) ( ) ( )4 cos cos

r i r

i i r

i r

fF S S D

ω ωθ θ θ α

θ θ

→′

=

Page 3

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Torrance-Sparrow Comparison

Aluminum

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Torrance-Sparrow Comparison

Magnesium Oxide

Page 4

CS348B Lecture 13 Pat Hanrahan, Spring 2004

“Diffuse” Reflection

ExperimentalPressed magnesium oxide powder used as an example of a diffuse reflectorReflection greater at high angles of incidence

TheoreticalBouguer - Microfacet distribution

No microfacet distribution can reflect rays equally in all directions

Multiple surface or subsurface reflections

Paint manufactures attempt to create ideal diffuse

Anisotropic Reflection Model

Page 5

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Anisotropic Reflection

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Quarterhorse

Page 6

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Reflection from a Cylinder

T

Ν

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Reflection from a Cylinder

T

Ν

Page 7

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Reflection from a Cylinder

T

L

ˆˆ( )NR L

N

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Reflection from a Cylinder

T

L

ˆˆ( )NR L

N

Page 8

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Reflection from a Cylinder

T

L

ˆˆ( )NR L

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Shape of Anisotropic Highlights

From Lu, Koenderink, Kappers

Fibers tangent to the plane defined by the halfway vector reflect light

Page 9

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Shape of Anisotropic Highlights

From Lu, Koenderink, Kappers

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Anisotropic Reflection

Page 10

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Kajiya-Kay Model

Diffuse

Specular

( )2ˆ ˆsin 1Lθ = − •T L

( ) ( )cos cos cos sin sin ssE L E L E Lθ θ θ θ θ θ− = +

L

E

T

ˆˆ( )NR L

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Herbert

Page 11

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Fiber Model

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Fiber Model

Page 12

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Fiber Model

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Caustics

Page 13

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Hair Appearance

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Hair Appearance

Page 14

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Page 15

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Multiple Importance Sampling

Page 16

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Multiple Importance Sampling

Reflection of a circular light source by a rough surface

Radius

Shin

ines

s

Sampling the light source Sampling the BRDF( ) ( )f x g x dx∫

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Multiple Importance Sampling

Two sampling techniques

Form weighted combination of samples

The balance heuristic

2, 2

2,2,

2 2,

~ ( )( )( )

i

ii

i

X p xf X

Yp X

=

1, 1

1,1,

1 1,

~ ( )( )( )

i

ii

i

X p xf X

Yp X

=

1 1, 2 2,i i iY w Y w Y= +

1 1 2 21 2

( )( ) ( ) ( ) ( ) ( ) ( )( ) ( )

ii

p xw x p x w x p x w x p xp x p x

= ⇒ = ++

Page 17

CS348B Lecture 13 Pat Hanrahan, Spring 2004

Multiple Importance Sampling

Source: Veach and Guibas