2IV60 Computer Graphics

Basic Math for CG

Jack van Wijk
TU/e
 Overview
• Coordinates, points, vectors
• Matrices

H&B A-1
 nD Space
nD Space: n (typically)
n : number of dimensions
Examples:
• 1D space: time, along a line or curve
• 2D space: plane, sphere
• 3D space: the world we live in
• 4D space: 3D + time
H&B A-2
 Coordinates
2D Cartesian coordinates:
x

y (x,y) (x,y)

x
Standard Screen (output, input)
H&B A-1
 Polar coordinates
Coordinate transformation
from (r , ) to ( x, y ) space :
y (x,y) x r cos
r y r sin 

Angle  : in radians
x

H&B A-1
 3D coordinates 1
3D Cartesian coordinates:

z (x,y,z) z (x,y,z)

y x
x y

Right-handed Left-handed
H&B A-1
 3D coordinates 2
3D Cartesian coordinates:

z (middle) (x,y,z) z (middle) (x,y,z)

x (thumb) y (index) x (thumb)

y (index)

Right-handed Left-handed
H&B A-1
 3D coordinates 3
Cylinder coordinates:

z (x,y,z) x r cos 
y r sin 
y
z z
x 

H&B A-1
 3D coordinates 4
Spherical coordinates: x r cos sin 
y r sin  sin 
z r cos 
z (x,y,z)

r  : azimuth or longitude
y  : elevation or latitude
x 

Also :   / 2  
 : colatitude H&B A-1
 Points
• Point: position in nD space
• Notation: P (H&B), also P, p, p en p
• (x,y,z) (H&B), also (x1, x2, x3), (Px, Py, Pz),
(r, , z), ( r, , ), …

H&B A-2
 Vectors 1
• Vector:
– “arrow”
– multiple interpretations (displacement, velocity,
force, …)
– has a magnitude and direction
– has no position
• Notation: V (H&B), also V, v, v en v
• (Vx, Vy, Vz) (H&B), also (x, y, z), (x1, x2, x3)
H&B A-2
 Vectors 2
V P2  P1
y
P2 ( x2  x1 , y2  y1 )
y2
(Vx , V y )
P1
y1
V: directed line segment, or
difference between two points

x1 x2 x
H&B A-2
 Vectors 3
Length of a vector:
V  Vx2  V y2 (2D : Pythagoras)

V  Vx2  V y2  Vz2 (3D)

H&B A-2
 Vectors 4
Direction of a vector: Direction angles.
Vx Vy Vz
cos   , cos   , cos   z
V V V 
Unit vector V : 
V y
V x 
|V|
Magnitude info is removed, direction is kept.
H&B A-2
 Vector addition
Add components, put vector head to tail

y W y V+W
W
V
V
x x

V  W (Vx  Wx ,V y  W y , Vz  Wz )
H&B A-2
 Scalar multiplication of vector
Multiplication components with scalar s

y y
2V
V

x x

sV ( sVx ,sV y , sVz )

H&B A-2
 Combining vector operations
Infinite line through P with direction V:
L(t) = P + Vt, t 

y V
P t

t : parameter along line

t [a, b]: line segment H&B A-2
 Vector multiplication 1
Scalar product or dot product:
Product of parallel components, gives 1 real value

V V W  V W cos

 VxWx  V yW y  VzWz

|W| cos 
|V|

H&B A-2
 Vector multiplication 2
Scalar product:

V W  V W cos
VxWx  V yW y  VzWz
Commutative : V W W V
Distributive : V (W  Z) V W  V Z

Associative : V (W Z) (V W ) Z ??

H&B A-2
 Vector multiplication 3
Vector product or cross product: gives a vector (in 3D)

n perpendicular to V and W

V W n V W sin
VW (V yWz  VzW y ,
W VzWx  VxWz ,


VxW y  V yWx )
n V

H&B A-2
 Vector multiplication 5

Scalar product:

Anticommutative : V W  W V
Not associative : V (W Z) (V W ) Z
Distributive : V (W  Z) V W  V Z

H&B A-2
 Vector multiplication 6
Scalar product: Vector product:
• scalar • vector
• Test if vectors are • Get a vector
perpendicular perpendicular to two
given vectors
• cos • sin
• project,… • surface area,…

H&B A-2
 Exercise 1
• Given a point P.
• Requested: Reflect a point Q with respect to P.

Q W=P–Q
W P Q’ = Q + 2W
y
W
Q’
= 2P – Q
or: = P + (P – Q )
x We don’t need coordinates!
 Exercise 1
• Given a point P.
• Requested: Reflect a point Q with respect to P.

Q Alternative
P P is halfway Q and Q’:
Q’
P = (Q + Q’)/2
2P = Q + Q’
Q’ = 2P – Q
 Exercise 2
• Given a line L: L(t) = P + Vt .
• Requested: Reflect a point Q with respect to L.

Q We know:
W Q’ = Q + 2 W
V L(t)
y P W W = L(t) – Q
t Q’
W. V = 0
x
 Exercise 2
We know: Substitute to get t:
L(t) = P + Vt (L(t) – Q).V = 0
Q’ = Q + 2 W (P + Vt – Q).V = 0
W = L(t) – Q V .V t + (P – Q).V = 0
W. V = 0 t = ((Q – P).V) / (V .V)
Then:
Q’ = Q + 2 (P + Vt – Q)
= 2P – Q + 2V((Q – P).V / V .V)
 Steps to be made
• Write down what you know
• Eliminate, substitute, etc. to get he result
• Check the result
– Does it make sense?
– Is there a simpler derivation?
 Exercise 3
• Given a triangle with vertices P, Q en R,
where the angle PQR is perpendicular.
• Requested: Rotate the triangle around the
line PQ over an angle What is the new
position R’ of R? P

Q R’


R
 Circle in space
y
C() C()
r 
B B
 A
P A x P

|A|=1, |B|=1, A.B = 0

C() = (r cos , r sin , 0)
= (0,0,0) + r cos  (1,0,0) + r sin  (0,1,0)
= P + r cos  A + r sin  B
 Exercise 3
A = (R – Q) / |R – Q|
B = (R – Q)(P – Q) / | (R – Q)(P – Q) |
R’ = Q + |R – Q| cos  A + |R – Q| sin  B

P
B
Q R’

A
R
 Use:
• Scalar product, cross product
• coordinate independent definitions
• vector algebra
Don’t use:
• arccos, arcsin
• y = f(x)
 Very short intro to Linear Algebra
System of linear equations:

Such systems occur in many, many applications.

They are studied in Linear Algebra.

u 2 x  3 y  4 z
v x  5 y  3z
w 5 x  y  z

H&B A-5
 Very short intro to Linear Algebra
System of linear equations:

Typical questions:
- Given u, v, w, what are x, y, z?
- Can we find a unique solution?

u 2 x  3 y  4 z
v x  5 y  3z
w 5 x  y  z

H&B A-5
 Very short intro to Linear Algebra
System of linear equations:

Crucial in computer graphics:

- Transforming geometric objects
- Change of coordinates

u 2 x  3 y  4 z
v x  5 y  3z
w 5 x  y  z

H&B A-5
 Example transformation
Transformation P  P ':
y
P : ( x, y ) P: (x,y)
P ' xU  yV , with x

U (3,2)
y’
V ( 2 ,3)
P’
In coordinates : V
x'  3x  2 y U
y x
y '  2 x  3 y x’

H&B A-5
 What is a matrix?
Matrix:
- Mathematical objects with operations

Matrix in computer graphics:

- Defines a coordinate frame
- Defines a transformation
- Handy tool for manipulating transformations

H&B A-5
 Matrix
Matrix: rectangular array of elements
Element: quantity (value, expression, function, …)
Examples:

 x
 3.60  0.01 2.1  e x
x
  5.46 0.00 1.6  ,  2 x 2
, a1 a2 a3  ,  y 
   e x 
 z 

H&B A-5
 Matrix
r  c matrix: r rows, c columns
mij : element at row i and column j.

 m11 m12  m1c 

m m  m 
M  21 22 2c 
   
 
 mr1 mr 2  mrc 
H&B A-5
 Matrix
r  c matrix: r rows, c columns
mij : element at row i and column j.

 m11 m12  m1c 

m m  m 
M  21 22 2c 
   
 
 mr1 mr 2  mrc  r  c
elements
H&B A-5
 Matrix
r  c matrix: r rows, c columns
mij : element at row i and column j.

 m11 m12  m1c 

m m  m 
M  21 22 2c 
   
 
 mr1 mr 2  mrc 
c columns H&B A-5
 Matrix
r  c matrix: r rows, c columns
mij : element at row i and column j.

 m11 m12  m1c 

m  
 21 m22 m2 c  r rows
M
   
 
 mr1 mr 2  mrc 
H&B A-5
 Matrix
Column vector: Row vector:
matrix with c =1 matrix with r =1
 v1  V v1 v2  vc 
v 
V 2
  Scalar:
  matrix with r=1and c=1
 vr 
V v 
Used for vectors
(and points)
H&B A-5
 Matrix
Matrix as collection of column vectors:

 ux vx wx 
M U V W   u y vy w y 
 u z vz wz  W

Matrix contains an axis-frame:

V U

H&B A-5
 Operations on matrices
• Multiplication with scalar
– simple
• Addition
– simple
• Matrix-matrix multiplication
– More difficult, but the most important

H&B A-5
 Scalar Matrix multiplication
Matrix M multiplied with scalar s:
 sm11 sm12  sm1c 
 sm sm2c 
 21 sm22 
A sM  , i.e., aij smij
    
 
 smr1 smr 2  smrc 
 3 x 12  x 4
example :   3 
 6 2 y 2 y 
H&B A-5
 Scalar Matrix addition
 a11 a12  a1c   b11 b12  b1c 
a   b b2c 
 21 a 22 a 2 c   21 b22 
M A  B  
       
   
a
 r1 a r2  a rc   br1 br 2  brc 
 a11  b11 a12  b12  a1c  b1c 
a b a  b  a  b 
 21 21 22 22 2c 2c 
, i.e., mij aij  bij
    
 
a 
 r 1 r1b a r2  b r2  a rc  brc 

 3 x 12   4  10  3 x  4 2 
example :     
 6 2 y  2 0   8 2 y 
H&B A-5
 Scalar Matrix addition
 a11 a12  a1c   b11 b12  b1c 
a   b b2c 
 21 a 22 a 2 c   21 b22 
M A  B  
       
   
 ar1 ar 2  arc   br1 br 2  brc 
 a11  b11 a12  b12  a1c  b1c 
a b a  b  a  b 
 21 21 22 22 2c 2c 
, i.e., mij aij  bij
    
 
 ar1  br1 ar 2  br 2  arc  brc 

• Just add elements pairwise

• A and B must have the same number of rows and columns
• Generalization of vector and scalar addition H&B A-5
 Matrix Matrix multiplication
 a11 a12  a1n   b11 b12  b1q 
a b 
a22  a2 n   21 b 22  b2q 
C AB  21
       
   
 amn am 2  amn   b p1 b p 2  b pq 
n
i = 2 ; k j = 1 ; k
cij a k 1
ik bkj

• cij: dot product of row vector ai and column vector bj

• #columns A must be the same as #rows B: n = p
• C: m  q matrix H&B A-5
 Matrix Matrix multiplication
Example:

 2 3  2 1  3 3 2 2  3 4  11 16
  1  4  1 2   1 1  4 3  1 2  4 4   13  16
   3 4    
 0 5    0 1  5 3 0 2  5 4  15 20
i=3 j=2 i=3, j=2

H&B A-5
 Matrix Matrix multiplication
Example:
 4
A 1 2 3 , B  5
 6

AB 

BA 

H&B A-5
 Matrix Matrix multiplication
Example:
 4
A 1 2 3 , B  5
 6
 4
AB 1 2 3  5 1 4  2 5  3 6 32
 6

BA 

H&B A-5
 Matrix Matrix multiplication
Example:
 4
A 1 2 3 , B  5 AB 
 6 BA
 4
AB 1 2 3  5 1 4  2 5  3 6 32
 6
 4  4 1 4 2 4 3  4 8 12
BA  5 1 2 3  5 1 5 2 5 3  5 10 15
 6  6 1 6 2 6 3  6 12 18
H&B A-5
 Matrix Matrix multiplication
• AB  BA
Matrix matrix multiplication is not commutative!
Order matters!

• A(B+C) = AB+AC
Matrix matrix multiplication is distributive

H&B A-5
 Example transformation sequence
(coordinate version)
X ' X : x'  px  qy, y ' rx  sy;
X ' '  X ': x' ' ax'by ' , y ' ' cx'dy '.

Then X ' '  X :

x' ' a ( px  qy )  b(rx  sy ), Unclear!

y ' ' c( px  qy )  d (rx  sy ); Error-prone!
or
x' ' (ap  br ) x  (aq  bs) y,
y ' ' (cp  dr ) x  (cq  ds) y.
H&B A-5
 Example transformation sequence
(matrix vector version)
 x'  p q   x 
X '  X :       ;
 y '  r s   y 
 x' '  a b   x'
X ' '  X ':       .
 y ' '  c d   y '

Then X ' '  X :

 x' '  a b   p q   x   ap  br aq  bs   x
 y ' '  c d   r s   y   cp  dr cq  ds   y
          
H&B A-5
 Example transformation sequence
(compact matrix vector version)
X '  X : X' AX ; Here some more detail :
X ' '  X ': X'' BX'  x  x'  x' '
X   , X' '   , X' '   ,
 y  y '  y ' '
Then X ' '  X :  p q  a b
A   ; B   .
 r s c d
X'' BAX

Matrix vector notation allows for compactness

and genericity!
H&B A-5
 Matrix Transpose
 a11 a12  a1c 
a  Transpose matrix:
a  a
M  21 22 2c 
Interchange rows and columns.
   
  r  c matrix M  c  r matrix
a
 r1 a r2  a rc 
MT
 a11 a21  ar1 
a a22  ar 2 
M T  12 , i.e., mTji mij
   
 
 a1c a2c  arc 

 1 4
T  x
 1 2 3  2 5 ,
examples :     [x y z ]T  y 
 4 5 6  H&B A-5
 3 6  z 
 Matrix Inverse
Simple algebra puzzle. Let ax = b. What is x?
ax b Multiply left and right with a  1
a  1ax a  1b Use a  1a equals 1
x a  1b Done

Linear algebra puzzle. Let MU = V. What is U?

MU V Multiply left and right with M  1
M  1MU M  1V Use M  1M equals I
U M  1V Done!

H&B A-5
 Matrix Inverse
Inverse of a n n (square) matrix M is denoted M  1 , with
MM  1 M  1M I,
where I is the identity matrix :
1 0  0
0 0
1 1 if i  j
I  , i.e., I ij  .
   0 if i  j
 
0 0  1
I is a diagonal matrix, with all 1' s on the diagonal.
M  1 " undoes" the effect of M,
multiplying with I has no effect.
H&B A-5
 Matrix Inverse examples
4 1 1 / 4
1
3 2  2  1
5   
 4   5 / 2 3 / 2 
3 2  2  1  3 2  2 5 / 2  3 1  2 3 / 2  1 0
5  
 4   5 / 2 3 / 2  5 2  4 5 / 2  5 1  4 3 / 2  0 1
   

H&B A-5
 Matrix Inverse examples
1
 a b
c d ?
 
1
 a b 1  d  b
c d  
  ad  bc   c a 

 a b   d  b   ad  bc  ab  ba   1 0
c d  c    (ad  bc)  
  a    ab  ba  cb  ad   0 1 
1
 2 1
 4 2 ? Does not exist, cannot be inverted.
  H&B A-5
 Matrix Inverse
- Does not always exist
- In general: if determinant = | M | = 0, matrix
cannot be inverted
- Inverse for n = 1 and n = 2: easy
- Inverse for higher n: use library function
- Important special case: orthonormal matrix.

H&B A-5
 Overview
• Coordinates, points, vectors
– definitions, operations, examples
• Matrices
– definitions, operations, examples