Graphics

2D Geometric
Transformations

고려대학교 컴퓨터 그래픽스 연구실

cgvr.korea.ac.kr Graphics Lab @ Korea

Contents
CGVR

 Definition & Motivation

 2D Geometric Transformation
 Translation
 Rotation
 Scaling
 Matrix Representation
 Homogeneous Coordinates
 Matrix Composition
 Composite Transformations
 Pivot-Point Rotation
 General Fixed-Point Scaling
 Reflection and Shearing
 Transformations Between Coordinate Systems

cgvr.korea.ac.kr Graphics Lab @ Korea

Geometric Transformation
CGVR

 Definition
 물체의 좌표를 바꾸는 것
 Translation, Rotation, Scaling

 Motivation – Why do we need geometric

transformations in CG?
 As a viewing aid
 As a modeling tool
 As an image manipulation tool

cgvr.korea.ac.kr Graphics Lab @ Korea

Example: 2D Geometric
Transformation CGVR

Modeling
Coordinates

World Coordinates

cgvr.korea.ac.kr Graphics Lab @ Korea

Example: 2D Scaling
CGVR

Modeling
Coordinates

Scale(0.3, 0.3)

World Coordinates

cgvr.korea.ac.kr Graphics Lab @ Korea

Example: 2D Rotation
CGVR

Modeling
Coordinates

Scale(0.3, 0.3)
Rotate(-90)

World Coordinates

cgvr.korea.ac.kr Graphics Lab @ Korea

Example: 2D Translation
CGVR

Modeling
Coordinates

Scale(0.3, 0.3)
Rotate(-90)
Translate(5, 3)

World Coordinates

cgvr.korea.ac.kr Graphics Lab @ Korea

Example: 2D Geometric
Transformation CGVR

Modeling
Coordinates
Again?

World Coordinates

cgvr.korea.ac.kr Graphics Lab @ Korea

Example: 2D Geometric
Transformation CGVR

Modeling
Coordinates
Scale
Translate

Scale
Rotate
Translate
World Coordinates

cgvr.korea.ac.kr Graphics Lab @ Korea

Basic 2D Transformations
CGVR

 Translation
 x  x  tx
 y y  ty
 Scale
 x  x sx
 y y sy
 Rotation
 x  x cosθ - y sinθ
 y y sinθ  y cosθ
 Shear
 x  x  hx y
 y y  hy x

cgvr.korea.ac.kr Graphics Lab @ Korea

Basic 2D Transformations
CGVR

 Translation
 x  x  tx
 y y  ty
 Scale
 x  x sx
 y y sy
 Rotation
 x  x cosθ - y sinθ Transformations
 y y sinθ  y cosθ
can be combined
(with simple algebra)
 Shear
 x  x  hx y
 y y  hy x

cgvr.korea.ac.kr Graphics Lab @ Korea

Basic 2D Transformations
CGVR

 Translation
 x  x  tx
 y y  ty
 Scale
 x  x sx
 y y sy
 Rotation
 x  x cosθ - y sinθ
 y y sinθ  y cosθ
 Shear x  x sx
 x  x  hx y y  y sy
 y y  hy x

cgvr.korea.ac.kr Graphics Lab @ Korea

Basic 2D Transformations
CGVR

 Translation
 x  x  tx
 y y  ty
 Scale
 x  x sx
 y y sy
 Rotation
 x  x cosθ - y sinθ
 y y sinθ  y cosθ
 Shear x  ((x sx) cos  (y sy) sin )
 x  x  hx y y ((x sx) sin  (y sy) cos )
 y y  hy x

cgvr.korea.ac.kr Graphics Lab @ Korea

Basic 2D Transformations
CGVR

 Translation
 x  x  tx
 y y  ty
 Scale
 x  x sx
 y y sy
 Rotation
 x  x cosθ - y sinθ
 y y sinθ  y cosθ
 Shear x  ((x sx) cos  (y sy) sin )  tx
 x  x  hx y y ((x sx) sin  (y sy) cos )  ty
 y y  hy x

cgvr.korea.ac.kr Graphics Lab @ Korea

Basic 2D Transformations
CGVR

 Translation
 x  x  tx
 y y  ty
 Scale
 x  x sx
 y y sy
 Rotation
 x  x cosθ - y sinθ
 y y sinθ  y cosθ
 Shear x  ((x sx) cos  (y sy) sin )  tx
 x  x  hx y y ((x sx) sin  (y sy) cos )  ty
 y y  hy x

cgvr.korea.ac.kr Graphics Lab @ Korea

Matrix Representation
CGVR

 Represent a 2D Transformation by a Matrix

 a b
c d
 
 Apply the Transformation to a Point
x ax  by  x  a b   x 
y cx  dy  y  c d   y 
    
Transformation
Point
Matrix

cgvr.korea.ac.kr Graphics Lab @ Korea

Matrix Representation
CGVR

 Transformations can be combined by matrix

multiplication

 x  a b   e f  i j  x
 y  c d   g     
h  k l   y 
   
Transformation
Matrix
Matrices are a convenient and efficient way
to represent a sequence of transformations

cgvr.korea.ac.kr Graphics Lab @ Korea

2×2 Matrices
CGVR

 What types of transformations can be

represented with a 2×2 matrix?
2D Identity
x  x  x 1 0  x 
 y  0 1  y 
y  y     
2D Scaling
x sx x  x  sx 0   x 
y sy y  y  0 sy   y 
    

cgvr.korea.ac.kr Graphics Lab @ Korea

2×2 Matrices
CGVR

 What types of transformations can be

represented with a 2×2 matrix?
2D Rotation
x cos x  sin  y  x  cos  sin    x 
 y  sin  cos   y 
y sin  x  cos y     
2D Shearing
x  x  shx y  x 1 shx   x 
y shy x  y  y  shy 1  y 
    

cgvr.korea.ac.kr Graphics Lab @ Korea

2×2 Matrices
CGVR

 What types of transformations can be

represented with a 2×2 matrix?
2D Mirror over Y axis
x  x  x   1 0  x 
     y
y  y y 
   0 1  
2D Mirror over (0,0)
x  x  x   1 0  x 
     y
y  y y 
   0  1  

cgvr.korea.ac.kr Graphics Lab @ Korea

2×2 Matrices
CGVR

 What types of transformations can be

represented with a 2×2 matrix?
2D Translation
x x  tx
y y  ty NO!!

Only linear 2D transformations

can be Represented with 2x2 matrix

cgvr.korea.ac.kr Graphics Lab @ Korea

2D Translation
CGVR

 2D translation can be represented by a 3×3

matrix
 Point represented with homogeneous coordinates

 x 1 0 tx   x 
x  x  tx  y  0 1 ty   y 
y y  ty     
1   0 0 1  1 

cgvr.korea.ac.kr Graphics Lab @ Korea

Basic 2D Transformations
CGVR

 Basic 2D transformations as 3x3 Matrices

 x 1 0 tx   x   x  sx 0 0   x 
 y  0 1 ty   y   y  0 sy 0  y 
         
1   0 0 1  1  1   0 0 1  1 
Translate Scale
 x  cos  sin  0   x   x 1 shx 0   x 
 y  sin  cos 0   y  y  shy 1 0  y 
         
1   0 0 1  1  1   0 0 1 1 

Rotate Shear

cgvr.korea.ac.kr Graphics Lab @ Korea

Homogeneous Coordinates
CGVR

 Add a 3rd coordinate to every 2D point

 (x, y, w) represents a point at location (x/w, y/w)
 (x, y, 0) represents a point at infinity
 (0, 0, 0) Is not allowed
y
2 (2, 1, 1) or (4, 2, 2) or (6, 3, 3)
1
x
1 2

Convenient Coordinate System to

Represent Many Useful Transformations

cgvr.korea.ac.kr Graphics Lab @ Korea

Linear Transformations
CGVR

 Linear transformations are combinations of …

 Scale
 x   a b 0   x 
 Rotation  y    c d 0   y 
 Shear, and     
 Mirror  w  0 0 1   w
 Properties of linear transformations
 Satisfies: T ( s p  s p ) s T ( p )  s T ( p )
1 1 2 2 1 1 2 2
 Origin maps to origin
 Lines map to lines
 Parallel lines remain parallel
 Ratios are preserved
 Closed under composition

cgvr.korea.ac.kr Graphics Lab @ Korea

Affine Transformations
CGVR

 Affine transformations are combinations of

 Linear transformations, and
 Translations x
   a b
c x
 y    d e f   y 
    
 w  0 0 1   w
 Properties of affine transformations
 Origin does not map to origin
 Lines map to lines
 Parallel lines remain parallel
 Ratios are preserved
 Closed under composition

cgvr.korea.ac.kr Graphics Lab @ Korea

Projective Transformations
CGVR

 Projective transformations…
 Affine transformations, and
 Projective warps x
   a
b c x
 y    d e f   y 
    
 w  g h i   w
 Properties of projective transformations
 Origin does not map to origin
 Lines map to lines
 Parallel lines do not necessarily remain parallel
 Ratios are not preserved
 Closed under composition

cgvr.korea.ac.kr Graphics Lab @ Korea

Matrix Composition
CGVR

 Transformations can be combined by matrix

multiplication
 x   1 0 tx   cos θ - sinθ 0   sx 0 0   x 
 y   0 1 ty   sinθ cosθ 0  0 sy 0    y 
       
 w   0 0 1   0 0 1   0 0 1    w 
p  T(tx, ty) R( ) S(sx, sy) p
 Efficiency with premultiplication
 Matrix multiplication is associative
p (T (R (S p))) p (T R S) p

cgvr.korea.ac.kr Graphics Lab @ Korea

Matrix Composition
CGVR

 Rotate by  around arbitrary point (a,b)

 M T(a, b) R(θ ) T(-a,-b)
(a,b)

 Scale by sx, sy around arbitrary point (a,b)


M T(a, b) S(sx, sy) T(-a,-b)

(a,b)

cgvr.korea.ac.kr Graphics Lab @ Korea

 Pivot-Point Rotation
CGVR

(xr,y (xr,y (xr,y (xr,y

r) r) r) r)

Translate Rotate Translate

T xr , yr R T  xr , yr  Rxr , yr , 

 1 0 x r   cos   sin  0  1 0  x r   cos   sin  x r (1  cos  )  y r sin  
 0 1 y   sin  cos  0  0 1  y   sin  cos  y r (1  cos  )  x r sin  
 r    r

 0 0 1   0 0 1  0 0 1   0 0 1 

cgvr.korea.ac.kr Graphics Lab @ Korea

General Fixed-Point Scaling
CGVR

(xf,y (xf,y (xf,y (xf,y

f) f) f) f)

Translate Scale Translate

T x f , y f S s x , s y T  x f , y f  S x f , y f , s x , s y 
1 0 xf   sx 0 0 1 0  x f   sx 0 x f (1  s x ) 
0 1 yf   0 sy 0  0 1  y f   0 sy y f (1  s y )
 
 0 0 1   0 0 1  0 0 1   0 0 1 

cgvr.korea.ac.kr Graphics Lab @ Korea

Reflection
CGVR

 Reflection with respect to the axis

• x 축에 대한 반사 • y 축에 대한 반사 • xy 축 ( 원점 )
에 대한
 1 0 0   1 0 0   1 0 0
 0  1 0  0 1 0  0  1 0
     
 0 0 1  0 0 1  0 0 1

y 1 y y
1 1’ 1’

2 3 2 3 3’ 2 3’ 2
x x 3 x
2’ 3’ 1

1’ 2

x 축에 대한 반사 y 축에 대한 반사 원점에 대한 반사

cgvr.korea.ac.kr Graphics Lab @ Korea

Reflection
CGVR

 Reflection with respect to a Line

y
 0 1 0
 1 0 0
  x
 0 0 1
y=x 에 대한 반사


 Clockwise rotation of 45  Reflection about the x
axis  Counterclockwise rotation of 45 
y y 1
y
2 3
x 2 3 x
x
’ ’
1
’

cgvr.korea.ac.kr Graphics Lab @ Korea

Shear
CGVR

 Converted to a parallelogram
y (1,1) y
(0,1)
 1 sh x 0 (2,1) (3,1)
 0 1 0
  (0,0) (1,0) x
(0,0) (1,0)
x
 0 0 1
x’ = x + shx · y, y’ = y
x 축으로 밀림 (Shx=2)
 Transformed to a shifted parallelogram
(Y = Yref)
y y
(1,1) (1,1) (2,1)
 1 sh x  sh x y ref  (0,1)
0 1  (1/2,0)
0 x
 
(0,0) (1,0) (3/2,0)x
 0 0 1 
(0,-1)

선분에 대한 밀림
x’ = x + shx · (y-yref), y’ = y
(Shx=1/2, yref=-1)

cgvr.korea.ac.kr Graphics Lab @ Korea

Shear
CGVR

 Transformed to a shifted parallelogram

(X = Xref)

(0,3/2) (1,2)
 1 0 0  y y
 sh 1  sh y x ref  (1,1)
 y  (0,1)
 0 0 1  (0,1/2) (1,1)
x x
(0,0) (1,0) (-1,0)

x’ = x, y’ = shy · (x-xref) + y
선분에 대한 밀림
(Shy=1/2, xref=-1)

cgvr.korea.ac.kr Graphics Lab @ Korea