TNM046: Datorgrafik Transformations

TNM046: Datorgrafik Transformations

Sasan Gooran

VT 2013

Computer Graphics

What?

“Almost everything on computers that is not text or sound “ The representation and manipulation of image data by

computer

Techniques to create and manipulate images

Where?

Television, Newspapers (weather reports), Medical investigations and surgical procedures

3

Computer Graphics

Applications:

Entertainments (Movies, Video games) Computer Aided Design (CAD)

Medical Applications Scientific Visualization Virtual Reality

….

Computer Graphics

5

Computer Graphics, TNM046

2D and 3D Transformations Triangle and Normal

Light Sources and Shading Perspective Projection

Culling and Clipping Texture Mapping

Linear Algebra

Transposition

x =

(

)

Scalar (dot) product:

x^t =

x₁ x₂ x₃

!

"

##

##

$

%

&

&

&&

y^t =

y₁ y₂ y₃

!

"

##

##

$

%

&

&

&&

y =

(

)

x • 

y =  x ∗ 

y^t = x₁y₁ + x₂y₂ + x₃y₃

Length (Norm):

Normalization:

x



7

Linear Algebra

x = 1, 2, 2

( )

(

)

Example: Find the angle between the two vectors below!

(answer: 1.68 radians or 96.38 degrees)

x • 

y = 

x ⋅ 

y ⋅ cos(θ⁾

x • 

y > 0 →θ < 90^ x • 

y = 0 →θ = 90^ x • 

y < 0 →θ > 90^

"

#$$

%

$$

Linear Algebra

Cross (vector) product:

x × 

y = ( x₂y₃ − x₃y₂ x₃y₁ − x₁y₃ x₁y₂ − x₂y₁ ) x × 

y =

ˆx ˆy ˆz

x₁ x₂ x₃ y₁ y₂ y₃

x y x × 

y

x y

 

x × 

y = 

x ⋅ 

y ⋅ sin(θ⁾ (

x × 

y)• 

x = ?

  

9

Basic Transformations (2D)

Translation

P₁

P₂ T

P₁ = x₁ y₁

!

"

##

$

%

&

& P₂ = x₂ y₂

!

"

##

$

%

&

&

T = ( t _x, t_y )

O

x₂ = x₁ + t_x y₂ = y₁ + t_y

!

"

#

$#

OP→₂ = →OP₁ + →T

x y

à

Basic Transformations (2D)

Translation

What is the translation vector, from A to B?

from B to A?

15 5

3 6

A

5 8

B

11

Basic Transformations (2D)

Rotation

P₁ = x₁ y₁

!

"

##

$

%

&

&

P₂ = x₂ y₂

!

"

##

$

%

&

&

ϕ

θ r

r

x₂ = r cos(ϕ +θ) = r cosϕ ^cosθ − r sinϕ ^sinθ y₂ = r sin(ϕ +θ) = r sinϕ ^cosθ + r cosϕ ^sinθ

"

#$

%$

x₁ = r cosϕ y₁ = r sinϕ

!

"

#

$#

x₂ = x₁ cosθ − y₁sinθ y₂ = x₁ sinθ + y₁ cosθ

"

#$

%$

Basic Transformations (2D)

Rotation

x₂ y₂

!

"

##

$

%

&

& = cosθ −sinθ

sinθ ^cosθ

!

"

# $

%& x₁ y₁

!

"

##

$

%

&

&

P₂ = R ∗ P₁

Counter-clockwise rotation by angle θ about the coordinate origin

à

13

Basic Transformations (2D)

Rotation

P₁ = x₁ y₁

!

"

##

$

%

&

&

P₂ = x₂ y₂

!

"

##

$

%

&

&

θ r

r

Rotation about an arbitrary point M

x₂ = x_M + (x₁ − x_M )cosθ − (y₁ − y_M )sinθ y₂ = y_M + (x₁ − x_M )sinθ + (y₁ − y_M )cosθ

"

#$

%$

M = x_M y_M

!

"

##

$

%

&

&

Basic Transformations (2D)

Scaling

Fixed point: Coordinate origin

x₂ = s_x ⋅ x₁ y₂ = s_y ⋅ y₁

"

#$

%$

P₂ = S ∗ P₁ S = s_x 0

0 s_y

!

"

##

$

%

&

&

à where

s_x and s_y > 0

Scaling values less than 1 reduces the size of the object Scaling values greater than 1 produce an enlargement

s_x=s_y: Uniform scaling, which maintains the relative object properties

15

Basic Transformations (2D)

Scaling

Arbitrary fixed point: (x_f, y_f) x₂ = x_f + s_x ⋅ (x₁ − x_f ) y₂ = y_f + s_y ⋅ (y₁ − y_f )

#

$%

&%

4 1

1 2

2 4

Object

Scaled sx=sy=2

Scaled sx=sy=0.5

fixed point: (0, 0)

4 1

1 2

2

Object Scaled sx=sy=2

Scaled sx=sy=0.5

fixed point: (2, 1) 3

Matrix representation

Homogenous coordinates (translation)

x₂ y₂

!

"

##

$

%

&

& = x₁ y₁

!

"

##

$

%

&

&+ t_x t_y

!

"

##

$

%

&

&

In order to represent this with a matrix multiplication we add a third coordinate, 1.

x₂ y₂ 1

!

"

##

##

$

%

&

&

&& =

1 0 t_x

0 1 t_y

0 0 1

!

"

##

##

$

%

&

&

&&

x₁ y₁ 1

!

"

##

##

$

%

&

&

&&

P₂ = T ∗ P₁ à

Matrix representation

Homogenous coordinates (translation)

T is the translation matrix. What is the inverse matrix?

T =

1 0 t_x

0 1 t_y

0 0 1

!

"

##

##

$

%

&

&

&&

T⁻¹ = ?

17

Matrix representation

Homogenous coordinates (Rotation)

x₂ y₂ 1

!

"

##

##

$

%

&

&

&& =

cosθ −sinθ ⁰

sinθ ^cosθ ⁰

0 0 1

!

"

##

#

$

%

&

&

&

x₁ y₁ 1

!

"

##

##

$

%

&

&

&&

P₂ = R ∗ P₁

R is the rotation matrix. What is the inverse matrix?

R⁻¹ = ?

Matrix representation

Homogenous coordinates (Scaling)

x₂ y₂ 1

!

"

##

##

$

%

&

&

&& =

s_x 0 0

0 s_y 0

0 0 1

!

"

##

##

$

%

&

&

&&

x₁ y₁ 1

!

"

##

##

$

%

&

&

&&

P₂ = S ∗ P₁

S is the scaling matrix. What is the inverse matrix?

S⁻¹ = ?

19

Composite Transformation

P₂ = T (t_{x 2}, t_y2)∗ T (t

{

}

{

}

If two successive translation vectors (t_x1, t_y1) and (t_x2, t_y2)

are applied to a coordinate position P₁, the final transformed location P₂ is calculated:

1 0 t_{x 2}

0 1 t_y2

0 0 1

!

"

##

##

$

%

&

&

&&

1 0 t_x1

0 1 t_y1

0 0 1

!

"

##

##

$

%

&

&

&& =

1 0 t_x1 + t_{x 2} 0 1 t_y1 + t_y2

0 0 1

!

"

##

##

$

%

&

&

&&

Composite Transformation

P₂ = R(θ₂)∗ R(

{

}

{

}

Two successive Rotations by θ₁ and θ₂:

s_{x 2} 0 0

0 s_y2 0

0 0 1

!

"

##

##

$

%

&

&

&&

s_x1 0 0

0 s_y1 0

0 0 1

!

"

##

##

$

%

&

&

&& =

s_x1 ⋅ s_{x 2} 0 0 0 s_y1 ⋅ s_y2 0

0 0 1

!

"

##

##

$

%

&

&

&&

Two successive Scaling-s:

21

General Pivot-Point Rotation

Rotation about an arbitrary point (x_M, y_M):

x₂ y₂ 1

!

"

##

##

$

%

&

&

&& =

1 0 x_M

0 1 y_M

0 0 1

!

"

##

##

$

%

&

&

&&

cosθ −sinθ ⁰

sinθ ^cosθ ⁰

0 0 1

!

"

##

#

$

%

&

&

&

1 0 −x_M

0 1 −y_M

0 0 1

!

"

##

##

$

%

&

&

&&

x₁ y₁ 1

!

"

##

##

$

%

&

&

&&

Step 1: Translate (x_M, y_M) to the coordinate origin (0,0) Step 2: Rotate about (0, 0)

Step 3: Translate back to (x_M, y_M)

x₂ y₂

!#

#

##

$&

&

&& =

cosθ −sinθ ^x_M (1− cosθ) + y_M sinθ sinθ ^cosθ ^y_M (1− cosθ) − x_M sinθ

!#

#

##

$&

&

&&

x₁ y₁

!#

#

##

$&

&

&&

General Fixed-Point Scaling

Scaling about an arbitrary point (x_f, y_f):

x₂ y₂ 1

!

"

##

##

$

%

&

&

&& =

1 0 x_f

0 1 y_f

0 0 1

!

"

##

##

$

%

&

&

&

&

s_x 0 0

0 s_y 0

0 0 1

!

"

##

##

$

%

&

&

&&

1 0 −x_f

0 1 −y_f

0 0 1

!

"

##

##

$

%

&

&

&

&

x₁ y₁ 1

!

"

##

##

$

%

&

&

&&

Step 1: Translate (x_f, y_f) to the coordinate origin (0,0) Step 2: Scale about (0, 0)

Step 3: Translate back to (x_f, y_f)

x₂ y₂ 1

!

"

##

##

$

%

&

&

&& =

s_x 0 x_f (1− s_x) 0 s_y y_f (1− s_y)

0 0 1

!

"

##

##

$

%

&

&

&

&

x₁ y₁ 1

!

"

##

##

$

%

&

&

&& 23

Ex: General Scaling Directions

We can scale an object in other directions by rotating the object to align the desired scaling directions

x y

S₁ S₂

θ Step 1: Rotate the object (by θ) to align

Step 2: Scale in the direction of (x, y)

Ex: General Scaling Directions

x₂ y₂ 1

!

"

##

##

$

%

&

&

&&=

cosθ ^sinθ ⁰

−sinθ ^cosθ ⁰

0 0 1

!

"

##

#

$

%

&

&

&

s₁ 0 0

0 s₂ 0

0 0 1

!

"

##

##

$

%

&

&

&&

cosθ −sinθ ⁰

sinθ ^cosθ ⁰

0 0 1

!

"

##

#

$

%

&

&

&

x₁ y₁ 1

!

"

##

##

$

%

&

&

&&

x₂ y₂ 1

!

"

##

##

$

%

&

&

&& =

s₁ cos²θ + s₂ sin²θ ^(s₂ − s₁)sinθ ^cosθ ⁰ (s₂ − s₁)sinθ ^cosθ ^s₁^sin2θ + s₂ cos²θ ⁰

0 0 1

!

"

##

##

$

%

&

&

&

&

x₁ y₁ 1

!

"

##

##

$

%

&

&

&&

What happens when s₁=s₂?

25

Reflection

1 0 0

0 −1 0

0 0 1

"

#

$$

$

%

&

'' '

Reflection matrix about y=0 (x-axis)

−1 0 0

0 1 0

0 0 1

"

#

$$

$

%

&

'' '

Reflection matrix about x=0 (y-axis)

−1 0 0

0 −1 0

0 0 1

"

$$

$

%' ''

Reflection relative to the coordinate origin

Ex: Reflection along y=x

Step 1: Rotate counter clockwise (45 degrees) (y=x) à y-axis

Step 2: Reflect bout y-axis Step 3: Rotate back

x₂ y₂ 1

!

"

##

##

$

%

&

&

&&=

cosθ ^sinθ ⁰

−sinθ ^cosθ ⁰

0 0 1

!

"

##

#

$

%

&

&

&

−1 0 0

0 1 0

0 0 1

!

"

##

#

$

%

&

&

&

cosθ −sinθ ⁰

sinθ ^cosθ ⁰

0 0 1

!

"

##

#

$

%

&

&

&

x₁ y₁ 1

!

"

##

##

$

%

&

&

&&

θ = π ^{/ 4} x₂

y₂ 1

!

"

##

##

$

%

&

&

&& =

0 1 0 1 0 0 0 0 1

!

"

##

#

$

%

&

&

&

x₁ y₁ 1

!

"

##

##

$

%

&

&

&& 27

Shear

x₂ y₂ 1

!

"

##

##

$

%

&

&

&& =

1 sh_x 0

0 1 0

0 0 1

!

"

##

##

$

%

&

&

&&

x₁ y₁ 1

!

"

##

##

$

%

&

&

&&

x₂ = x₁ + sh_x ⋅ y₁ y₂ = y₁

"

#$

%$

1 1

sh_x = 2

3D transformations, Translation

x₂ y₂ z₂ 1

!

"

##

##

#

$

%

&

&

&

&

&

=

1 0 0 t_x

0 1 0 t_y

0 0 1 t_z

0 0 0 1

!

"

##

##

#

$

%

&

&

&

&

&

x₁ y₁ z₁ 1

!

"

##

##

#

$

%

&

&

&

&

&

29

3D Rotation

x

P₂ =

x₂ = a y₂ z₂

!

"

##

##

$

%

&

&

&&

θ r r

y z

About x-axis

Counter clockwise

P₁ =

x₁ = a y₁ z₁

!

"

##

##

$

%

&

&

&&

When rotating around x-axis the x-coordinate is unchanged

3D Rotation

x₂ y₂ z₂ 1

!

"

##

##

#

$

%

&

&

&

&

&

=

1 0 0 0

0 cosθ −sinθ ⁰

0 sinθ ^cosθ ⁰

0 0 0 1

!

"

##

##

$

%

&

&

&

&

x₁ y₁ z₁ 1

!

"

##

##

#

$

%

&

&

&

&

&

Rotation about x-axis:

x₂ y₂ z₂ 1

!

"

##

##

#

$

%

&

&

&

&

&

=

cosθ ^{0 sin}θ ⁰

0 1 0 0

−sinθ ^{0 cos}θ ⁰

0 0 0 1

!

"

##

##

$

%

&

&

&

&

x₁ y₁ z₁ 1

!

"

##

##

#

$

%

&

&

&

&

&

Rotation about y-axis:

Why changing the signs in front of sinθ?

Rotation matrix about z-axis?

31

General 3D Rotation

Rotation about an axis parallel to one of the coordinate axes Step 1: Translate so that the rotation axis coincides with the parallel coordinate axis

Step 2: Perform the specified rotation about that axis Step 3: Translate back

R(α) = T⁻¹ ∗ R_co−axis(α^)*T

General 3D Rotation

Example: Counterclockwise rotation by 30 degrees about the axis passing through M1:(2, 1, 3) and M2:(2, 2, 3)

(vector from M1 to M2)

The vector going from M1 to M2 is (2 2 3)-(2 1 3)=(0 1 0), which is obviously parallel to y-axis

R(θ) =

1 0 0 2 0 1 0 1 0 0 1 3 0 0 0 1

!

"

##

##

$

%

&

&

&

&

cos(π

6 ) 0 sin(π

6 ) 0

0 1 0 0

−sin(π

6 ) 0 cos(π

6 ) 0

0 0 0 1

!

"

##

##

##

#

$

%

&

&

&

&

&

&

&

1 0 0 −2 0 1 0 −1 0 0 1 −3

0 0 0 1

!

"

##

##

$

%

&

&

&

&

Are there any other translations that work? ₃₃

General 3D Rotation

Rotation about an axis that is not parallel to any coordinate axes Step 1: Translate the object so that the rotation axis passes through the coordinate origin

Step 2: Rotate the object so that the axis of rotation coincides with one of the coordinate axes

Step 3: Perform the specified rotation about the coordinate axis Step 4: Rotate back

Step 5: Translate back

R(α) = T⁻¹ ∗ Raxis −>co−axis

−1 ∗ R_co−axis(α)∗ Raxis−>co−axis ∗T

General 3D Rotation

x

M₂ =

x_m2 y_m2 z_m2

!

"

##

##

$

%

&

&

&&

y z

M₁ =

x_m1 y_m1 z_m1

!

"

##

##

$

%

&

&

&

&

Eye

→U = →

M₁M₂ = (x_m2 − x_m1, y_m2 − y_m1, z_m2 − z_m1)

ˆu = U

U = (a, b, c) The vector is normalized

a² + b² + c² = ? ³⁵

General 3D Rotation

x

y z

After translation

(Step 1) ˆu = (a, b, c)

T =

1 0 0 −x_m1

0 1 0 −y_m1

0 0 1 −z_m1

0 0 0 1

"

#

$$

$$

$

%

&

'' '' '

General 3D Rotation

x

y z

ˆu = (a, b, c)

Two rotations are needed to make u coincide with one of the coordinate axes. (step 2)

Here we find the rotations to coincide it with z-axis

v = (a, b, 0) ϕ

The first rotation

is about the z-axis by φ to rotate u to the xz-plane (clockwise)

θ

The second rotation is about the y-axis by θ to align that with z-axis (clockwise)

37

General 3D Rotation

y z

ˆu = (a, b, c)

v = (a, b, 0) ϕ

The first rotation:

ˆx • 

v = ˆx ⋅ 

v cosϕ = a² + b² cosϕ ˆx • 

v = (1, 0, 0)⋅ (a, b, 0) = a ^à cosϕ = a d set d = a² + b²

ˆx × 

v = (1, 0, 0) × (a, b, 0) = (0, 0, b) ˆx × 

v = ˆx ⋅ 

v sinϕ = a² + b² sinϕ sinϕ = b

d

General 3D Rotation

The second rotation:

ˆz• ˆu = ˆz ⋅ ˆu cosθ = cosθ

ˆz• ˆu = (0, 0,1)⋅ (a, b, c) = c ^à cosθ = c

sinθ ^{= 1− cos2}θ ^{= 1− c2 = d}

x

y z

ˆu = (a, b, c) θ

Since 0 ≤θ ≤ π ^{then sin}θ ≥ 0

Observe: a² + b² + c² = 1 and d = a² + b²

39

General 3D Rotation

R(α) = T⁻¹ ∗ Raxis −>co−axis

−1 ∗ R_co−axis(α)∗ Raxis−>co−axis ∗T

Raxis−>co−axis =

cosθ 0 −sinθ ⁰

0 1 0 0

sinθ ^{0 cos}θ ⁰

0 0 0 1

"

#

$$

$$

%

&

'' ''

cosϕ ^sinϕ ^{0 0}

−sinϕ ^cosϕ ^{0 0}

0 0 1 0

0 0 0 1

"

#

$$

$$$

%

&

'' '''

Raxis−>co−axis =

c 0 −d 0

0 1 0 0

d 0 c 0

"

$$

$

%' ''

a / d b / d 0 0

−b / d a / d 0 0

0 0 1 0

"

$$

$

%' ''

Notice: Clockwise rotations Step 2

General 3D Rotation

R(α) = T⁻¹ ∗ Raxis −>co−axis

−1 ∗ R_co−axis(α)∗ Raxis−>co−axis ∗T

Raxis−>co−axis

−1 =

a / d −b / d 0 0 b / d a / d 0 0

0 0 1 0

0 0 0 1

"

#

$$

$$

%

&

'' ''

c 0 d 0

0 1 0 0

−d 0 c 0

0 0 0 1

"

#

$$

$$

%

&

'' ''

Why?

T⁻¹ =

1 0 0 x_m1

0 1 0 y_m1

0 0 1 z_m1

0 0 0 1

"

#

$$

$$

$

%

&

'' '' '

41

Step 4

Step 5

General 3D Rotation

R(α) = T⁻¹ ∗ Raxis −>co−axis

−1 ∗ R_co−axis(α)∗ Raxis−>co−axis ∗T

R_co−axis =

cos(α) −sin(α^{) 0 0}

sin(α^{) cos(}α^{) 0 0}

0 0 1 0

0 0 0 1

"

#

$$

$$$

%

&

'' '''

Notice that in the approach presented here we coincided the axis of rotation with z-axis. Therefore the coordinate axis is z-axis.

Step 3

Parallel Projection

Just to make it consistent with the assignments in the labs, here we rearranged the x, y and z-axes. The xy plane is the view plane and the z-axis points towards the viewer.

If the parallel projection on the view plane

(xy-plane or z=0) occurs along the z-direction, then:

z

x y

P₁ = x₁ y₁ z₁

!

"

##

##

$

%

&

&

&&

P₂ =

x₂ y₂ z₂

!

"

##

##

$

%

&

&

&&

x₂ = x₁ y₂ = y₁ z₂ = 0

!

"

#

$

##

x₂ y₂ z₂ 1

!

"

##

##

#

$

%

&

&

&

&

&

=

1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1

!

"

##

##

$

%

&

&

&

&

x₁ y₁ z₁ 1

!

"

##

##

#

$

%

&

&

&

&

&

43

Parallel Projection

z

x y

If the view plane is parallel to xy-plane, z=z_v.

If the parallel projection on the view plane (z=z_v) occurs along the z-direction, then:

P₁ = x₁ y₁ z

!#

###

$&

&

&&

x₂ = x₁ y₂ = y₁ z₂ = z_v

!

"

#

$#

#

x₂ y₂ z₂

!#

##

##

$&

&

&

&

&

=

1 0 0 0

0 1 0 0

0 0 0 z_v

!#

##

#

$&

&

&

&

x₁ y₁ z₁

!#

##

##

$&

&

&

&

&

P₂ =

x₂ y₂ z₂

!

"

##

##

$

%

&

&

&&

Coordinate System Transformation

z

x

y N

U V

The axes of the two coordinate systems are parallel.

O

O' =

x_o' y_o' z_o'

!

"

##

##

$

%

&

&

&

&

Assume that a point P has coordinates x

(

)

and coordinates u

(

)

World

View

T =

1 0 0 −x_o'

0 1 0 −y_o'

0 0 1 −z_o'

0 0 0 1

"

#

$$

$$

$

%

&

'' '' '

T transforms (translates) View system to World system.

45

Coordinate System Transformation

x₁ y₁ z₁ 1

!

"

##

##

#

$

%

&

&

&

&

&

=

1 0 0 x_o'

0 1 0 y_o'

0 0 1 z_o'

0 0 0 1

!

"

##

##

#

$

%

&

&

&

&

&

u₁ v₁ n₁ 1

!

"

##

##

#

$

%

&

&

&

&

&

u₁ v₁ n₁ 1

!

"

##

##

#

$

%

&

&

&

&

&

=

1 0 0 −x_o'

0 1 0 −y_o'

0 0 1 −z_o'

0 0 0 1

!

"

##

##

#

$

%

&

&

&

&

&

x₁ y₁ z₁ 1

!

"

##

##

#

$

%

&

&

&

&

&

In order not to get confused think of the coordinates of the origin of the View

system (O’), which is (0,0,0) in View, but

O' = x

(

)

in World system.

The axes of the two coordinate systems are parallel.

View World

Coordinate System Transformation

z

x

y N

U V

One of the axes are parallel (here z and N) but UV plane is rotated by θ in relation to xy palne.

O

O' = x( _o', y_o', z_o')^t World

View The view coordinate

system is transformed to World by a translation followed by a clockwise rotation about z-axis.

θ

47

Coordinate System Transformation

The z-axes are parallel but xy are rotated by θ.

x₁ y₁ z₁ 1

!

"

##

##

#

$

%

&

&

&

&

&

=

1 0 0 x_o'

0 1 0 y_o'

0 0 1 z_o'

0 0 0 1

!

"

##

##

#

$

%

&

&

&

&

&

cosθ −sinθ ^{0 0}

sinθ ^cosθ ^{0 0}

0 0 1 0

0 0 0 1

!

"

##

##

$

%

&

&

&

&

u₁ v₁ n₁ 1

!

"

##

##

#

$

%

&

&

&

&

&

u₁ v₁ n₁ 1

!

"

##

##

#

$

%

&

&

&

&

&

=

cosθ ^sinθ ^{0 0}

−sinθ ^cosθ ^{0 0}

0 0 1 0

0 0 0 1

!

"

##

##

$

%

&

&

&

&

1 0 0 −x_o'

0 1 0 −y_o'

0 0 1 −z_o'

0 0 0 1

!

"

##

##

#

$

%

&

&

&

&

&

x₁ y₁ z₁ 1

!

"

##

##

#

$

%

&

&

&

&

&

View World

Coordinate System Transformation

None of the axes are parallel

z

x

y N

U V

O

O' = x( _o', y_o', z_o')^t World

View First a translation is needed. Assume then that the unit vectors of the

coordinate axes of the View system can be expressed in World

coordinates.

ˆu = (u_x, u_y, u_z) ˆv = (v_x, v_y, v_z)

ˆn = (n_x, n_y, n_z)

49

Coordinate System Transformation

x₁ y₁ z₁ 1

!

"

##

##

#

$

%

&

&

&

&

&

=

1 0 0 x_o'

0 1 0 y_o'

0 0 1 z_o'

0 0 0 1

!

"

##

##

#

$

%

&

&

&

&

&

u_x v_x n_x 0 u_y v_y n_y 0 u_z v_z n_z 0

0 0 0 1

!

"

##

##

#

$

%

&

&

&

&

&

u₁ v₁ n₁ 1

!

"

##

##

#

$

%

&

&

&

&

&

u₁ v₁ n₁ 1

!

"

##

##

#

$

%

&

&

&

&

&

=

u_x u_y u_z 0 v_x v_y v_z 0 n_x n_y n_z 0

0 0 0 1

!

"

##

##

#

$

%

&

&

&

&

&

1 0 0 −x_o'

0 1 0 −y_o'

0 0 1 −z_o'

0 0 0 1

!

"

##

##

#

$

%

&

&

&

&

&

x₁ y₁ z₁ 1

!

"

##

##

#

$

%

&

&

&

&

&

View World

None of the axes are parallel

Coordinate System Transformation

Few points:

In order to better understand the transformations in previous page, think of a point that has the coordinate (0,0,1) in the View system, then you can see that it is first transformed to (n_x,n_y,n_z) and then translated.

What is the World coordinates of a point with coordinates (1,0,0) in the View system?

Make sure that you have understood these transformations by trying some obvious coordinates.

Notice that the inverse of a matrix whose rows (or columns) make an orthonormal base is equal to its transpose.

51

Coordinate System Transformation

x y

N V’

O

O' = x( _o', y_o', z_o')^t World

View

In practice you usually have the position of the camera (O’), the direction it is looking towards (N) and an up direction (V’) which is not necessarily perpendicular to N. If we now find an orthonormal base for the View system then we have the transformation matrix.

Coordinate System Transformation

z

x y

N V’

O World

ˆn = N

N = (n_x, n_y, n_z)

In some literature this might be defined as ˆn = − N N

Then U has to be perpendicular to the plane of N and V’:

ˆu =

V '× 

 N V '× 

N = (u_x, u_y, u_z)

U

V has to be perpendicular to both U and N.

ˆv = ˆn × ˆu = (v_x, v_y, v_z)

V

The transformations are now as shown on Page 50 53