'

3D Transformations
Translation Rotation Scaling Composite transformations Reection and shear Coordinate system transformations OpenGL transformation functions

&
1

'

3D Transformations

All the 2D transformations can be extended to three dimensions Translation and Scaling are extended by adding a third value for the z-direction Rotation in 3D is more complicated Homogeneous coordinates for 3 dimensions require 4 components. Any sequence of transformations can be represented as a composite of the individual transformations.

&
2

'

3D Translation

Represent a 3D translation by a 3-tuple whose components are the shifts in the x, y and z directions. (tx , ty , tz )

x y z

= x + tx = y + ty = z + tz
z (x, y, x)

(x, y, z) T = (tx, ty, tz) x

&
3

'

In matrix form, this is x 1 y 0 = z 0 1 0 or P = TP 0 1 0 0 0 0 1 0 tx x ty y tz z 1 1

&
4

'

3D Rotation

Can rotate about any arbitrary line in space. Easiest to visualize are rotations about one of the coordinate axes 2D rotation is equivalent to a 3D rotation about the z-axis A positive rotation about a coordinate axis is counterclockwise when viewed from the positive direction.
y

&
5

'

Rotation about Coordinate Axis

For a rotation about the z-axis, the x and y coordinates change in the same way as for a 2D rotation and the z coordinate is unchanged. x y z x cos y sin = z 0 1 0 sin cos 0 0 0 0 1 0 0 x 0 y 0 z 1 1 = x cos y sin = x sin + y cos = z

Rotations about other coordinate axes have a similar form. &

'

Cyclic Permutations

We can generate the matrices by doing a cyclic permutation of the coordinate parameters. x y z x
y z x

z y y

&
7

'

Rotation about x and y Axes

For a rotation about the x-axis, we get y z x = y cos z sin = y sin + z cos = x

For a rotation about the y-axis, we get z x y = z cos x sin = z sin + x cos = y

As in 2D, the inverse transformation can be obtained by replacing by . As before, R1 = RT &

'

Rotation About Line Parallel to Axis

Do this with a composite transformation 1. Translate rotation axis to coordinate axis 2. Rotate 3. Reverse the translation The composite transformation matrix is R() = T1 Rx () T

&
9

'

Transformation Sequence

translate

rot

atio

inverse translation

&
10

'

Rotation about an Arbitrary Line

Use the same general approach but need both a translation and a rotation to align Transformation sequence: 1. Translate rotation axis to intersect the origin 2. Rotate axis of rotation to align with a coordinate axis 3. Do the desired rotation about the coordinate axis 4. Reverse the rotation in step 2 5. Reverse the translation in step 1

&
11

'

Transformation Sequence

P2 P1 P 1 P2 P 2

P 1

P2 P 2 P 1 P1

P 1 P2

&
12

'

Specifying a Rotation Axis

Rotation axis can be dened by two points

Use these points to generate a vector V = = P2 P1 (x2 x1 , y2 y1 , z2 z1 )

P2 u P1

Now construct a unit rotation axis vector

u = = where a = c =
z1 z2 |V |

V |V |

(a, b, c) b =
y1 y2 |V |

x1 x2 |V | ,

and

&
13

'

Translation to the Coordinate Origin

1 0 T = 0 0

0 1 0 0

0 x1 0 1 0

y1 z1 1

&
14

'

Aligning Rotation and Coordinate Axes

In general, need two coordinate axis rotations to accomplish this alignment. One of several possible sequences 1. Rotate rotation axis about x-axis into x-z plane 2. Rotate about y-axis to align with z-axis

u= (0, b, c)

u = (a,b,c)

u z = (0,0,1) u = (a,0,d)

u z = (0,0,1)

&
15

'

Rotation Angles

is the angle between the projection of u onto the y-z plane and the z-axis is the angle between the projection of u onto the x-z plane and the z-axis Use dot and cross products to determine the sin and cos terms needed for the rotation matrices. u v = |u| |v| cos u v = |u| |v| sin 1 0 Rx () = 0 0 0
c d b d

0
b d c d

0 0 1

d 0 a 0 0 1 0 0 Rx () = a 0 d 0 0 0 0 1

where a, b and c are as dened earlier and d = &

b2 + c2 %

'

3D Scaling

Similar to those in 2D A 3D scaling matrix sx 0 0 0 s 0 y 0 0 sz 0 0 0

(xf, yf, zf)

translate

0 0 1

(xf, yf, zf)

scale

inverse

Example: Scaling relative to a xed point

translation (xf, yf, zf) (xf, yf, zf)

&
17

'

3D Reection

Reection can be performed relative to a reection axis or relative to a plane Reection relative to a line is eectively a 180 deg rotation about that line Reection relative to a plane is like a rotation in 4D - it is like a conversion between left-handed and right-handed coordinate systems.
y reflection relative to x-y plane z y

x z

&
18

'

3D Shear

Shear transformations are used for modifying shapes and for perspective transformations Similar to shearing in 2D for x and y-axis shears; can also have z-axis shear. A z-axis 1 0 0 0 shearing matrix 0 shzx 1 0 0 shzy 1 0 shzx zref Example: shzx = shzy = 1 and zref = (0, 0, 0)

shzy zref 0 1

shear

&
19

'

Ane Transformation

Transformations of the form x = axx x + axy y + axz z y = z = ayx x + ayy y + ayz z azx x + azy y + azz z

Parallel lines remain parallel. Finite points map to nite points. Any ane transformation can be expressed in terms of translation, rotation, scaling, reection and shear.

&
20

'

Coordinate Transformations

Scene processing in computer graphics involves transformations between different coordinate systems. Conversions between world and device coordinates Objects are often modeled in their own coordinate system Objects that are created from many simpler objects need coordinate transformations to properly orient the components. Two steps to this transformation in 2D 1. Translate the origin of one system to that of the other 2. Rotate to align the x-axes

&
21

'

Specifying a 2D Coordinate System

There are several ways of specifying one coordinate system relative to another one With the position of the origin and the angle between the x and x axes With the position of the origin and a vector in the direction of the y axis With the coordinates of two points
y axis
y s axi

y axis
s axi y s axi x s axi

y axis
s axi y s axi x

V y0 P0 x0

P1 y0 P0 x0

y0 x0 x axis

x axis

x axis

&
22

'

3D Coordinate Transformations

In 3D, need the equivalent of two orthogonal vectors to specify a coordinate system. (The third one is determined by the cross product of the original two.)

uz=u uy ux

Can use a coordinate system whose z-axis is the rotation axis to derive the transformation matrix for rotation about an arbitrary axis. &
23