CAP4730: Computational
Structures in Computer Graphics
2D Transformations
2D Transformations
World Coordinates
Translate
Rotate
Scale
Viewport Transforms
Putting it all together
Transformations
Rigid Body Transformations - transformations that do
not change the object.
Translate
If you translate a rectangle, it is still a rectangle
Scale
If you scale a rectangle, it is still a rectangle
Rotate
If you rotate a rectangle, it is still a rectangle
Vertices
We have always represented vertices as
(x,y)
An alternate method is:
x
( x, y )
y
Example:
2.1
(2.1,4.8)
4.8
Matrix * Vector
x' a b x x' a b c x
y ' c d y y ' d e f y
z ' g h i z
x' ax by x' ax by cz
y ' dx ey fz
y ' cx dy
z ' gx hy iz
1 0 0
1 0
I I 0 1 0
0 1 0 0 1
Matrix * Matrix
a b x y Does A*B = B*A?
A ,B
c d z w What does the identity do?
ax bz ay bw
A* B
cx dz cy dw
a b 1 0
? 0 1
c d
Practice
2 3 2
A ,X
1 5 3
AX ?
2 3 0.5 1
A ,B
1 5 3 2 . 5
A* B ?
A* I ?
Translation
Translation - repositioning an object along a
straight-line path (the translation
distances) from one coordinate location to
another. (x,y)
(tx,ty)
(x,y)
Translation
P (3.7,4.1)
P ( x, y )
Given: T (7.1,8.2)
T (t x , t y )
x ' 3 .7 7 .1
x' x t x
y ' 4 . 1 8 .2
We want: y' y t y
x' 3.7 7.1
x' x t x y ' 4.1 8.2
y ' y t
Matrix form: y x ' 3 .4
P' P T y ' 4 .1
Translation Examples
P=(2,4), T=(-1,14), P=(?,?)
P=(8.6,-1), T=(0.4,-0.2), P=(?,?)
P=(0,0), T=(1,0), P=(?,?)
Which one is it?
(x,y)
(tx,ty)
(tx,ty)
(x,y) (x,y)
Recall
A point is a position specified with
coordinate values in some reference frame.
We usually label a point in this reference
point as the origin.
All points in the reference frame are given
with respect to the origin.
Applying to Triangles
(tx,ty)
What do we have here?
You know how to:
Scale
Scale - Alters the size of an object.
Scales about a fixed point
(x,y)
(x,y)
Scale
P ( x, y ) P (1.4,2.2)
Given:
S (sx , s y ) S (3,3)
x' s x x x ' 3 * 1. 4
We want: y' s y y y ' 3 * 2.2
x' 3 0 1.4
y ' 0 3 2.2
x' s x 0 x
y ' 0 s y y x' 4.2
Matrix form:
P' S P y ' 6. 6
Non-Uniform/Differential Scalin
(x,y)
(x,y)
S=(1,2)
Rotation
Rotation - repositions an object along a
circular path.
Rotation requires an and a pivot point
Rotation
P ( x, y ) x' r cos cos r sin sin
R ( ) y ' r cos sin r sin cos
x r cos x' x cos y sin
y r sin y ' x sin y cos
x' r cos( ) x' cos sin x
y ' r cos( ) y ' sin cos y
P' R P
Example
P=(4,4)
=45 degrees
What is the difference? Revisited
V(-0.6,0) V(0,-0.6) V(0.6,0.6) Translate (1.2,0.3)
Translate (1.2,0.3) V(0.6,0.3) V(1.2,-0.3) V(1.8,0.9)
V(0,0.6) V(0.3,0.9) V(0,1.2) V(0,0.6) V(0.3,0.9) V(0,1.2)
Rotations
V(-0.6,0) V(0,-0.6) V(0.6,0.6)
Rotate -30 degrees
V(0,0.6) V(0.3,0.9) V(0,1.2)
Combining Transformations
Q: How do we
specify each
transformation?
Specifying 2D Transformations
Translation
T(tx, ty)
Translation distances
Scale
S(sx,sy)
Scale factors
Rotation
R()
Rotation angle
Combining Transformations
Using translate, rotation, and scale, how do
we get:
Combining Transformations
Note there are two ways to combine rotation
and translation. Why?
Lets look at the equations
P ( x, y ) x" x t x cos y t y ' sin
T (t x , t y ) y" x t x sin y t y ' cos
R( ) P' R P
P' P T P" P 'T
P' ' R P' x' x cos y sin
x' x t x y ' x sin y cos
y' y t y x" x cos y sin t x
x" x' cos y ' sin y" x sin y cos t y
y" x' sin y ' cos
Combining them
We must do each step in turn. First we
rotate the points, then we translate, etc.
Since we can represent the transformations
by matrices, why dont we just combine
them?
P' P T
P' R P
P' S P
2x2 -> 3x3 Matrices
We can combine transformations by
expanding from 2x2 to 3x3 matrices.
1 0 t x
x t x
T t x , t y 0 1 t y
y t y 0 0 1
s x 0 0
sx 0
S s x , s y 0 s y 0
0 s y 0 0 1
cos sin 0
cos sin
R sin cos 0
sin cos
0 0 1
Homogenous Coordinates
We need to do something to the vertices
By increasing the dimensionality of the
problem we can transform the addition
component of Translation into
multiplication.
xh 6
x 3
xh h 4 6 2
x y h 4 3 14
P yh y Ex. 2, Ex. 14 7
y h h 2
1 7
2 2
2
h
h 2
Homogenous Coordinates
Homogenous Coordinates - term used in
mathematics to refer to the effect of this
representation on Cartesian equations. Converting
a pt(x,y) and f(x,y)=0 -> (xh,yh,h) then in
homogenous equations mean (v*xh,v*yh,v*h) can
be factored out.
What you should get: By expressing the
transformations with homogenous equations and
coordinates, all transformations can be expressed
as matrix multiplications.
Final Transformations -
Compare Equations
x ' 1 0 t x x
x ' t x x
T t x , t y y ' 0 1 t y y
y ' t y y 1 0 0 1 1
P T t x , t y P P T t x , t y P
x' s x 0 0 x
x' s x 0 x
S s x , s y y ' 0 sy 0 y
y ' 0 s y y 1 0 1 1
0
P S s x , s y P P S s x , s y P
x' cos sin 0 x
x' cos sin x
R y ' sin cos 0 y
y ' sin cos y 1 0 1 1
0
P R P P R P
Combining Transformations
P' A P, P" B P ' P" A B P
P(3,4), T (0.4641,2), R(60)
x' 1 0 t x x x" cos sin t x cos t y sin x
y ' 0 1 t y y" sin cos t sin t cos y
y
x y
1 0 0 1 1 1 0 0 1 1
x" cos sin 0 x ' x" 1 0 t x cos sin 0 x
y" sin cos 0 y ' y" 0 1 t sin cos 0 y
y
1 0 0 1 1 1 0 0 1 0 0 1 1
x" cos sin 0 1 0 t x x' x" cos sin t x x
y" sin cos 0 0 1 t y y ' y" sin cos t y
y
1 0 0 1 0 0 1 1 1 0 0 1 1
How would we get:
How would we get:
Coordinate Systems
Object Coordinates
World Coordinates
Eye Coordinates
Object Coordinates
World Coordinates
Screen Coordinates
Coordinate Hierarchy
Screen Coordinates
Transformation
World->Screen
World Coordinates
Transformation Transformation Transformation
Object #1 -> Object #2 -> Object #3 ->
World World World
Object #1 Object #2 Object #3
Object Coordinates Object Coordinates Object Coordinates
Lets reexamine assignment 2b
Transformation Hierarchies
For example:
Transformation Hierarchies
Lets examine:
Transformation Hierarchies
What is a better way?
Transformation Hierarchies
What is a better way?
Transformation Hierarchies World Coordinates
Transformation
Green -> World
We can have transformations be
Green
in relation to each other Object's Coordinates
Transformation
Red -> Green
Red
Object Coordinates
Transformation
Blue -> Red
Blue
Object Coordinates
More Complex Models