University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell
Affine Transformations
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 2
Logistics
Required reading:Watt, Section 1.1.
Further reading:Foley, et al, Chapter 5.1-5.5.David F. Rogers and J. Alan Adams, MathematicalElements for Computer Graphics, 2nd Ed., McGraw-Hill, New York, 1990, Chapter 2.
Logistics:HW #1 handed out MondayProject #1 due on Monday, artifact on followingMonday.
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 3
Geometric transformations
Geometric transformations will map points in onespace to points in another: (x',y',z') = f(x,y,z).These transformations can be very simple, such asscaling each coordinate, or complex, such as non-linear twists and bends.We'll focus on transformations that can berepresented easily with matrix operations.We'll start in 2D...
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 4
Representation
We can represent a point, p = (x,y), in the plane
as a column vector
as a row vector!
x
y
"
# $ %
& '
!
x y[ ]
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 5
Representation, cont.
We can represent a 2-D transformation M by a matrix
If p is a column vector, M goes on the left:
If p is a row vector, MT goes on the right:
We will use column vectors.
!
" p = pMT
" x " y [ ] = x y[ ]a c
b d
#
$ %
&
' ( !
" p = Mp
" x
" y
#
$ % &
' ( =
a b
c d
#
$ %
&
' (
x
y
#
$ % &
' ( !
M =a b
c d
"
# $
%
& '
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 6
Two-dimensional transformations
Here's all you get with a 2 x 2 transformationmatrix M:
So:
We will develop some intimacy with theelements a, b, c, d…
!
" x
" y
#
$ % &
' ( =
a b
c d
#
$ %
&
' (
x
y
#
$ % &
' (
!
" x = ax + by
" y = cx + dy
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 7
Identity
Suppose we choose a=d=1, b=c=0:Gives the identity matrix:
Doesn't move the points at all!
1 0
0 1
"
# $
%
& '
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 8
Scaling
Suppose b=c=0, but let a and d take on any positive value:Gives a scaling matrix:
Provides differential (non-uniform) scaling in x and y:
!
a 0
0 d
"
# $
%
& '
!
" x = ax
" y = dy
!
2 0
0 2
"
# $
%
& '
!
1 2 0
0 2
"
# $
%
& '
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 9
Reflection
Suppose b=c=0, but let either a or d go negative.Examples:
!
"1 0
0 1
#
$ %
&
' (
!
1 0
0 "1
#
$ %
&
' (
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 10
Shear
Now leave a=d=1 and experiment with bThe matrix
gives:!
1 b
0 1
"
# $
%
& '
!
" x = x + by
" y = y
!
1 1
0 1
"
# $
%
& '
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 11
Effect on unit square
Let's see how a general 2 x 2 transformationM affects the unit square:
!
a b
c d
"
# $
%
& ' p q r s[ ] = ( p ( q ( r ( s [ ]
a b
c d
"
# $
%
& ' 0 1 1 0
0 0 1 1
"
# $
%
& ' =
0 a a + b b
0 c c + d d
"
# $
%
& '
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 12
Effect on unit square, cont.
Observe:Origin invariant under MM can be determined just by knowing how thecorners (1,0) and (0,1) are mappeda and d give x- and y-scalingb and c give x- and y-shearing
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 13
Rotation
From our observations of the effect on the unit square, itshould be easy to write down a matrix for “rotation aboutthe origin”:
Thus !
1
0
"
# $ %
& ' (
cos())
sin())
"
# $
%
& '
0
1
"
# $ %
& ' (
*sin())
cos())
"
# $
%
& '
!
MR
= R(") =cos(") #sin(")
sin(") cos(")
$
% &
'
( )
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 14
Linear transformationsThe unit square observations also tell us the 2x2 matrix transformationimplies that we are representing a point in a new coordinate system:
where u=[a c]T and v=[b d]T are vectors that define a new basis for alinear space.The transformation to this new basis (a.k.a., change of basis) is alinear transformation.!
" p = Mp
=a b
c d
#
$ %
&
' ( x
y
#
$ % &
' (
= u v[ ]x
y
#
$ % &
' (
= x )u + y ) v
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 15
Limitations of the 2 x 2 matrix
A 2 x 2 linear transformation matrix allowsScalingRotationReflectionShearing
Q: What important operation does thatleave out?
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 16
Affine transformations
In order to incorporate the idea that both the basis and theorigin can change, we augment the linear space u, v withan origin t.Note that while u and v are basis vectors, the origin t is apoint.We call u, v, and t (basis and origin) a frame for an affinespace.Then, we can represent a change of frame as:
This change of frame is also known as an affinetransformation.How do we write an affine transformation with matrices?!
" p = x #u + y # v + t
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 17
Homogeneous CoordinatesTo represent transformations among affine frames, we can loft theproblem up into 3-space, adding a third component to every point:
Note that [a c 0]T and [b d 0]T represent vectors and[tx ty 1]T, [x y 1]T and [x' y' 1]T represent points.
!
" p = Mp
=
a b tx
c d ty
0 0 1
#
$
% % %
&
'
( ( (
x
y
1
#
$
% % %
&
'
( ( (
= u v t[ ]
x
y
1
#
$
% % %
&
'
( ( (
= x )u + y ) v +1) t
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 18
Homogeneous coordinatesThis allows us to perform translation as well as the linear
transformations as a matrix operation:
!
" p = MTp
" x
" y
1
#
$
% % %
&
'
( ( (
=
1 0 tx
0 1 ty
0 0 1
#
$
% % %
&
'
( ( (
x
y
1
#
$
% % %
&
'
( ( (
" x = x + tx
" y = y + ty
!
1 0 1
0 1 1 2
0 0 1
"
#
$ $ $
%
&
' ' '
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 19
Rotation about arbitrary points
1. Translate q to origin2. Rotate3. Translate backLine up the matrices for these step in right to left order and multiply.
Note: Transformation order is important!!
Until now, we have only considered rotation about the origin.
With homogeneous coordinates, you can specify a rotation, Rq,about any point q = [qx qy 1]T with a matrix:
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 20
Points and vectorsFrom now on, we can represent points as have an additional coordinate of w=1.
Vectors have an additional coordinate of w=0. Thus, a change of origin has noeffect on vectors.
Q: What happens if we multiply a matrix by a vector?
These representations reflect some of the rules of affine operations on points andvectors:
One useful combination of affine operations is:
Q: What does this describe?
!
vector + vector "
vector # vector "
point $point "
point + vector "
point + point "
!
p(t) = p0 + tv
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 21
Barycentric coordinatesA set of points can be used to create an affine frame. Consider atriangle ABC and a point p:
We can form a frame with an origin C and the vectors from C to theother vertices:
We can then write P in this coordinate frame
The coordinates (α, β, γ) are called the barycentric coordinates ofp relative to A, B, and C.
A
BC
p
!
•
!
p ="u+ #v + t
!
u = A"C v = B"C t = C
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 22
Computing barycentric coordinatesFor the triangle example we can compute the barycentric
coordinates of P:
Cramer’s rule gives the solution:
Computing the determinant of the denominator gives:
!
"A + #B + $C =
Ax Bx Cx
Ay By Cy
1 1 1
%
&
' ' '
(
)
* * *
"
#
$
%
&
' ' '
(
)
* * *
=
px
py
1
%
&
' ' '
(
)
* * *
!
BxCy " ByCx + AyCx " AxCy + AxBy " AyBx!
" =
px Bx Cx
py By Cy
1 1 1
Ax Bx Cx
Ay By Cy
1 1 1
# =
Ax px Cx
Ay py Cy
1 1 1
Ax Bx Cx
Ay By Cy
1 1 1
$ =
Ax Bx px
Ay By py
1 1 1
Ax Bx Cx
Ay By Cy
1 1 1
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 23
Cross productsConsider the cross-product of two vectors, u and v. What is
the geometric interpretation of this cross-product?A cross-product can be computed as:
What happens when u and v lie in the x-y plane? What is thearea of the triangle they span?
!
u" v =
i j k
ux uy uz
vx vy vz
= (uyvz # uzvy )i + (uzvx # uxvz)j+ (uxvy # uyvx )k
=
uyvz # uzvyuzvx # uxvzuxvy # uyvx
$
%
& & &
'
(
) ) )
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 24
Barycentric coords from area ratiosNow, let’s rearrange the equation from two slides ago:
The determinant is then just the z-component of(B-A) × (C-A), which is two times the area of triangle ABC!Thus, we find:
Where SArea(RST) is the signed area of a triangle, which canbe computed with cross-products.
!
BxCy " ByCx + AyCx " AxCy + AxBy " AyBx
= (Bx " Ax )(Cy " Ay ) " (By " Ay )(Cx " Ax )
!
" =SArea(pBC)
SArea(ABC)# =
SArea(ApC)
SArea(ABC)$ =
SArea(ABp)
SArea(ABC)
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 25
Affine and convex combinationsNote that we seem to have added points together, which we said wasillegal, but as long as they have coefficients that sum to one, it’s ok.
We call this an affine combination. More generally
is a proper affine combination if:
Note that if the αi ‘s are all positive, the result is more specifically called aconvex combination.
Q: Why is it called a convex combination?
1
1n
i
i
!=
="
!
p ="1p1 +K+"
npn
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 26
Basic 3-D transformations: scaling
Some of the 3-D transformations are just likethe 2-D ones.
For example, scaling:
!
" x
" y
" z
1
#
$
% % % %
&
'
( ( ( (
=
sx 0 0 0
0 sy 0 0
0 0 sz 0
0 0 0 1
#
$
% % % %
&
'
( ( ( (
x
y
z
1
#
$
% % % %
&
'
( ( ( (
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 27
Translation in 3D
!
" x
" y
" z
1
#
$
% % % %
&
'
( ( ( (
=
1 0 0 tx
0 1 0 ty
0 0 1 tz
0 0 0 1
#
$
% % % %
&
'
( ( ( (
x
y
z
1
#
$
% % % %
&
'
( ( ( (
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 28
Rotation in 3DRotation now has more possibilities in 3D:
xR
yR
zR
Use right hand rule
!
Rx (") =
1 0 0 0
0 cos(") #sin(") 0
0 sin(") cos(") 0
0 0 0 1
$
%
& & & &
'
(
) ) ) )
Ry (") =
cos(") 0 sin(") 0
0 1 0 0
#sin(") 0 cos(") 0
0 0 0 1
$
%
& & & &
'
(
) ) ) )
Rz(") =
cos(") #sin(") 0 0
sin(") cos(") 0 0
0 0 1 0
0 0 0 1
$
%
& & & &
'
(
) ) ) )
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 29
Shearing in 3D
Shearing is also more complicated. Here is oneexample:
We call this a shear with respect to the x-z plane.
!
" x
" y
" z
1
#
$
% % % %
&
'
( ( ( (
=
1 b 0 0
0 1 0 0
0 0 1 0
0 0 0 1
#
$
% % % %
&
'
( ( ( (
x
y
z
1
#
$
% % % %
&
'
( ( ( (
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 30
Preservation of affine combinationsA transformation F is an affine transformation if it preserves affine
combinations:
where the pi are points, and:
Clearly, the matrix form of F has this property.One special example is a matrix that drops a dimension. For example:
This transformation, known as an orthographic projection, is an affinetransformation.
We’ll use this fact later…
1
1n
i
i
!=
="
!
" x
" y
1
#
$
% % %
&
'
( ( (
=
1 0 0 0
0 1 0 0
0 0 0 1
#
$
% % %
&
'
( ( (
x
y
z
1
#
$
% % % %
&
'
( ( ( (
!
F("1p1 +K+"
npn ) ="
1F(p1 ) +K+"
nF(pn )
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 31
Properties of affine transformations
Here are some useful properties of affinetransformations:
Lines map to linesParallel lines remain parallelMidpoints map to midpoints (in fact, ratios arealways preserved)
!
ratio =pq
qr=
s
t=
" p " q
" q " r
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 32
Summary
What to take away from this lecture:All the names in boldface.How points and transformations are represented.What all the elements of a 2 x 2 transformation matrixdo and how these generalize to 3 x 3 transformations.What homogeneous coordinates are and how they workfor affine transformations.How to concatenate transformations.The rules for combining points and vectorsThe mathematical properties of affine transformations.
University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell 33
Next class: Shading
Topics we’ll cover:
- How does light interact with surfaces?
- What approximations do we use to model this interaction in computer graphics?Read:
Watt, sections 6.2 – 6.3
Optional Reading:Watt, chapter 7.