Foundations of Computer Graphics

Online Lecture 2: Review of Basic Math Vectors and Dot Products Ravi Ramamoorthi

Complete HW 0

Course: Next Steps

Sets up basic compilation issues Verifies you can work with feedback/grading servers This lecture is a revision of basic math concepts

First few lectures core math ideas in graphics HW 1 has few lines of code (but start early) Textbooks: None required

Use some ideas discussed in lecture, create images OpenGL/GLSL reference helpful (but not required)

Motivation and Outline

Many graphics concepts need basic math like linear algebra Vectors (dot products, cross products, ) Matrices (matrix-matrix, matrix-vector mult., ) E.g: a point is a vector, and an operation like translating or rotating points on object can be matrix-vector multiply Should be refresher on very basic material for most of you Only basic high school math required

Vectors
=
Usually written as a or in bold. Magnitude written as a Length and direction. Absolute position not important

Use to store offsets, displacements, locations

But strictly speaking, positions are not vectors and cannot be added: a location implicitly involves an origin, while an offset does not.

Vector Addition
a = +b
a
Geometrically: Parallelogram rule In cartesian coordinates (next), simply add coords

Cartesian Coordinates
A=4X+3Y

b+

a
b

x A = y

A =

x y

A = x2 + y 2

X and Y can be any (usually orthogonal unit) vectors

Vector Multiplication
Dot product Cross product Orthonormal bases and coordinate frames

Dot (scalar) product

b

aib = bia = ?
Note: We use right-handed (standard) coordinates

Dot (scalar) product

b

Dot (scalar) product

b

aib = bia = ?

a i b = a b cos aib = cos1 a b

aib = bia = ? a i (b + c) = a i b + a i c (ka) i b = a i (kb) = k(a i b)

a i b = a b cos aib = cos1 a b

Dot product in Cartesian components

Dot product in Cartesian components

x x a b =? a b= ya yb

x x a b =? a b= ya yb
x x a b = xa xb + y a y b a b= ya yb

Dot product: some applications in CG

Find angle between two vectors (e.g. cosine of angle between light source and surface for shading)

Projections (of b on a)
b

Finding projection of one vector on another (e.g. coordinates of point in arbitrary coordinate system) Advantage: computed easily in cartesian components

ba =? ba =?

b a = b cos = ba= ba

aib a

a aib = a 2 a a

Foundations of Computer Graphics

Online Lecture 2: Review of Basic Math Vectors: Cross Products Ravi Ramamoorthi

Vector Multiplication
Dot product Cross product Orthonormal bases and coordinate frames Note: We use right-handed (standard) coordinates

Cross (vector) product

a b = b a a b = a b sin b
Cross product orthogonal to two initial vectors Direction determined by right-hand rule Useful in constructing coordinate systems (later)

Cross product: Properties

x y = +z y x = z y z = +x z y = x z x = +y x z = y a b = b a aa = 0 a (b + c) = a b + a c a (kb) = k(a b)

Cross product: Cartesian formula?

y z y z b a a b a b = xa y a za = za xb xa zb xb y b zb xa y b y a xb 0 z y a xb a 0 xa y b a b = A*b = za 0 zb y a xa x y z

Cross product: Cartesian formula?

Dual matrix of vector a

Foundations of Computer Graphics

Online Lecture 2: Review of Basic Math Vectors: Orthonormal Basis Frames Ravi Ramamoorthi

Vector Multiplication
Dot product Cross product Orthonormal bases and coordinate frames Note: We use right-handed (standard) coordinates

Orthonormal bases/coordinate frames

Important for representing points, positions, locations Often, many sets of coordinate systems (not just X, Y, Z) Global, local, world, model, parts of model (head, hands, ) Critical issue is transforming between these systems/bases Topic of next 3 lectures

Coordinate Frames
Any set of 3 vectors (in 3D) so that

u = v = w =1 u iv = v iw = u iw = 0 w = u v p = (p i u)u + (p i v)v + (p i w)w

Constructing a coordinate frame

Often, given a vector a (viewing direction in HW1), want to construct an orthonormal basis Need a second vector b (up direction of camera in HW1) Construct an orthonormal basis (for instance, camera coordinate frame to transform world objects into in HW1)

Constructing a coordinate frame?

We want to associate w with a, and v with b But a and b are neither orthogonal nor unit norm And we also need to find u

Constructing a coordinate frame?

We want to associate w with a, and v with b But a and b are neither orthogonal nor unit norm And we also need to find u

Constructing a coordinate frame?

We want to associate w with a, and v with b But a and b are neither orthogonal nor unit norm And we also need to find u

w=

a a

w= u=

a a

bw bw

Constructing a coordinate frame?

We want to associate w with a, and v with b But a and b are neither orthogonal nor unit norm And we also need to find u

Foundations of Computer Graphics

Online Lecture 2: Review of Basic Math Matrices Ravi Ramamoorthi

a w= a u=

bw bw

v =w u

Matrices
Can be used to transform points (vectors)
Translation, rotation, shear, scale (more detail next lecture)

What is a matrix
Array of numbers (mn = m rows, n columns)

1 3 5 2 0 4
Addition, multiplication by a scalar simple: element by element

Matrix-matrix multiplication
Number of columns in first must = rows in second
1 3 3 6 9 4 5 2 2 7 8 3 0 4

Matrix-matrix multiplication
Number of columns in first must = rows in second
9 27 33 13 1 3 3 6 9 4 5 2 2 7 8 3 = 19 44 61 26 8 28 32 12 0 4

Element (i,j) in product is dot product of row i of first matrix and column j of second matrix

Element (i,j) in product is dot product of row i of first matrix and column j of second matrix

Matrix-matrix multiplication
Number of columns in first must = rows in second
9 27 33 13 1 3 3 6 9 4 = 19 44 61 26 5 2 2 7 8 3 8 28 32 12 0 4

Matrix-matrix multiplication
Number of columns in first must = rows in second
9 27 33 13 1 3 3 6 9 4 = 19 44 61 26 5 2 2 7 8 3 8 28 32 12 0 4

Element (i,j) in product is dot product of row i of first matrix and column j of second matrix

Element (i,j) in product is dot product of row i of first matrix and column j of second matrix

Matrix-matrix multiplication
Number of columns in first must = rows in second
1 3 3 6 9 4 5 2 NOT EVEN LEGAL!! 2 7 8 3 0 4 Non-commutative (AB and BA are different in general)
Associative and distributive A(B+C) = AB + AC (A+B)C = AC + BC

Matrix-Vector Multiplication
Key for transforming points (next lecture) Treat vector as a column matrix (m1) E.g. 2D reflection about y-axis
1 0 x x = 0 1 y y

Transpose of a Matrix (or vector?)

1 2 1 3 5 3 4 = 2 4 6 5 6
T

Identity Matrix and Inverses

I33 1 0 0 = 0 1 0 0 0 1

(AB)T = BT AT

AA1 = A1A = I (AB)1 = B 1A1

Vector multiplication in Matrix form

Dot product?
a b = aT b

(x
Cross product?

ya

za

0 a b = A b = za y a
*

x b y b = xa xb + y a y b + za zb zb

za 0 xa

Dual matrix of vector a

y a xb xa y b 0 zb