OpenGL Notes a

Stu Pomerantz
smp@psc.edu
http://www.psc.edu/~smp
December 8, 2004

a Mostmaterial is adapted from: OpenGL ARB, et. al, “The OpenGL Pro-
gramming Guide”, Third Ed., Reading: Addison-Wesley, 1999

1
 2D Parametric Curves

In two dimensions a parametric curve is defined by:

x = x(u)
y = y(u)
Where u is the parameter that is free to vary. For example:

x = cos(θ)
y = sin(θ)
x and y will trace out a circle as 0 ≤ θ ≤ 2π .

222
2D Parametric Curves

Drawing of the parametric circle:

223
 2D Parametric Curves

Code for the parametric circle:

glBegin(GL_LINES) ;
du = PI/32 ;
for( u = 0 ; u < 8*PI ; u += du) {
x = cos(u) ; y = sin(u) ; z = 0 ;
glVertex3f(x,y,z) ;

x = cos(u+du) ; y = sin(u+du) ; z = 0 ;
glVertex3f(x,y,z) ;
}
glEnd() ;

224
 3D Parametric Curves

In two dimensions a parametric curve is defined by:

x = x(u)
y = y(u)
z = z(u)
Where u is the parameter that is free to vary. For example:

x = cos(θ)
y = sin(θ)
z=θ
x, y and z will trace out a circlular helix as 0 ≤ θ ≤ 2π .

225
3D Parametric Curves

Drawing of the parametric helix:

226
 3D Parametric Curves

Code for the parametric helix:

glBegin(GL_LINES) ;
du = PI/32 ;
for( u = 0 ; u < 8*PI ; u += du) {
x = cos(u) ; y = sin(u) ; z = u ;
glVertex3f(x,y,z) ;

x = cos(u+du) ; y = sin(u+du) ; z = u+du ;

glVertex3f(x,y,z) ;
}
glEnd() ;

227
 3D Parametric Surfaces

In three dimensions a parametric surface is defined by:

x = x(u, v)
y = y(u, v)
z = z(u, v)
Where u and v are the parameters that are free to vary. For
example:

x = cos(θ) sin(φ)
y = sin(θ) sin(φ)
z = cos(φ)
x, y and z will trace out a sphere as 0 ≤ θ ≤ 2π and 0 ≤ φ ≤ 2π .

228
3D Parametric Surfaces

Drawing of the parametric sphere:

229
 Polynomial Parametric Curves

n
X
p~(u) = uk~c k

k=0

where

 
cxk
k
 
~c =  cyk 


czk

230
 Polynomial Parametric Curves

For example if n = 3,
3
X
p~(u) = uk~c k

k=0

expands to:

       
cx3 cx2 cx1 cx0

3
 2
     
 + u  cy2  + u  cy1  +  cy0 
p~(u) = u  cy3       

cz3 cz2 cz1 cz0

231
 Polynomial Parametric Curves

Continuing,
       
cx3 cx2 cx1 cx0
3
 
2
    
p~(u) = u  cy3  + u  cy2  + u  cy1  +  cy0 
     

cz3 cz2 cz1 cz0

expands to

x(u) = cx3 u3 + cx2 u2 + cx1 u + cx0

p~(u) = y(u) = cy3 u3 + cy2 u2 + cy1 u + cy0
z(u) = cz3 u3 + cz2 u2 + cz1 u + cz0

Notice that 4 coefficient vectors which must be specified.

232
 Polynomial Parametric Surfaces

 
x(u, v) n X m
  X i j
p~(u, v) = 
 y(u, v) =
 ~
c ij u v
i=0 j=0
z(u, v)

So, if n = m = 3 then 16 coefficient vectors will be needed to

evaluate the formula and determine the value of p~
• u for curves, and u and v for surfaces have an infinite domain.
• Generally, though, only an interval of values for the domain of
u and v is desired.
• So, without lost of generality, restrict the range of u and v to
[0, 1].

233
 Interpolation Revisited

Two points, p~0 and p~1 , can be linearly interpolated like this:

p~ = a p~0 + (1 − a) p~1
where a ∈ [0, 1]
In this context, interpolation is also called blending. The points are
mixed together by the a and 1 − a which are weights.
The points p~0 and p~1 can be thought of as control points since they
influence (constrain) the final interpolated value.
This is similar, but not identical to, OpenGL’s blending functions.
And, in this case the purpose is not to produce transparent
surfaces but rather to produce a new interpolated point.

234
 Interpolation using Cubic Polynomials

Recall the cubic polynomial

3
X
p~(u) = uk~c k

k=0

Requires 4 vectors of coefficients in order for it to be completely

determined.
Rewrite this equation...

235
 Interpolation using Cubic Polynomials

..using a matrix representation of the coefficient vectors.

p~(u) = [x(u) y(u) z(u)] =

   
m11 m12 m13 m14 g1x g1y g1z
   
 m21 m22 m23 m24
¤    g g2y g2z 
£ 3 2   2x
u u u 1 · ·


 m31 m32 m33 m34   g3x g3y g3z 
   
m41 m42 m43 m44 g4x g4y g4z

Or more briefly,
p~(u) = U · M · G

Call M the basis matrix. Call G the Geometry matrix.

236
 Interpolation using Cubic Polynomials

For example,

x(u) = (u3 m11 + u2 m21 + um31 + m41 )g1x +

(u3 m12 + u2 m22 + um32 + m42 )g2x +
(u3 m13 + u2 m23 + um33 + m43 )g3x +
(u3 m14 + u2 m24 + um34 + m44 )g4x +

This form emphasizes that the curve is a weighted sum of the

elements of G (geometry) matrix.

237
 Interpolation using Cubic Polynomials

• The geometry matrix, G, contains 4 points. They are called

control points.
• Analogous to the 2 control points in the previous linear
interpolation example.
• The meaning of these 4 points is dependant on the basis matrix
M.
• The basis matrix M affects
– The continuity of the curve.
– Which and whether control points are interpolated.

238
 The Hermite Basis

The Hermite basis is named after the Mathematician. The

construction the Hermite basis matrix is as follows:
Given 4 control points: p~1 , p~2 , p~3 , p~4
• Constrain the curve so that it interpolates p~1 at u = 0 and p~4
at u = 1
• Let p~2 and p~3 be the tangent vectors for p~1 and p~4 respectively.

239
 The Hermite Basis

Recall,
£ 3 2
¤
U= u u u1
so,
0
£ 2
¤
U = 3u 2u 1 0

There are 4 equations and 4 unknowns:

p~1 = [0 0 0 1] (u = 0)
p~4 = [1 1 1 1] (u = 1)
p~2 = [0 0 1 0] (u0 = 0)
p~3 = [3 2 1 0] (u0 = 1)

240
 The Hermite Basis

Rewriting the 4 equations above as a matrix:

   
p~1 0 0 0 1
   
 p~   1 1 1 1 
 4  
=

 
 p~2   0 0 1 0 
   
p~3 3 2 1 0

For these equations to be satisfied M , the Hermitian basis matrix

must be the inverse of this matrix.

241
 The Hermite Basis

The Hermitian basis matrix:

 
2 −2 1 1
 
 −3 3 −2 −1 
M =
 

 0 0 1 0 
 
1 0 0 0
So,

(2u3 − 3u2 + 1)~

p1 + (−2u3 + 3u2 )~
p4 +
p~(u) =
(u3 − 2u2 + u)~
p3 + (u3 − u2 )~
p4

242
 The Hermite Basis

A picture of the Hermitian basis functions:

243
Example Hermite Curves

244
 Comparison of Curve Types

Curves differ by basis matrix.

Hermite Bezier Non-Uniform B-Spline Catmull-Rom
A Y Y N Y
B Y N N Y
C C0 C0 C2 C1
A = Interpolates some control points
B = Interpolates all control points
C = Inherent Continuity

If the direction and magnitude of the nth derivative are equal at

the joint point, the curve is called C n continuous.

245
Example NURBS Surface

246