Professional Documents
Culture Documents
PDPM
INDIAN INSTITUTE OF INFORMATION TECHNOLOGY,
DESIGN AND MANUFACTURING JABALPUR
(An Institute of National Importance (INI) established by MHRD, Govt. of INDIA)
9-Nov-17 Dr. P. K. Jain (IIITDMJ) 4
P (t ) [ x (t ) y (t ) z (t )] 0 t t max
function of the single parameter t.
Rotation about an axis causes the location of the point to also be a
function of the rotation angle Ø.
Thus, a point on a surface of revolution is specified by two
parameters t and Ø.
It is bi-parametric function.
Rotation about the x-axis of an entity, initially lying in the xy plane,
the surface equation is;
Q (t , ) [ x (t ) y (t ) cos y (t ) sin ]
Surface of revolution
Sphere
Rotating plane curves also yields surfaces
of revolution.
A sphere is obtained by rotating an
origin-centered semicircle in the xy
plane about the x-axis.
the parametric equation of the circle
x r cos 0
y r sin
the parametric equation of the sphere is
Ellipsoid Paraboloid
An ellipsoid of revolution is obtained by
rotating parametric equation of an A paraboloid of revolution is obtained by rotating the parametric
origin centered semi ellipse in the xy parabola
plane.
x a 2 0 max
the parametric equation of the semi ellipse
y 2 a
x a cos 0 about the x-axis. The parametric surface is given by
y b sin
Q( , ) a 2 2a cos 2a sin 0 max
parametric equation for any point on the
ellipsoid of revolution is 0 2
A hyperboloid of revolution is obtained by rotating the parametric Any parametric curve can be used to create a surface of revolution.
hyperbola Like cubic spline, Bezier and B-spline curves.
x a sec 0 max
y b tan
about the x-axis. The parametric surface is given by
Surface of Revolution
in matrix form a parametric space curve is given by Derivatives: Surface of Revolution
P(t)=[T] [N] [G] Formally differentiating yields the parametric derivatives for a surface of
where [T], [N] and [G] are parameter, blending function and geometry revolution.
matrices, respectively. Specifically, the derivative in the axial direction is
The general form of the matrix equation for a surface of revolution is Qt(t, ø)=[T’] [N] [G] [S]
Q(t, ø)=[T] [N] [G] [S] and in the radial direction
where [S] represents the contribution due to rotation about an axis by Q ø(t, ø)=[T] [N] [G] [S’]
the angle ø. For the specific case of rotation about the x-axis
where the prime denotes appropriate differentiation.
1 0 0 0
0 1 0 0
[T ( s)] 0 s 1
0 0 1 0
0 0 ns 1
9-Nov-17 Dr. P. K. Jain (IIITDMJ) 17 9-Nov-17 Dr. P. K. Jain (IIITDMJ) 18
Consider the line segment in the xy plane and parallel to the y-axis defined by end P(t)=P1+(P2- P1)t
points P1[0 0 0] and P2[0 3 0]. Determine the point at t=0.5, s=0.5 on the sweep = [0 0 0 1]+[0-0 3-0 0-0 1-1]t
surface generated by simultaneously translating the line 10 units along the x-axis and = [0 3t 0 1]
rotating it through 2 about the x-axis. The sweep surface is given by
Q(t, s)=[P(t)] [T(s)]
1 0 0 0
0 cos 2 s sin 2 s 0
0 3 t 0 1
0 sin 2 s cos 2 s 0
ls 0 0 1
1 0 0 0
0 1 0 0
Q ( 0 .5 , 0 .5 ) 0 1 .5 0 1
0 0 1 0
5 0 0 1
5 1 .5 0 1
cos sin 0 0
sin cos 0 0
T ( s )
0 0 1 0
10 s cos(s ) 1 0 1
Contd…
9-Nov-17 Dr. P. K. Jain (IIITDMJ) 21 9-Nov-17 Dr. P. K. Jain (IIITDMJ) 22
0 and
2 4 2
The shape of the interior of the surface near each corner is controlled
by the twist vector or cross derivative at the corner. For the spherical
surface patch, the cross derivative or twist vector is
The normal to a surface patch at any point is given by the cross
2 product of the parametric derivatives.
Q
Q , , sin cos cos cos 0 Specifically for a spherical surface.
i j k
Consequently a quadrilateral surface patch can be completely Q Q sin sin cos sin 0
described by
cos cos sin cos sin
The 4 position vectors at the corners.
The 8 tangent vectors, two at each corner.
The 4 twist vectors at the corners.
cos sin2 sin sin2 sin cos
All ruled surfaces are not developable surface, however all developable Euler the great Swiss mathematician, Showed that unique directions for
surfaces are ruled. which the curvature is a minimum and a maximum exists.
To find whether surface or a portion of a surface is developable or not The curvatures in these directions are called the principle curvatures, Kmin
curvature of parametric surface is considered. and Kmax.
The principal curvature directions are orthogonal.
At any point P on a surface, the curve Two combinations of the principal curvatures are of particular interest,
of intersection of a plane containing
Average
the normal to the surface at P and the min max
H
surface, has a curvature K. 2
As the plane is rotated about the and
normal, the curvature changes. Gaussian curvatures.
There will be unique directions for K min max
which the curvature is a minimum and For a developable surface the Gaussian curvature K is everywhere
a maximum exists. zero, i.e., K=0.
Developable Surfaces
Dill has shown that for bi-parametric surfaces the average and Show that an elliptic cone is a developable surface.
Gaussian curvatures are given by: Rewriting equation for a parametric elliptic cone in terms of u and w
2 2 yields Q (u , w ) au cos w bu sin w cu
A Qw 2 BQu Qw C Qu
H 3
2 Qu Qw The partial derivatives are:
AC B 2 Qu a cos w b sin w c
K 4
Qu Qw Qw au sin w bu cos w 0
where Quw a sin w b cos w 0
A B C Qu Qw Quu Quw Qww Quu 0 0 0
Linear Coons Surface For a single normalized cubic spline segment with known tangent and
Are not sufficiently flexible for many position vectors at the ends, each of the four boundary curves, P(u, 0),
applications.
P(u, 1), P(0, w) and P(1, w), is given by
The coons bicubic surface patch uses
P(t) = [T] [N] [G]
normalized cubic splines for all four
boundary curves. 2 2 1 1 P1
Cubic blending functions are used to 3 3 2 1 P2
define the interior of the patch. t 3
t 2
t 1
0 0 1 0 P1'
0 t 1
Thus, each boundary curve is of the
1 0 0 0 P2'
general form
P(t) = B1t + B2t + B3t2 + B4t3 where t becomes u or w as appropriate and P1, P2, P1’, P2’ are the position
0≤ t ≤ 1 and tangent vectors at the ends of the appropriate boundary curve.
Bezier Surfaces
Bezier Surfaces
Coons bicubic surfaces provide a flexible and powerful surface design tool. However, A Bezier surface is given by
n m
practical usage suffers from the necessity of specifying precise, non-intuitive Q (u , w ) B i, j J n ,i ( u ) K m , j ( w )
mathematical information, e.g., position, tangent and twist vectors. i0 j 0
where Jn, i(u) and Km, j(w) are the Bernstein basis functions in the u and w
parametric directions. Repeating the definition previously given for
convenience yields
n m
J n , i ( u ) u i (1 u ) n i and Km, j (w) wi (1 w) m j
i j
Many properties of the surface are similar to curves and are known.
The degree of the surface in each parametric direction is one less than the
number of defining polygon vertices in that direction.
The continuity of the surface in each parametric direction is two less than
the number of defining polygon vertices in that direction.
The surface generally follows the shape of the defining polygon net.
Only the corner points of the defining polygon net and the surface are
coincident.
The surface is contained within the convex hull of the defining polygon
net.
The surface does not exhibit the variation diminishing property. The
variation diminishing property for bivariant surfaces is both undefined and
unknown.
The surface is invariant under an affine transformation.
B 0 ,0 ... B0 ,m
B ... ... ...
B n , 0 ... B n , m
The derivatives of a Bezier surface are obtained by formal The relationship between a bicubic Bezier and bicubic Coons surface is
differentiation of equations. easily found.
First and second parametric derivatives are Recalling equations and equating them yields
n m
QCoons(u, w) = QBezier(u, w)
Qu (u , w ) Bi , j J 'n ,i (u ) K m , j ( w )
i0 j 0
n m
[U] [NC] [P] [NC]T [W] = [U] [NB] [B] [NB]T [W]
Qw (u , w) Bi , j J n ,i (u ) K 'm , j ( w )
i 0 j 0
n m
where [NC] and [NB] are known equations. Hence the bicubic Coons
Quw (u , w ) Bi , j J 'n , i (u ) K ' m , j ( w )
i0 j 0 surface geometric matrix [P] is given in terms of the Bezier surface
n m
Quu (u , w) Bi , j J ' 'n ,i (u ) K m , j ( w) polygon net as
P N C 1 N B B N B T N C T
1
i 0 j 0
n m
Qww (u , w ) Bi , j J n , i (u ) K ' ' m , j ( w)
i 0 j 0
Bezier surface
In matrix form
P(0,0) P(0,1) Pw (0,0) Pw (0,1)
Using previous equations this can also be shown
P(1,0) P(1,1) P (1,0) P (1,1)
w w
Pu (0,0) Pu (0,1) Puw (0,0) Puw (0,1)
Pu (1,0) Pu (1,1) Puw (0,1) Puw (1,1)
P(0,0) P(0,1) Pw (0,0) Pw (0,1) Q(0,0) Q(0,1) Qw(0,0) Qw (0,1)
B0 ,0 B0,3 3( B0,1 B0,0 ) 3( B0,3 B0 ,2 ) P(1,0) P(1,1) Pw(1,0) Pw (1,1) Q(1,0) Q(1,1) Qw (1,0) Qw (1,1)
B3,0 B3,3 3( B3,1 B3,0 ) 3( B3,3 B3,2 )
Pu (0,0) Pu (0,1) Puw(0,0) Puw(0,1) Qu (0,0) Qu (0,1) Quw(0,0) Quw(0,1)
3( B1,0 B0,0 ) 3( B1,3 B0,3 ) 9( B0 ,0 B1,0 B0,1 B1,1 ) 9( B0, 2 B1,2 B0 ,3 B1,3 )
3( B3,0 B2,0 ) 3( B3,3 B2,3 ) 9( B2,0 B3,0 B2 ,1 B3,1 ) 9( B2 ,2 B3,2 B2,3 B3,3 ) Pu (1,0) Pu (1,1) Puw(0,1) Puw(1,1) Qu (1,0) Qu (1,1) Quw(0,1) Quw(1,1)
• Lower right 2x2 submatrix confirms that the center four defining polygon net
vertices influence the twist at the bicubic Bezier patch corners.
• The twist at a corner controlled by not only the center polygon vertices but also by
the adjacent tangent vectors.
• In fact, the twist at the corner is controlled by the shape of the nonplanar
quadrilateral formed by the corner, the two adjacent boundary points and the
adjacent center point.
9-Nov-17 Dr. P. K. Jain (IIITDMJ) 82 9-Nov-17 Dr. P. K. Jain (IIITDMJ) 83
Similarly the inverse relationship between [P] and [B] which gives the Bezier For C0 continuity along the edge of two boundary curves and hence two
polygon net vertices in terms of the Coons bicubic surface parameters is boundary polygons along edge must be coincident.
For C1 continuity across the patch boundary, surface normal direction along
B0,0 B0,1 B0,2 B0,3 boundary edge must be same. Two conditions may be used to achieve this:
B Four polygon net lines that meet at and cross the boundary edge to be collinear.
B1,1 B1,2 B1,3
1,0 Three polygon net edges meeting at the ends of the boundary curve to be coplanar.
B2,0 B2,1 B2, 2 B2,3
B3,0 B3,1 B3,2 B3,3
B-Spline surface
9-Nov-17 Dr. P. K. Jain (IIITDMJ) 94 9-Nov-17 Dr. P. K. Jain (IIITDMJ) 95
B Spline surface
B1,1[-15 0 15] B2,1[-5 5 15] B3,1[5 5 15] B4,1[15 0 15]
4 4
Q(u, w) Bi, j Ni, 4 (u)M j ,3 (w)
B1,2[-15 5 15] B2,2[-5 10 5] B3,2[5 10 5] B4,2[15 5 5] i 1 j 1
B1,3[-15 5 -5] B2,3[-5 10 -5] B3,3[5 10 -5] B4,3[15 5 -5] N1,4 B1,1M1,3 B1,2M2,3 B1,3M3,3 B1,4M4,3
N2,4 B2,1M1,3 B2,2M2,3 B2,3M3,3 B2,4M4,3
B1,4[-15 0 -15] B2,4[-5 5 -15] B3,4[5 5 -15] B4,4[15 0 -15]
N3,4 B3,1M1,3 B3,2M2,3 B3,3M3,3 B3,4M4,3
The surface is fourth-order in the u direction (k=4) and third order in the w
direction (l=3). Thus the surface is composed of two subpatches: one for 0 u 1, N4,4B4,1M1,3 B4,2M2,3 B4,3M3,3 B4,4M4,3
0 w 1 and the other for 0 u 1, 1 w 2.
Determine the surface point at the center of the surface, i.e., at u=1/2, w=1.
h i, j B h i , j N i , k ( u ) M j ,l ( w ) n 1 m 1
i 1 j 1
Q (u , w ) n 1 m 1
Q (u , w ) Bi , j S i , j (u , w)
i 1 j 1
h
i 1 j 1
i, j N i , k ( u ) M j ,l ( w )
where the Bi, j’s are the 3D defining polygon net points and the Si, j(u, w) are
bivariant rational B-spline surface basis functions
h N i , k (u ) M
i, j j ,l ( w)
S i , j (u , w ) n 1 m 1
hi1, j1 N i1,k (u ) M
i 1 j 1
j 1, l ( w)
9-Nov-17 Dr. P. K. Jain (IIITDMJ) 100 9-Nov-17 Dr. P. K. Jain (IIITDMJ) 101
Spline surfaces have similar analytic and geometric properties to their A rational B-spline surface is invariant with respect to a projective
nonrational counterparts. transformation; i.e., any projective transformation can be applied to the
Specifically, surface by applying it to the defining polygon net. Note this is a stronger
The sum of the rational surface basis functions for any u, w values is condition than that for a nonrational B-spline surface.
n 1 m 1 The surface lies within the convex hull of the defining polygon net formed by
Si 1 j 1
i, j (u , w ) 1 taking the union of all convex hull of k, l neighboring polygon net vertices.
The variation diminishing property is not known for rational B-spline surface.
Each rational surface basis function is positive or zero for all parameter values
u, w, i.e., Si, j 0. The influence of a single polygon net vertex is limited to k/2, l/2 spans in
each parametric direction.
Except for k=1 or l=1, each rational surface basis function has precisely one
maximum. If triangulated, the defining polygon net forms a planar approximation to the
surface.
The maximum order of a rational B-spline surface in each parametric
direction is equal to the number of defining polygon vertices in that direction. If the number of defining polygon net vertices is equal to the order in each
parametric direction and there are no duplicate interior knot values, the
A rational B-spline surface of order k, l (degree k-1, l-1) is Ck-2 , Cl-2 continuous
rational B-spline surface is a rational Bezier surface.
everywhere.
9-Nov-17 Dr. P. K. Jain (IIITDMJ) 102 9-Nov-17 Dr. P. K. Jain (IIITDMJ) 103
Rational B-spline surfaces with n+1 = 5, m+1 = 4, k = l= 4 Rational B-splined ruled surface
9-Nov-17 Dr. P. K. Jain (IIITDMJ) 106 9-Nov-17 Dr. P. K. Jain (IIITDMJ) 107
N N u Du
Qu
D N D
N N w Dw
Qw
D N D
N N uw N u D w N w D u D u D w D uw
Q uw
D N N D N D 2 D D D
2
N N uu N D D D
Q uu 2 u u 2 u 2 uu
D N N D D D
2
N N ww N D D D
Q ww 2 w w 2 w2 ww
D N N D D D
9-Nov-17 Dr. P. K. Jain (IIITDMJ) 108 9-Nov-17 Dr. P. K. Jain (IIITDMJ) 109
n 1 m 1 n 1 m 1
N w hi , j Bi , j N i , k (u ) M ' j ,l ( w ) Dw hi , j N i ,k (u ) M ' j ,l ( w)
i 1 j 1 i 1 j 1
n 1 m 1 n1 m1
N uw hi , j Bi , j N 'i , k (u ) M ' j , l ( w ) Duw hi , j N 'i ,k (u)M ' j ,l (w)
i 1 j 1
i 1 j 1
n 1 m 1 n 1 m 1
N uu hi , j Bi , j N ' 'i , k (u ) M j ,l ( w) Duu hi , j N ' 'i , k (u ) M j ,l ( w )
i 1 j 1 i 1 j 1
n 1 m 1 n 1 m 1
N ww hi , j Bi , j N i , k (u ) M ' ' j ,l ( w ) Dww hi , j N i ,k (u ) M ' ' j ,l ( w) A quadratic surface element within a more general rational B-spline surface. Airfoil leading edge. (a) Polygon net; (b) surface. Ship stern
i 1 j 1 i 1 j 1 (c) Polygon net; (d) surface. A cylinder as part of a more general surface. (e) Polygon net; (f) surface.
9-Nov-17 Dr. P. K. Jain (IIITDMJ) 110 9-Nov-17 Dr. P. K. Jain (IIITDMJ) 111
48 degrees of freedom or algebraic coefficients need to be supplied we can generate 16 of these equations, one for each of the 16 points.
Let us use the u, w values. Thus,
a33 (0)3(0)3 + a32 (0)3(0)2 + ……..... + a00 = p(0, 0)
a33 (1/3)3(0)3 + a32 (1/3)3(0)2 + ….. + a00 = p(1/3, 0)
- -
- -
a33 (2/3)3(1/3)3 + a32 (2/3)3(1/3)2 + .. + a00 = p(2/3, 1/3)
- -
- -
a33 (1)3(1)3 + a32 (1)3(1)2 + ……….. + a00 = p(1, 1)
References
9-Nov-17 Dr. P. K. Jain (IIITDMJ) 114 9-Nov-17 Dr. P. K. Jain (IIITDMJ) 130