Advanced Modeling in MultiSurf 6.0

MultiSurf 6.
0 Reference: Advanced Modeling

MULTISURF 6.0 REFERENCE: ADVANCED MODELING ..................................1
ADVANCED MODELING .....................................................................................4

Points ............................................................................................................................................................. 4
Blend Point ................................................................................................................................................. 4
Intersection Point........................................................................................................................................ 6
Tabulated Point........................................................................................................................................... 7
Tangent Magnet.......................................................................................................................................... 8
XYZ Magnet............................................................................................................................................. 10
Proximity Points ......................................................................................................................................... 11

Proximity Bead, Ring, Magnet ................................................................................................................. 11
Curves and Snakes ..................................................................................................................................... 14

B-Fitted Curve.......................................................................................................................................... 14
Conic ........................................................................................................................................................ 15
Expanded Curve ....................................................................................................................................... 18
NURBS Curve.......................................................................................................................................... 19
Radius Arc................................................................................................................................................ 20
Tabulated Curve ....................................................................................................................................... 22
X-Curve .................................................................................................................................................... 24
Arc Snake ................................................................................................................................................. 26
Foil Snake................................................................................................................................................. 27
Geodesic Snake ........................................................................................................................................ 29
NURBS Snake.......................................................................................................................................... 32
PolySnake................................................................................................................................................. 33
NUBFitCurve, NUBFitSnake, NUBFitSurf entities................................................................................. 34

NUBFitCurve ........................................................................................................................................... 34
NUBFitSnake ........................................................................................................................................... 36
NUBFitSurf .............................................................................................................................................. 36
Tools and NUBFit Surf ............................................................................................................................ 37
Surfaces ....................................................................................................................................................... 38
Foil Lofted Surfaces ................................................................................................................................. 38
B-Fit Surface ............................................................................................................................................ 41
B Surface .................................................................................................................................................. 43
NURBS Surface ....................................................................................................................................... 44
Projected Surface...................................................................................................................................... 47
Relative Surface ....................................................................................................................................... 48
Tabulated Surface..................................................................................................................................... 50
X-Spline Lofted Surface........................................................................................................................... 52
Parametric Solids ....................................................................................................................................... 55

Additional Solid Attributes....................................................................................................................... 55
divisions: u, v, and w divisions and subdivisions ..................................................................................... 55
Orientation................................................................................................................................................ 56
1
Using Solids ................................................................................................................................................. 56
Some applications..................................................................................................................................... 56
Discussion of tank and grid examples ...................................................................................................... 57
Solid Entities ............................................................................................................................................... 57

Boundary Solid......................................................................................................................................... 57
BlockSolid ................................................................................................................................................ 58
B-spline Lofted Solid ............................................................................................................................... 59
Bsolid ....................................................................................................................................................... 60
CopySolid................................................................................................................................................. 61
Ruled Solid ............................................................................................................................................... 61
Triangle Mesh Entity Types ...................................................................................................................... 63

Triangle Mesh basics................................................................................................................................ 63
Recursive subdivision............................................................................................................................... 64
Locations on a Triangle Mesh .................................................................................................................. 64
Common data ........................................................................................................................................... 64
Triangle Mesh........................................................................................................................................... 65
Light Triangle Mesh ................................................................................................................................. 65
Surface Triangle Mesh ............................................................................................................................. 66
Copy TriMesh........................................................................................................................................... 67
PolyTriMesh............................................................................................................................................. 67
Triangle Mesh Magnet ............................................................................................................................. 68
Triangle Mesh Projected Magnet ............................................................................................................. 68
Triangle Mesh Ring.................................................................................................................................. 69
Snakes on Triangle Meshes........................................................................................................................ 69

Triangle Mesh B-spline Snake ................................................................................................................. 70
Triangle Mesh Edge Snake....................................................................................................................... 71
Triangle Mesh Intersection Snake ............................................................................................................ 71
Triangle Mesh Projected Snake................................................................................................................ 72
Triangle Mesh SubSnake.......................................................................................................................... 73
Formulas ..................................................................................................................................................... 74
Constants, variables and formulas ............................................................................................................ 74
Variables................................................................................................................................................... 75
Formulas................................................................................................................................................... 75
Basic syntax.............................................................................................................................................. 75
Expression ................................................................................................................................................ 75
Unit consistency ....................................................................................................................................... 77
Evaluation logic........................................................................................................................................ 77
Complete syntax ....................................................................................................................................... 78
Variables and formulas as parents ............................................................................................................ 78
2
P.O. Box 684 / 54 Herrick Rd.
Southwest Harbor, Maine 04679 U.S.A.
voice 207-244-4100
fax 207-244-4171
email [email protected]
website www.aerohydro.com
© 2006 AeroHydro, Inc.
3
Advanced Modeling
Points
Blend Point
Characteristic data point1 ... pointN = component point names
weight1 … weightN = corresponding weights
type = 0, 1, or 2
Description The Blended Point is a weighted sum of one or more points. Each
coordinate of the blended point is the same weighted sum of the
component point coordinates:
X = w1 X1 + w2 X2 + ...... + wN XN
Y = w1 Y1 + w2 Y2 + ...... + wN YN
Z = w1 Z1 + w2 Z2 + ...... + wN ZN
type specifies how the weight values are used in the formulas:
type = 0 weights are used without wi = weighti
modification
type = 1 weights sum to 1(forced) wi = weighti EXCEPT weightN is ignored -- it
is replaced by 1 - (sum of the other weights)
type = 2 each weight is divided by wi = weighti / sum
the sum of all weights
When weights sum to 1, it is generally easier to predict the result of the
blend; when weights do not sum to one, the blend is more complex. For
most applications, you probably want the weights to sum to 1. Thus, to
construct the midpoint between two points, you could make a type-1
BlendPoint with weights of 0.5 and 0.5 . The weights used in the
formulas would be w1 = 0.5, w2 = 1 - 0.5 = 0.5 (in a type-1 BlendPoint
weightN is ignored and 1 minus the sum of the other weights is used).
You could also use a type-1 BlendPoint to construct the mirror point of
point1 across point2. You would use weights -1.0 and +2.0; the weights
used in the formulas would then be w1 = -1.0, w2 = 1 - (-1.0) = 2.0 .
Example 1 Blend Point2.ms2
This model has two, type-1 (weights sum to 1) BlendPoints: ‘p3’ and ‘ctr’.
'p3' (cyan) is used to make the fourth corner of the parallelogram; its
component points are ‘p0’, ‘p1’, and ‘p2’ (weights -1, +1, +1). Since ‘p3’
is a type-1 BlendPoint, the weights used in the formulas are: w1 = -1, w2
= +1, and wN = 1 - (-1 +1) = 1. You can drag RelPoints ‘p1’ and ‘p2’
around as you wish, but the figure formed by BCurve 'c0' is durably
parallel-sided.
4
Z
blended points Z
'p3'
'p2' 'p3'
'ctr'
'ctr'
X X
BLENDPT2.MS2 at Lat 0, Lon -90 orthographic: initial model (left); with ‘p2’
moved (right).
The second BlendPoint 'ctr' stays at the center of the figure; it is the
average of the four corners ‘p0’, ‘p1’, ‘p2’, and ‘p3’ (weights .25, .25, .25,
.25). Since ‘ctr’ is a type-1 BlendPoint, the weights used in the formulas
are: w1 = .25, w2 = .25, w3 = .25, and wN = 1 - (.25 + .25 + .25) = .25 .
Example 2 Blend Point3.ms2
This example uses three type-1 BlendPoints and one type-2 BlendPoint to
make the standard yacht designer’s “center of effort” calculation.
Z
4 blended points
'Mctr'
'Zctr'
'CE' 'Jctr'
BLENDPT3.MS2 at Lat 0, Lon 90 orthographic.

The “ketch” rig has three triangular sails: ‘jib’, ‘main’ and ‘mizzen’. Each
sail is made from its 3 corner points. The center of each sail is located by
a type-1 BlendPoint, equally weighting the 3 corners. (This is a valid way
to locate the centroid of a triangle.) The boundary of each sail is a type-1
BCurve, and the surface of the sail is a RuledSurf between the centroid
and the boundary. The “Center of Effort” (CE, the supposed
aerodynamic center of sail force) is defined as the weighted sum of the
individual sail centroids, each weighted by the area of that sail. The sail
areas are obtained from Calcs/ Mass Properties:
Sail Area Area/sum of areas
(weights for the (w1, w2, w3 values
BlendPoint) used in the formulas)
‘Jib’ 275 sq.ft. .3681
‘Main’ 288 sq.ft. .3855
‘Mizzen’ 184 sq.ft. .2463
5
Total 747 sq.ft.
A type-2 BlendPoint ‘CE’ makes the center of effort calculation; the three
sail areas are entered as the weights. Since ‘CE’ is type-2, the weights
used in the formulas are w1 = .3681, w2 = .3855, and w3 = .2463 (as shown
in the table above). You can read the absolute coordinates of the center of
effort ‘CE’ from the Edit/Attributes box or the status line.
[Variables and Formulas provide a way to represent this complete
relationship durably. The critical element is the AREA function.]
Other Examples See use of BlendPoint in Entity Description for ProcPtSurf - Example 1.
Intersection Point
An Intersection Point is located at the common intersection point of three
planes and/or surfaces. (Actually, it solves a considerably more general
intersection problem, as it allows offsets to be specified from any
combination of the 3 planes and/or surfaces.)
Characteristic Data Magnet/surface/plane/frame1
Magnet/surface/plane/frame2
Magnet/surface/plane/frame3
Offset1, Offset2, Offset3
Description If a magnet is specified in place of a surface, the magnet location is used
as a starting point in the search for an intersection point.
As usual, a frame can serve as a plane; in this case, the plane is the x-y
plane of the frame.
When the intersector is a surface, the offset is positive in the direction of
the positive normal (like an OffsetSurf). Bear in mind that such an offset
will be reversed if the surface's orientation attribute is reversed.
Planes have an orientation, too, though this has not so far been made a
visible characteristic of a plane, so you need to follow some rules if you
want to make use of an offset from a plane:
*X=0, *Y=0, *Z=0 -- A positive offset is in the positive coordinate
direction.
Plane2 -- A positive offset is in the direction from Point1 toward Point2
Plane3 -- A right-hand rule prevails; if you made a Frame3 from the
same three points, picked in the same order, positive offset is in the
direction of the positive z-axis.
OffsetPlane -- Orientation is the same as that of the parent plane.
Frame -- A positive offset is in the direction of the positive z-axis.
Associated errors
228 -- IntPoint has duplicate Parents; no solution.
In general, if the same object is used for two of the Parents, there is
no unique solution for the intersection.
6
229 -- IntPoint: Solution failed, equations are singular.
This usually indicates a badly posed intersection, for example: two of
the planes are parallel, or a surface coincides with a plane.
232 -- IntPoint: Failed to converge.
This can also result from a badly posed intersection. Bad starting
locations are another cause; use a magnet near the intersection, in place
of a surface.
Sample files
Intersection Point1.ms2. This has 4 IntPoints;:
ip0 is at the intersection of 3 planes.
ip1 is at the intersection of 2 planes and 1 surface.
ip2 is at the intersection of 1 plane and 2 surfaces.
ip3 is at the intersection of 3 surfaces.
Tabulated Point
Characteristic data filename = filename of .3DA or .PAT file

point number = number of points in file
frame/point = name of a frame or point object or ‘*’ for the default
Description filename is the filename of a .3DA (or .PAT) file. If no extension is given,
MultiSurf assumes .3DA.
A .3DA file (see sample contents in example below) consists of a series of
records (file lines) each having four numbers: a pen number and X, Y, Z
coordinates. TabPoint ignores the pen value, and just counts each record
as an individual point. The file must have at least as many points in it as
the point no. specified, or a geometry error occurs.
point no. is the sequence number of the point to be taken from the .3DA
file: 1 for the first, 2 for the second, etc.
frame/point is the name of a frame or point object or ‘*’ for the default:
When frame/point is a point, it specifies an insert origin, and point’s
coordinates are added to the TabPoint coordinates. This results
in a shift (but no rotation).
When frame/point is a frame, the file coordinates are interpreted as
frame coordinates x,y,z. This generally results in a shift plus a
rotation.
When frame/point is the default ‘*’, there is no shift or rotation.
Example tabpoint.ms2 (needs demopts.3da)
The model generated by this file is identical to the DEMO model. If you
use Edit/ Model File, you can see the difference is that where the
DEMO.MS2 has all of its control points “hard-wired,” TABPOINT.MS2
gets its control points by reading the file DEMOPTS.3DA (see contents
7
below). Thus, the shape of the TABPOINT model can be controlled in
many ways by changing the contents of DEMOPTS.3DA.
The 4 TabPoints defining the bow master curve are named ‘P11’ - ‘P14’
(from top to bottom), those for the middle master curve ‘P21’ - ‘P24’ (top
to bottom), those for the stern master curve ‘P31’ - ‘P34’ (also top to
bottom). For all the TabPoints, filename is DEMOPTS, and frame/point is
the default ‘*’ which produces no shift or rotation of the points. point no.
for ‘P11’ is 1, for ‘P12’ is 2, for ‘P13’ is 3, for ‘P14’ is 4, for ‘P21’ is 5, for
‘P22’ is 6, etc.
Z
Y
all points are tabulated points
X
TABPOINT.MS2
Contents of DEMOPTS.3DA
0 0.000 0.000 3.600
1 1.367 0.000 0.602
1 2.324 0.000 -0.800
1 3.000 0.000 -0.900
0 15.000 4.815 2.560
1 15.000 5.046 0.628
1 15.000 3.603 -0.870
1 15.000 0.000 -1.175
0 30.000 3.500 2.760
1 30.000 3.500 1.320
1 30.000 2.500 0.120
1 30.000 0.000 0.120
See also TabCurve, TabSurf, WireFrame
Tangent Magnet
The TanMagnet (tangent magnet) entity type constructs a point on a
surface along the direction tangent to a snake. It is highly analogous to a
TanPoint in 3-D (a point along the 3-D tangent to a curve, at a place on
the curve specified by a bead). In the TanMagnet case, the tangency
direction is really in the 2-D u,v-parameter space of the surface.
TanMagnet is a key to constructing snakes that are tangent to each other
in both u,v parameter space and in 3-D.
8
A TanMagnet is a draggable point. The tangent direction is fixed by the
ring and snake parents, so only the u,v-offset (a single degree of
freedom) is controlled by dragging or nudging.
Characteristic data
Ring -- This designates both the snake to be tangent to, and the point
along the snake where the tangent direction is taken. The host surface of
the TanMagnet is the host surface of the ring (and therefore of the snake).
u,v-offset -- a distance in u,v parameter space from the ring parent. This
is a signed distance, positive along the positive tangent direction at ring.
Construction
The program evaluates the ring parent, obtaining the host snake and a t
parameter value. Then it makes a level-1 evaluation of the snake at this
parameter value, obtaining the host surface, a u,v location for the ring,
and the first derivative of the u,v location. It normalizes the first
derivative to obtain a unit vector U in u,v space. It multiplies U by the
u,v-offset and adds the resulting 2D vector to the ring's u,v coordinates,
to obtain the u,v coordinates of the TanMagnet.
Error conditions
(1) The ring can fail to evaluate, or the snake can fail to evaluate at this t
position (though this seems unlikely, if the ring is valid), or the surface
can fail to evaluate at this u,v location. These failures produce error 284
(parent failed).
(2) The first derivative of the snake can be zero, so the normalization fails
and a tangent direction can't be established. This produces a new error,
473.
Sample files Tangent Magnet.ms2

AbsRing 'r0' (bright white) is on BSnake 'n0' (bright magenta). 'r0' is used
as parent for TanMagnet 'm0' (bright red); the u,v-offset is 0.200. 'r0',
'm0', and two additional AbsMagnets 'm1' and 'm2' are used as control
magnets for BSnake 'n1' (yellow). Use of the TanMagnet as the second
control point of 'n1' has made 'n1' durably tangent to 'n0' at 'r0'.
SubSnake 'n2' (bright green) is the portion of 'n0' from t=0 to 'r0'.
SubSnake 'n2' and BSnake 'n1' are assembled into PolySnake2 'n3' (bright
9
blue). Breakpoints command reveals only 1 degree-2 breakpoint for 'n3',
at the location of 'r0'. This shows that an accurate tangency has been
achieved between its component snakes 'n1' and 'n2'.
On layer 1, an image of the u,v parameter space has been constructed

using a 4-point BSurf 's0' as the parameter square, and CopySnakes 'N0'
and 'N1', CopyRing 'R0', and CopyMagnets 'M1' and 'M2' as images of
the corresponding 3-D objects.
XYZ Magnet
Characteristic Data
Type (1 to 3)
Surface or magnet
Frame
Point
Location1, Location 2
Description Type determines whether you are making the YZ, ZX, or XY variety
Type-1 locates the magnet by its Y and Z locations
Type-2 locates the magnet by its Z and X locations
Type-3 locates the magnet by its X and Y locations.
Note these designations are based on an “odd-man-out” rule:
Type-1 is the one that doesn’t use X
Type-2 is the one that doesn’t use Y
Type-3 is the one that doesn’t use Z
Surface or magnet designates the host surface for the XYZMagnet. If a

magnet is used here, it also provides a starting location for the iterative
solution. Unless there is clearly only one solution (considering also the
extension of the surface beyond the 0 to 1 nominal parameter range), it is
always preferable to specify a magnet that is somewhere in the
neighborhood of the intended intersection. If you just specify the surface,
the u = 0.5, v = 0.5 position on the surface will be used as a starting value
for the XYZMagnet. If the true intersection is far from 0.5, 0.5, the
XYZMagnet may fail to converge, or may go to an unexpected location
on the surface extension.
Frame specifies the coordinate frame in which to measure X, Y or Z. The

default
10
frame ‘*’ is the global coordinate system, in which case the coordinates
used are just
the global X, Y or Z. If you choose to measure offsets or location in
another Frame not aligned with the global coordinates, Location1 and
Location2 will be frame coordinates, x, y or z.
Point is an optional reference point. If you choose the default point ‘*’,
the Locations
will be measured from the origin of Frame. This creates the “absolute”
variety of
XYZMagnet. If you choose another point for Point, the Locations will be
relative to
that Point. This makes the “relative” variety of XYZMagnet.
Location1, Location2 are the two coordinate offsets.
Example Suppose you need to locate a point ‘m1’ on the deck surface, exactly 3.20
ft. from centerplane and 10.46 ft. aft of the forward reference station. The
most direct way to do that now is an XYZMagnet. Since X and Y are
specified, the type has to be 3. Use ‘*’ for both Frame and Point
(coordinates measured in the global coordinate system), and 10.46 and
3.20 for the Locations.
Suppose you need another point ‘m2’ on the deck, exactly 2.25 ft. aft of
‘m1’ and the same distance from the centerplane. The simplest way to do
this is an XYZMagnet, again type-3 since X and Y are specified, with
Frame = ‘*’, Point = ‘m1’, and Locations of 2.25 and 0.
Sample file: XYZ MAGNET.MS2
Proximity Points
Proximity Bead, Ring, Magnet
Characteristic Data
ProxBead
type (kind), 0|1 = minimum/maximum signed distance
type (kind), 0|1 = unconstrained/constrained
bead/curve
mirror/surface
ProxRing
11
ring/snake
mirror/surface
ProxMagnet
magnet/surface
mirror/surface
Description The ProxBead, ProxRing and ProxMagnet entity types solve the
following problems:
ProxBead (ProxRing):
(Type 0) Find the point on a curve (snake) that has the minimum
signed distance from a mirror/surface.
(Type 1) Find the point on a curve (snake) that has the maximum
signed distance from a mirror/surface.
ProxMagnet
(Type 0) Find the point on a surface that has the minimum signed
distance from a mirror/surface.
(Type 1) Find the point on a surface that has the maximum signed
distance from a mirror/surface.
"Signed distance" is a measure of the position of a point in relation

to a mirror or surface, that takes the orientation of the mirror or
surface into account. In the case of a point or line mirror, signed
distance is always positive, and increases outward from the mirror.
In the case of a plane mirror, signed distance is zero on the plane,
positive on one side and negative on the other. In the case of a
surface mirror, signed distance is zero on the surface, positive in the
direction of the positive normal and negative in the other direction.
Constrained/ Unconstrained
There is an issue of whether the ProxBead is constrained to the
interval [0,1], or is free to go onto the extensions of the curve;
similarly for a ProxMagnet, whether it can use the extension of the
surface. Both alternatives appear useful in different circumstances,
so both are provided by means of a second type: 0 for unconstrained,
1 for constrained.If a bead (ring, magnet) is given for the bead/curve
12
(ring/snake, magnet/surface), MultiSurf uses it for the starting
location, and searches for a nearby local minimum or maximum.
Otherwise, MultiSurf starts with a tabulation of the host curve
(snake, surface) and identifies the closest/farthest point in the table
as a startinglocation for the search.
Associated errors
(1) Failed to converge.
(2) No solution exists. This will be a fairly common occurrence; say,
from choosing the wrong type. On a sailboat hull (unconstrained),
the maximum Y point will exist and be well defined, but in general
there is no minimum Y -- the extended surface goes on to minus
infinity. The optimization will be trying to run to minus infinity, and
we have to detect that and stop it before the numbers get outrageous.
Sample files
ProxBead1.ms2 This model contains 4 ProxBeads hosted on a circle.
'e1' is at the closest point on the circle to Point 'p3', and 'e2' is the
point farthest away from 'p3'. 'e1' and 'e2' both use the circle as
parent, so do not specify a starting point. 'e3' is the point closest to
Point 'p4', and 'e4' is the point farthest away from Point 'p5'. 'e3' and
'e4' both use AbsBead 'e0' as a starting point.
ProxBead3.ms2 This model contains 3 ProxBeads hosted on a

semicircle, each constrained to the 0-to-1 parameter interval (their
second type is 1). 'e1' is the point on 'c0' that is closest to Point 'p3';
'e2' is the point that is farthest away from 'p3'. 'e3' is also at a local
maximum distance from 'p3'; its parent is AbsBead 'e0', providing a
starting position for the search, which therefore finds the local rather
than global maximum. If 'e0' is moved to the opposite side of 'e1',
then 'e3' jumps to the t=0 end of 'c0'.
ProxRing1.ms2 This model has 2 ProxRings hosted on a BSnake.

Each is at the minimum distance from a magnet. 'r1' uses the snake
as parent; 'r2' uses AbsRing 'r0', which gives it a starting location for
the search.
ProxMagnet1.ms2 This model has a single ProxMagnet 'm1', used to

locate the maximum beam location on a boat hull. The mirror is the
predefined plane *Y=0, whose positive orientation is toward positive
Y. Therefore the ProxMagnet's first type is 1, seeking a maximum
signed distance.
13
MaxBeam.ms2 In this model a construction, using a Proximity Ring,
is set up to find a curve showing the maximum beam on a vessel’s
topsides. The Proximity Ring ‘Prox1’ draws out a Procedural Snake
caused by the action of moving its parent snake from sheer to chine.
The Parents of the Procedural Snake are the Proximity Ring and the
Ring ‘MovingPoint’, which is the parent of the ProxRing’s host UV
Snake.
Curves and Snakes

B-Fitted Curve
Characteristic data type = B-spline type

nc-specifier = integer; specifies choices for number of control points to use
for fit (0 if free)
log-tolerance = log (base 10) of tolerance
curve = basis curve
Description A BFitCurve is a uniform B-spline curve, least-squares fitted to the
tabulated data of its basis curve. Its principal function is for exporting
models to NURBS-based systems. Creating BFitCurves before you File/
Export 3D/ IGES gives you more control over the accuracy of the
exported model:
• you can see the NURBS approximation results right in MultiSurf,
prior to exporting IGES
• you can specify type and number of control points for NURBS curve
approximation
• you can specify different tolerances for different curves
log-tolerance is the tolerance specified with a base-10 logarithm, e.g. -3 for
tolerance of 10-3. This notation permits a wide dynamic range. (Tolerance
itself is measured as the RMS (root-mean-squared) deviation of the fitted
curve from tabulated points of the basis curve.)
type is the type of B-spline : 1 = linear, 2 = quadratic, 3 = cubic, etc.
nc-specifier is an integer value which specifies choices for the number of
control points to be used. There are two choices:
When you specify a positive value (it must be greater than the
corresponding type), the program will use exactly this number of
control points for the fit.
When you specify zero, the program will start at type +1 control
points and cycle nc upward until it either achieves the fit within
the tolerance or reaches the limit of 128 control points.
If the BFitCurve does not meet the specified tolerance, the Error window
will pop up and let you know the problem. Double-Clicking the Entity
14
gives you quick access to the Property Manager, where you can try a
different set of values.
When the BFitCurve is displayed, you can see where the control points
are by turning on the "polygon" (visibility = polygon) for the curve − the
control points are located where the polygon lines meet
If you would like MultiSurf to report information about the quality of fit,
select the BFitCurve and then choose Tools/ Clearance.
Example bfitcv1.ms2
This example is a cubic (type-3) B-spline curve fitted to one quadrant of
an ellipse basis curve. Since nc-specifier is 0, the program incremented the
number of control points until the tolerance (1E-4 = .0001) was met. The
fit was accomplished with 8 control points.
Z
polygon for BFitCurve
BFitCurve and basis curve
The deviation of the fitted curve from the basis curve, as provided by Tools/
Clearance, is: rms = 6.80e-005.
See also Fitted Surface
Conic
Characteristic data center = point to locate center point of complete conic section (P)
point2 = point to locate end of primary axis (P2)
point3 = point to locate end of secondary axis (P3)
s0 = conic parameter (degrees) for the start (t = 0)
s1 = conic parameter (degrees) for the end (t = 1)
type = 1, 2, 3, or 4
type-1 = ellipse
type-2 = hyperbola, positive branch
type-3 = hyperbola, negative branch
type-4 = catenary
Description All types of conics are constructed by specifying the center (center), the
ends of the semimajor and semiminor axes (point2 and point3), and two
parameter values (s0 and s1). The conic curve always lies in the plane of
15
the three points. The end of the primary axis is at point2 (P2). A temporary
point Q is created for the end of the secondary axis by moving from P2
parallel to the primary axis as required to make the secondary axis
perpendicular to the primary axis. Thus the user is freed from locating
the three points such that P2-P-P3 is a right angle. The length of the
secondary axis is the distance of P3 from the primary axis, also the
distance from P to Q.
For a precise description of the conic curves we define the three vectors
using bold face letters:
a is the vector from P to P2 (primary axis)
b is the vector from P to Q (secondary axis)
c is the vector from the origin to P
a, b, and c are their magnitudes, respectively.
Ellipse (type-1)
For an ellipse, the primary axis can be either the semimajor or semiminor
axis, depending on the relative distances of P2 and Q from P. The full
ellipse passes through points P2 (at s = 0) and Q (at s = 90 deg.). The
equation for the ellipse (type-1) is:
x(t) = a cos s + b sin s + c
s = s0 * (1 - t) + s1 * t
The ellipse foci are at c +/- a SQR(1 - b2 / a2) .
Hyperbola (type-2 and type-3)

For a hyperbola, the primary axis is always the semimajor axis. The full
hyperbola passes through point P2 (at s = 0).
The equation for the positive hyperbola (type-2) is:
x(t) = a cosh s + b sinh s + c
s = s0 * (1 - t) + s1 * t
The equation for the negative hyperbola branch (type-3) is:
x(t) = - a cosh s - b sinh s + c
s = s0 * (1 - t) + s1 * t
The hyperbola foci are at c +/- a SQR (1 + b2 / a2) . The
asymptotes are the two lines:
x = c + r * (a +/- b)
where r is a real parameter.
Catenary (type-4)
Although not actually a conic section, the catenary is a related analytic
curve which is easily supported in the same context. Its equation is:
x(t) = a cosh s + b s + c
16
s = s0 * (1 - t) + s1 * t
(s in radians). If s0 and s1 are fairly small, the curve is very similar in
shape to the positive hyperbola.
The catenary is the equilibrium curve assumed by a hanging chain or
cable under its own weight, with the force of gravity acting in the -a
direction. It has engineering applications in the design of bridges and
arches.
For all types of Conic Sections, relabel is used to relabel the curve. The
program’s default relabel ‘∗’ produces the “natural” labeling, in which the
parameter t is linearly related as indicated in the formulas above.
Note that the parabola is also a conic section. It can be regarded as the
limiting case of either an ellipse or a hyperbola as the center moves to
infinity. Parabolas can be created in MultiSurf in two ways — any type-2
B-spline or C-spline curve with three control points is a parabola.
Example 1 conics.ms2 (ellipse and hyperbola)
Z
t=1 t=1
'minor'
'hyperbola_neg' 'hyperbola_pos'
= type-3 t=1 = type-2 conic section
conic section
'major'
X
'center'
t=0
'ellipse'
= type-1
t=0 conic t=0
section
CONIC.MS2 in <Y> view; Conic types 1 - 3.
All three of the conic sections in this example are formed using the same
control points and conic parameters:
‘center’ is the center point
‘major’ is point2, the end of the primary axis
‘minor’ is point3, the end of the secondary axis
s0, the conic parameter for the start (t = 0) of the curve is -90
s1, the conic parameter for the end (t = 1) of the curve is 90
‘ellipse’ is the type-1 conic section, ‘hyperbola_pos’ is the type-2 conic
section, and ‘hyperbola_neg’ is the type-3 conic section.
17
Example 2 catenary.ms2 (catenary)
‘catenary’ is a type-4 conic section formed by the following three control
points and two conic parameters:
‘center’ is the center point
‘contact’ is point2, the end of the primary axis
‘minor’ is point3, the end of the secondary axis
s0, the conic parameter for the start (t = 0) of the curve is 0
s1, the conic parameter for the end (t = 1) of the curve is 90
In this example, ‘bottom’ is a line depicting the bottom of the bay (or lake
or ...). ‘catenary’ is a type-4 conic; it is the portion of anchor chain not
resting on the bottom. ‘chain’ is a Line; it is the portion of anchor chain
that is resting on the bottom. ‘boat’ is a bead at the t=1 end of ‘catenary’.
Z
'boat'
at t=1
X
'catenary'
'chain'
'bottom'
'contact'
'anchor'
at t=0
'center' 'minor'
CATENARY.MS2 in <Y> view.
See also Arc, ArcSnake, BCurve, CCurve
Expanded Curve
Characteristic data
Relabel
Type (-3 to -1, 1 to 3) -- Specifies which axis of the frame to use, and
which direction along that axis (+/-) corresponds to the positive t
direction.
Bead/curve -- Specifies the basis curve; if a bead, it also specifies a
particular point on the curve to end up at the frame origin.
Frame
Scale factor
Description An ExpdCurve (Expanded curve) is to a curve as a development
(flattening) is to a developable surface: It is a copy of the curve, rolled
18
out flat to make a straight line, with preservation of arc-length of all
elements.
Note that this construction implies that, in absence of a relabel, the
velocity profile of an ExpdCurve will be identical to that of its basis
curve.
One application is the layout of structural elements that will be formed
by bending straight tubes, shapes or extrusions into curves. In this case,
the 3-D basis curve should follow the neutral axis of the shape.
Sample files ExpdCurve1.ms2
This is the forward master curve of Demo.MS2, flattened into a vertical
line. Point 'p0' is used as the frame. The type is -3, therefore the positive-t
direction on the ExpdCurve is the negative z axis, i.e., downward. A
bead 'e0' at t = 0.257 specifies which point of the ExpdCurve should be at
the origin.
NURBS Curve
Characteristic data point1 ... pointN = control points

wt1 ... wtN = corresponding weights
knotlist = name of a KnotList object or ‘∗’ for uniform knot spacing
type = B-spline type: 1 = linear, 2 = quadratic, 3 = cubic, etc.
Description The NURBCurve is a non-uniform rational B-spline curve. The control
points guide or shape the surface in much the same way as the control
points shape a B-spline Curve. In general, the curve does not interpolate
any of its control points except the first and last (point1 and pointN).
wt1, wt2, etc. are the weights associated with each control point.
knotlist controls the knot spacing. If not default ‘∗’ (uniform spacing), the
knotlist should have (number of control points) - (type) + 1 components,
and the list should begin with 0 and end with 1.
The program’s default relabel produces the “natural” labeling, in which
the parameter t is concentrated in highly curved areas formed either by
closely spaced control points or by high weights.
Example nurbcrv1.ms2
Z Z Z
NURB curve
Y Y Y
'p2' wt=1 'p2' wt=2 'p2' wt=5

polyline
NURBCRV1.MS2 in <x> view: initial model (left); ‘p2’ weight edited (middle and right).
19
‘midsection’ is a NURBCurve formed by three points ‘p1’, ‘p2’, and ‘p3’,
all of which have wt = 1 (left figure).
In the middle figure, the weight for ‘p2’ has been changed from 1 to 2,
pulling the curve deeper. In the right-hand figure, the weight for ‘p2’ has
been increased to 5, giving ‘midsection’ a tight bilge curve.
See also BCurve, NURBSnake, NURBSurf, KnotList, BSurf.
Radius Arc
Characteristic data point1, poin2, point3 = control points

radius = radius of arc
type = kind of radius arc
Description The RadiusArc is a circular arc of a specified radius. It is defined by three
control points and a radius, and can be built in several different ways, as
designated by type.
radius is a decimal number, positive or negative. For type-1 and type-2,
radius must be less in absolute value than twice the distance from point1
to point3.
type specifies how the radius arc is built from the control points:
type-1 The arc runs from point1 to point3, in the plane of the 3
points; type-1 is the shorter arc, with angle < 180 deg.
type-2 The arc runs from point1 to point3, in the plane of the 3
points; type-2 is the longer arc, with angle > 180 deg.
type-3 This arc rounds the corner formed at point2 by the directions
to point1 and point3. Its t=0 end is tangent to the line from
point1 to point2; its t=1 end is tangent to the line from point2
to point3.
For type-1 and type-2 RadiusArcs, note that in a given plane, there
are four possible arcs of a given radius between two points:
• type-1, positive radius
• type-1, negative radius
• type-2, positive radius
• type-2, negative radius
Z Z
'Type2_PosRadius'
'Type1_PosRadius' 'p3'
'p2' 'p2'
'p3' 'p1'
X X
'p1'
'Type1_NegRadius'
'Type2_NegRadius'
20
RadiusArc types 1 and 2: four possible arcs between two points.
The program’s default relabel produces uniform labeling with respect to

arc length or angle.
Example 1 radiusarc1.ms2
Example: radiusarc1.ms2
This is an example of a constant radius deck made by sweeping a
RadiusArc to generate a ProcCvSurf (Procedural Curve Surface).
‘deck_radius’ (red) is a type-1 RadiusArc. It has a radius of 20 and is
built between:
‘e0’ (bright green; an AbsBead on ‘sheer’) = point1 for the arc
and
‘p0’ (cyan; a MirrPoint of ‘e0’ across the centerplane) = point3 for the arc.
‘p1’ (yellow; a RelPoint off ‘e0’, at dX=0, dY=0, dZ= .1) is point2 for the
arc. It orients the arc in a plane parallel to the X=0 plane.
The deck is a ProcCvSurf for which the moving curve is ‘deck_radius’ and
bead = ‘e0’. To generate the ProcCvSurf, MultiSurf varies t for ‘e0’ from 0
to 1, and the ProcCvSurf is the surface swept out by ‘deck_radius’.
Z
RadiusArc (type-1)
'p0'
X
'p1'
'e0'
RADIUSARC1.MS2 at Lat -20, Lon 30.
Example 2 radiusarc3-polycurve2.ms2
Example 2: radiusarc3-polycurve2.ms2
This model is a hard-chine barge with raked ends. A type-3 RadiusArc is
used to form the 20-foot-radius curve in the chine.
‘radiusarc’ (bright green) is a type-3 RadiusArc made from ‘vert_rake’
(blue, a bead on ‘l2’), ‘long_rake’ (blue, a bead on ‘l0’), and ‘p2’ (cyan, a
ProjPoint of a point at the origin). ‘radiusarc’ is tangent to ‘l3’ (yellow) at
its t=0 end and to ‘l1’ (magenta) at its t=1 end.
‘chine’ (cyan) is a PolyCurve2 made from ‘l1’, ‘radiusarc’, and ‘l0’.
You can vary the longitudinal and vertical extents of the rake by
dragging ‘long_rake’ and vert_rake’ respectively. The RadiusArc adjusts
itself to maintain its specified radius of 20 ft and it remains tangent to the
lines ‘l1’ and ‘l3’.
21
Z
t=0 'vert_rake'
X 'l3'
RadiusArc
'p1' 'l1' t=1 'long_rake'
Y
RADIUSARC3-POLYCURVE2.MS2 at Lat 10, Lon 100.
For discussion of the PolyCurve2 in this model, see the end of the first
PolyCurve2 Entity Description example.
See also Arc
Tabulated Curve
Characteristic data filename = filename of .3DA or .PAT file

polyline number = number of polyline in file
Description filename is the filename of a .3DA (or .PAT) file. If no extension is given,
MultiSurf assumes .3DA.
A .3DA file (see sample contents and notes in example below) consists of
a series of records (file lines) each having four numbers: a pen number
and X, Y, Z coordinates. A zero pen means “move to this point without
drawing”. Therefore, if we let nlinks equal the number of divisions in a
polyline, a polyline with nlinks divisions is represented by one pen = 0
record followed by nlinks records with nonzero pen. The product of t-
divisions × t-subdivisions must be correctly set equal to nlinks, or a
geometry error occurs. Each 0-pen record is counted as the start of a new
polyline.
polyline no. is the number of the polyline to be taken from the .3DA file: 1
for the first, 2 for the second, etc.
coordinates are added to the polyline coordinates. This results in
a shift (but no rotation).
rotation.
relabel is used to relabel the curve. The program’s default relabel ‘∗’
produces a labeling which corresponds to the labeling of the original
22
polyline points; i.e., the first point is at t = 0; the second is at t = 1/nlinks,
the third is at 2/nlinks.... and the last is at t = 1.
Note: A TabCurve has less inherent smoothness than most native
MultiSurf curves. It is actually a polyline, like a type-1 BCurve. You
cannot subdivide a TabCurve to produce a smoother version of the same
curve.
Example tabcurve.ms2 (needs demomcs.3da)
This example produces Z
a model identical to Y
the DEMO model. tabulated curves
However, in this case, 'c3', 'c2', and 'c1'
instead of generating
the master curves
itself, the model reads
them in from a file
DEMOMCS.3DA (see 'e1'
contents below).
Therefore, the shape
of the hull depends X
entirely on the
contents of that file. TABCURVE.MS2 at Lat -30, Lon 60.
The three TabCurves are named ‘c1’, ‘c2’, and ‘c3’. For each of them,
filename is DEMOMCS, and frame/point is the default ‘*’ which produces
no shift or rotation of the curves. polyline no. for ‘c1’ is 1, for ‘c2’ is 2, and
for ‘c3’ is 3.
The bead ‘e1’ is included to show that a TabCurve, like any other curve,
will support beads and other objects depending on curves.
Contents of DEMOMCS.3DA
0 0.000 0.000 3.600 pen=0 point (1st point) on 1st
polyline ‘c1’
11 0.511 0.000 2.493 1st of 10 records with nonzero pen
for 1st polyline; nlinks for ‘c1’ = 10
11 0.951 0.000 1.569
11 1.321 0.000 0.829
11 1.619 0.000 0.273
11 1.845 0.000 -0.099
11 2.045 0.000 -0.355
11 2.260 0.000 -0.564
11 2.491 0.000 -0.724
11 2.738 0.000 -0.836
11 3.000 0.000 -0.900
0 15.000 4.815 2.560 pen=0 point (1st point) on 2nd
polyline ‘c2’
11 15.000 4.869 1.835
11 15.000 4.847 1.204
11 15.000 4.749 0.667
23
11 15.000 4.575 0.226
11 15.000 4.325 -0.121
11 15.000 3.921 -0.403
11 15.000 3.286 -0.649
11 15.000 2.421 -0.860
11 15.000 1.326 -1.035
11 15.000 0.000 -1.175
0 30.000 3.500 2.760 pen=0 point (1st point) on 3rd
polyline ‘c3’
11 30.000 3.480 2.218
11 30.000 3.420 1.742
11 30.000 3.320 1.334
11 30.000 3.180 0.994
11 30.000 3.000 0.720
11 30.000 2.720 0.504
11 30.000 2.280 0.336
11 30.000 1.680 0.216
11 30.000 0.920 0.144
11 30.000 0.000 0.120
See also TabPoint, TabSurf, WireFrame
X-Curve
Characteristic data point1, point2 ... pointN = control points

type = direction of orientation of XCurve
ecc = end condition code for the two directions other than the direction of
orientation (specified by type)
s/m1, s/m2, s/m3, s/m4 = slope or moment at each end of the curve for the
two directions other than the direction of orientation (specified by type)
Description The XCurve was created to add 100% forward compatibility from FL/2B
to MultiSurf. The XCurve, “explicit spline” or X-spline, is a cubic spline
passing through its data points, with knots at the interior data points. In
this respect, it is like a type-3 CCurve. However, there are several
important differences from a CCurve:
(1) The XCurve is an explicit function of either X, Y, or Z, according to
its type. Its natural or default t parameter is distributed in proportion
to distance along the specified axis.
(2) A consequence of the above is that an XCurve must be oriented with
respect to its selected axis, and must have that coordinate either
strictly increasing or strictly decreasing along its length; it isn’t
allowed to double back on itself.
(3) If you rotate the set of points that parent an XCurve about any line
not parallel to the selected axis, the curve changes shape, as well as
position. If you rotate the points far enough, the XCurve will cause a
geometry error.
24
(4) At each end of an XCurve, you have explicit control over either the
slope or the curvature (bending moment) in two directions. To be
more specific, suppose we have a type-1 XCurve, i.e., oriented with
respect to the X-axis. At each end of this curve, you will explicitly
specify either the slopes Y’, Z’ or the moments Y”, Z”.
(5) An XCurve will not in general lie in a plane, even if all its control
points do. There are somewhat intricate constraints on the end
condition values that are required to make the curve lie in a plane.
(6) The Y and Z projections of a type-1 XCurve are completely
independent. That is, the Y-projection depends only on the Z-
coordinates and the Z’ or Z” end condition values, and vice versa.
type specifies the direction of orientation of the XCurve:
1 for X orientation
2 for Y orientation
3 for Z orientation
Note: this usage of type has nothing to do with “spline” type as used in
BCurve, CCurve entity specifications. XCurves are always cubic splines.
ecc is the end condition code for the two directions other than the
direction of orientation (specified by type):
0 = moment both ends
1 = slope at t=0 end, moment at t=1 end
2 = moment at t=0 end, slope at t=1 end
3 = slope at both ends
s/m1, s/m2, s/m3, s/m4 are the slope or moment at each end of the curve
for the two directions other than the direction of orientation (specified by
type):
type-1 type-2 type-3

(X orientation) (Y orientation) (Z orientation)
s/m1 Y’ or Y” Z’ or Z” X’ or X” at t = 0 end
s/m2 Z’ or Z” X’ or X” Y’ or Y” at t = 0 end
s/m3 Y’ or Y” Z’ or Z” X’ or X” at t = 1 end
s/m4 Z’ or Z” X’ or X” Y’ or Y” at t = 1 end
While inclusion of the XCurve was motivated primarily by the desire for
full compatibility with FAIRLINE/2, it certainly has some other
worthwhile uses. For example... boat hull with straight frames...
cambered deck as ALoftSurf... accurately flat spray chines...
Example xcurve.ms2
This example is a candidate sheerline for an 8-meter sailing yacht. You’ll
probably want to use Edit/ Model File to follow the discussion. The
XCurve is type-1 (oriented to the x-axis), and has end condition code = 1
(slope at t=0 end, moment at t=1 end). It passes through the five points in
sequence. At t=0, the Y-slope (dY/dX) is 0.5, and the Z-slope (dZ/dX) is
25
-0.13. At t=1, the Y-moment (d2Y/dX2) is -.02, and the Z-moment
(d2Z/dX2) is .005.
Z
Y
type-1 X-spline curve
'p1'
'p2' t=0
'p3'
'p4'
'p5'
t=1
XCURVE.MS2 at Lat -30, Lon 60.

See also XLoftSurf
Arc Snake
Characteristic data magnet1, magnet2, magnet3 = magnet or ring control points

type = 1 - 6
Description An Arc Snake is a continuous curve lying in a surface and defined by 3
magnets, which must all lie in the same surface. In the u-v parameter
space, the Arc Snake is a circular arc. There are 6 types:
type-1 arc (through 3 magnets) — the arc passes through all three
magnets
type-2 arc (start-center-guide) — starts at magnet1, uses magnet2 for
its center, and ends (at the correct radius) on a line from the
center through magnet3
type-3 circle — the arc is a complete circle starting at magnet1 and
using magnet2 for its center; magnet3 has no role except to set
the direction of the circle; magnet3 is used to set the direction
of the circle — the closest point on the circle to magnet3 is at
a parameter value t that is less than 0.5
type-4 arc (tangent at start) — the arc starts at magnet1, uses
magnet2 for the middle point of its polygon, ends at magnet3,
and is tangent to its polygon at the t=0 end
type-5 arc (tangent at end) — the arc starts at magnet1, uses magnet2
for the middle point of its polygon, ends at magnet3, and is
tangent to its polygon at the t=1 end
type-6 semi-circle — the arc is a semi-circle starting at magnet1 and
using magnet2 as its center. magnet3 determines which of the
two possible semi-circles you get — the one that passes
closest to magnet3
26
The program’s default relabel produces the “natural” labeling, in which
the parameter t is uniformly distributed with respect to arc length or
angle, in the u-v parameter space.
Example arcsnake.ms2
‘garter’ (red), a type-3 arc snake, is shaped by the three magnets ‘m1’,
‘m2’, and ‘center’ (all cyan). ‘garter’ is a circle in the u-v parameter space
mapped onto a rectangle ‘patch’. This produces a true ellipse. If the
shape of ‘patch’ is changed, ‘garter’ will change its position in space but
will remain always on ‘patch’, in a corresponding position. ‘garter’ will
remain an ellipse as long as ‘patch’ remains a rectangle.
Z
'center'
type-3
arc snake
X u
0,0 u 0,0
'm1' 'm2'
t=0 and t=1
ARCSNAKE.MS2 in <Y> view.
See also Arc
Foil Snake
Characteristic data magnet1 ... magnetN = control points

type = type of airfoil family, 1 to 5; types 101 - 255 reserved for user-
defined foils
Description A Foil Snake is a true NACA airfoil section or user-defined foil section in
the u-v parameter plane. This maps into a foil shape lying in the surface
its 3 or 5 control point magnets are attached to. If the net of u-v
parameter lines in the vicinity is roughly rectangular, the Foil Snake will
be close to a true NACA section or user-defined foil in space.
Foils can be constructed with either 3, 4, or 5 control magnets — 3 for a half section, 4 for a symmetric full
section, or 5 for a cambered or symmetric full section. In all cases:
• The snake lies in the same surface as its magnets.

• magnet1 is the trailing edge.
• magnet3 is the leading edge.
• The “chord line” is the Line Snake joining magnet1 and magnet3.
• point2, and point4 of a 5-point foil, determine thickness and camber.
27
• The last control point in a 4- or 5-point foil nominally closes the foil
(usually same point as point1).
type:
type-1 NACA 4-digit series; max camber at 40%
type-2 NACA 63 series with a=0.3 mean line
type-5 NACA 0010-34; max camber at 40%
types 101-255 user-defined foils
relabel is used to relabel the snake. The default relabel ‘∗’ produces a
labeling in which the parameter t is concentrated in the highly curved
leading edge area, in the u-v parameter space. Subdivision control may
be important if you are creating panels for a potential-flow code.
Notice The implementation of FSnake may be changed in a future version. We
have found FSnakes to be of marginal value because of the distortion of
the foil shape that occurs when the true foil outline is expressed in u, v
parameters and mapped onto the distorted mesh of u, v grid lines. We
now see the possibility of compensating for this distortion, producing a
much truer rendition of the foil.
The ProjSnake entity provides a way to overcome this problem: instead
of using an FSnake, use a ProjSnake made from an FCurve. For example,
in the FSNAKE2.MS2 example below, you could make an FCurve lying
in a horizontal plane such as Z = 0, then project it vertically onto the hull
surface.
Example fsnake2.ms2
‘hull’ is a C-lofted surface made from 5 master curves ‘MC1’ to ‘MC5’. u
= 0 is the sheerline, and v = 0 is ‘MC1’.
‘trailing_edge’, ‘side’, and ‘leading_edge’ are three magnets on ‘hull’.
‘trailing_edge’ and ‘leading_edge’ are located on the u = 1 edge of the
surface.
‘keel_root’ is a type-1 half-section foil curve, defined by the three
magnets and lying on ‘hull’.
28
v
Z 1,1
Y
half-section foil snake 0,0 'trailing_edge'
v
'leading_edge'
u u
0,0
'leading_edge'
t=1
'side'
X
Y
'trailing_edge'
t=0
X
Z
FSNAKE2.MS2 at Lat -30, Lon 50 (top) and Lat 90, Lon -90 (bottom).
If ‘hull’ is changed by modification of one or more of the master curves,

‘keel_root’ will change its position in space but will remain always on
‘hull’, in a corresponding position.
See also FCurve
Geodesic Snake
Characteristic data curvature = a geodesic curvature value (defined below)

graph = name of a BGraph object or ‘*’ for default graph
log-tolerance = logarithm (base 10) of tolerance for matching prescribed
curvature
magnet1 locates the t=0 end of the snake
magnetN locates the t=1 end of the snake
Description A Geodesic Snake is a curve embedded in a surface, with a specified
distribution of geodesic curvature along its length (for definition of
geodesic curvature, see “Geodesic Curvature” below). The geodesic
curvature can be zero (constant) in which case the GeoSnake is a
geodesic on the host surface.
Note that calculation of a GeoSnake is a computationally intensive
searching process, without guarantee of success (details below in
“Failure to converge”). Therefore, unless your task really requires
accurate control over geodesic curvature, we recommend that you use a
simpler snake entity, such as a LineSnake or BSnake.
A GeoSnake is defined by 2 or more magnets which all must lie in the
same surface. The program uses a type-2 BSnake made from the magnets
as its starting path for computing the GeoSnake. If you give only 2
29
magnets, start and end, the type-2 BSnake is just the LineSnake from
magnet1 to magnet2. For many GeoSnake applications, especially those
with zero curvature, two magnets will be sufficient. If the resulting
GeoSnake fails to converge, you can add one or more intermediate
magnets to get a better starting configuration.
A GeoSnake can have zero (constant) geodesic curvature for its entire
length, or it can have a specified distribution of geodesic curvature.
curvature and graph are the controls:
curvature is a geodesic curvature value, units = 1/length. 0 curvature
makes an actual geodesic satisfying Geodesic conditions (1), (2) below.
graph modulates the geodesic curvature; i.e., multiplies the curvature
value. The default graph ‘*’ is interpreted as a constant, 1.
log-tolerance allows you to specify a tolerance for the convergence of the
iterative search MultiSurf uses to compute the GeoSnake. The value is
specified as the log (base 10) of tolerance (using a logarithm here allows
a wide dynamic range). Example: log-tolerance = -4 means tolerance = 10-4
= .0001. log-tolerance is provided as a control because practical accuracy
requirements for geodesics vary widely between applications.
Tolerance itself is the largest tolerable deviation of calculated geodesic
curvature from specified geodesic curvature, multiplied by the total arc
length of the GeoSnake (this makes it non-dimensional). Example: a
curve with 10-meter arc length and maximum curvature of 10-3/meter
(curvature x arc length = 10-4) could deviate from a straight line by no
more than 0.125 mm.
relabel is used to relabel the snake. The default relabel ‘*’ produces the
“natural” relabeling of the snake which is uniform with respect to arc
length.
Geodesic curvature Geodesic curvature is a pointwise scalar property of a curve lying in a
surface, determined as follows:
(1) At any point along the curve, construct the plane tangent to the
surface.
(2) Project the local portion of the curve normally onto the tangent
plane.
(3) Geodesic curvature is the curvature of the projection.
Geodesic A geodesic is a curve embedded in a surface, with a special property
which can be expressed in either of two mathematically equivalent ways:
(1) A geodesic has zero geodesic curvature throughout its length.
(2) A geodesic is the shortest curve in the surface joining its endpoints.
Failure to converge The typical way a GeoSnake fails to evaluate is “failure to converge.”
Features which can promote this problem are:
• coarse subdivisions of the host surface
• irregular mesh on the host surface
• lack of smoothness of the host surface
• passing through or close to a coordinate singularity
• starting path far from the geodesic path
• tight tolerance
30
Usage Geodesics have significant applications in manufacturing:
(1) Ribbands — they must lie along geodesic paths if they are to
conform to the hull surface between frames (see “Example”).
(2) Planked wooden boat construction — geodesic seams produce
straight planks.
(3) Cold-molded plywood construction — veneers have to lie along
geodesic paths. (Take a look at VENEERS.MS2 which shows a hull in
the process of veneer layout. After making the first veneer-edge
GeoSnake, we saved it as a component, then loaded it successively,
moving each new GeoSnake to its proper location by moving its
AbsRing control. Then we’d make DevSurfs (the veneers) between
adjacent pairs of GeoSnakes.)
(4) Sail making — geodesic seams are key to efficient use of cloth and
flat panel development. (Take a look at SAILSEAM.MS2 which
shows a jib in the process of being seamed; eventually, DevSurfs
would be created between each adjacent pair of seams.)
(5) Plate expansion — a mesh of geodesic triangles is key step in the
process.
(6) Compounding by line heating — using geodesics for heat lines
minimizes residual stress.
(7) Automated lamination — reinforcing tape has to be laid along
geodesic paths, or it wrinkles and buckles.
Example ribbands.ms2
Z
Y
GeoSnakes 'm0' v 0,0
u
'm1'
t=0
X
t=1
In this example, GeoSnakes are used to layout the paths for ribbands.
The use of geodesics makes ribbands that will naturally lie flat on the
frames and conform to the hull surface between frames.
‘m0’ is an AbsMagnet on the v = 0 edge of the hull; ‘m1’ is a RelMagnet
on the v = 1 edge of the hull (at du = 0, dv = 1 from ‘m0’). ‘n0’ is the
GeoSnake between ‘m0’ and ‘m1’.
To make the set of ribbands, we saved this first ribband as a component
(objects ‘m0’, ‘m1’, and ‘n0’; RIBBAND.MC2), then loaded the
component successively to make the other ribbands. Since we made the
initial ribband so that it can be moved by moving just the magnet ‘m0’, it
is easy to move each component ribband to its own position on the hull.
31
Note that where the hull has relatively high curvature at the turn of the
bilge, we had to play with the position of the control magnet to get the
geodesic we wanted. In this region, a small movement up or down of
‘m0’ makes a big difference in the geodesic path.
See also LineSnake
NURBS Snake
Characteristic data magnet1 ... magnetN = control points and weights

wt1 ... wtN = corresponding weights
knotlist = name of a KnotList object or ‘*’ for uniform knot spacing
type = B-spline type: 1 = linear, 2 = quadratic, 3 = cubic, etc.
Description A NURBSnake is a continuous curve lying in a surface and defined by a
series of magnets for control points. The magnets must all lie in the same
surface. The curve ends at the two end control points (magnet1 and
magnetN), but in general does not pass through the others (unless snake
is type-1). Instead, it is “molded” or “shaped” by them. In the u-v
parameter space, the NURBSnake is a true NURBCurve.
wt1, wt2, etc. are the weights associated with each control point.
knotlist controls the knot spacing. If not default ‘∗’ (uniform spacing), the
knotlist should have (number of control points) - (type) + 1 components,
and the list should begin with 0 and end with 1.
The program’s default relabel ‘∗’ produces the “natural” labeling, in
which the parameter t is concentrated in highly curved areas formed
either by closely spaced control points or by high weights, in the u-v
parameter space.
Example nurbsnk1.ms2
NURB snake v 'csm2'

u 0,0
'csm1'
t v
u
'csm1' 0,0
'csm2'
t=0 and t=1
v 'csm2'
'csm7' wt=3
u 0,0
v
u
0,0
'csm2' wt=3 'csm7'
NURBSNK1.MS2 in Lat 0, Lon 60 orthographic view: initial model (top); with two weights edited
(bottom).
32
‘cabin_side’ is the side of a boat cabin. ‘portlight’ is a type-2 NURBSnake
formed by the seven magnets ‘csm1’ to ‘csm7’ which all lie on
‘cabin_side’ and all have wt = 1 (top figure). If the shape of ‘cabin_side’ is
modified, ‘portlight’ will change its position in space but will remain
always on ‘cabin_side’, in a corresponding position.
The lower figure shows the effect of increasing the weights of ‘csm2’ and
‘csm7’ to 3.
See also BCurve, NURBCurve, NURBSurf, KnotList, BSurf
PolySnake
Characteristic data snake1, snake2 .... snakeN = component snakes (t-orientation doesn’t
matter)
Description A PolySnake is a single curve made by joining two or more other snakes
(called the “component snakes”) end-to-end. The component snakes
must all lie in the same surface and be selected in the order in which they
are to be assembled.
It is anticipated that the end of snake1 is actually the same point as one of
the ends of snake2, etc. However, this is not enforced or checked. If you
specify snakes that do not share endpoints, the moving point will jump
discontinuously from the end of snake1 to the start of snake2 as the
parameter t passes t1 , etc. (see next paragraph).
relabel is used to relabel the snake. The program’s default relabel ‘∗’
produces the “natural” labeling, in which the parameter t is t1 at the end
of the first component snake, t2 at the end of the second component
snake, etc., in the u-v parameter space.
Example polysnk1.ms2
‘t_reversed’ (cyan) is the same subsnake we created in the second
example for the “SubSnake” entity description. ‘t_reversed’ begins (t = 0)
at ‘end1’ and ends (t = 1) at ‘end2’. ‘lsnake’ (yellow) is a line snake that
begins at magnet ‘M1’ and ends at magnet ‘M4’. Both ‘t_reversed’ and
‘lsnake’ lie in the surface ‘patch’. ‘end2’ (t = 1 for ‘t_reversed’) and ‘M1’ (t
= 0 for ‘lsnake’) are at the same position. ‘polysnake’ (magenta) is formed
by joining ‘t_reversed’ and ‘lsnake’. t = 0 for ‘polysnake’ is at ‘end1’, t =
.5 is at the end of ‘t_reversed’ (‘end2’), and t = 1 is at the end of ‘lsnake’
(‘M4’).
In the lower figure, the shape of ‘patch’ has been changed, but you can
see that ‘polysnake’ still lies in the surface.
33
'end1'
polysnake t=0
v 'M4'
'M4'
polysnake t=1 v 'end2'
& 'M1'
0,0
u
u 0,0
'end1'
'end2' & 'M1'

polysnake t=.5 v
0,0
u
0,0
u
POLYSNK1.MS2 at Lat -60, Lon 60: initial model (top); with ‘patch’ surface edited (bottom).
See also PolyCurve, SubSnake
NUBFitCurve, NUBFitSnake, NUBFitSurf entities

The “NUBFit” entities are essentially replacements for the “BFit” entities.
BFitCurve, BFitSnake and BFitSurf — with substantially greater
flexibility. You can think of the “NU” as standing for “non-uniform”, or
for “New”. The BFits still exist for backwards compatibility, but the
NUBFits do everything the BFits did, and much more.
NUBFitCurve
Characteristic data Type
Knotlist
nc-specifier
Log-tolerance
Curve
34
Description Compared with a BFitCurve, the only thing new here is the Knotlist. There are 3
alternatives for this entry:
1. ‘*’ (the “default” knotlist) - This allows the fitting to use the best set
of nonuniform knots it can come up with.
2. ‘*UNIFORM’ - a new “predefined” knotlist, available in the System
Objects: this forces the fitting to use uniform knots, essentially
duplicating the function of BFitCurve.
3. You can choose a KnotList or KnotList2 object containing a suitable
specified list of knots.
As in the BFitCurve, for nc-specifier, there are two basic choices:
A nc-specifier can be a positive integer (greater than type), specifying a
definite number of control points to use for the fit.
B nc-specifier can be 0, which signifies “increase the number of control
points until tolerance is met”.
Since there are 3 alternatives for Knotlist and 2 alternatives for nc-
specifier, altogether there are 6 combinations which produce different
behaviors for the
NUBFitCurve entity:
1A. ‘*’ knotlist, combined with positive nc-specifier: The fitting will use
exactly nc-specifier control points, and will use the best non-uniform
knots it can generate.
1B. ‘*’ knotlist, combined with nc-specifier = 0: The fitting will increase
the number of control points (from type+1 up to the number of data
points), until tolerance is met. At each step, it will optimize the non-
uniform knotlist.
2A. ‘*UNIFORM’ knotlist, combined with positive nc-specifier: The
fitting will use exactly nc-specifier control points and uniform knots.
2B. ‘*UNIFORM’ knotlist, combined with nc-specifier = 0: The fitting will
increase the number of control points (from type+1 up to the number of
data points), until tolerance is met. At each step, it will use uniform
knots.
3A. KnotList object, combined with positive nc-specifier: The fitting will
use exactly nc-specifier control points, along with the specified KnotList.
In this case, there is a required relationship between nc-specifier, type,
and number of knots: the number of interior knots (i.e., knots which are
> 0 and < 1) = number of control points + type -1.
3B. KnotList object, combined with nc-specifier = 0: This case results in an error
condition; because of the relationship noted under (3A); the number of
control points is not free to escalate when the type and number of knots
are
both specified.
Tools/Clearance works with NUBFits (as with BFits); with a single
NUBFit object
35
selected, it reports the number of control points used and the goodness-
of-fit RMS
and worst-point statistics.
Tools/Special/Freeze Fit (and the BFreeze command) work with
NUBFits just as for
BFits. They will generate a NURBCurve and its control points (or
NURBSnake and
its control magnets) as freestanding MultiSurf objects. When the fit is
non-uniform, it
generates a KnotList in addition.
NUBFitSnake
The NUBFitSnake entity is very similar to the NUBFitCurve. Its parent
must be a snake, and the resulting approximation is in the u,v, space of
the supporting surface.
Its control points are u,v locations on the supporting surface. The
“tolerance” in logtolerance is a dimensionless value in u,v-space, rather
than a distance.
NUBFitSurf
The NUBFitSurf entity works similarly to NUBFitCurve, but it
approximates a surface.
Characteristic data
u-type
u-knotlist
ncu-specifier
v-type
v-knotlist
ncv-specifier
log-tolerance
Surface
Description The fittings in the u- and v-directions are relatively independent. As with
NUBFitCurve, there are 3 alternatives for each knotlist entry:
1. ‘*’ (the “default” knotlist) - This allows the fitting to use the best set
of nonuniform knots it can come up with.
2. ‘*UNIFORM’ - a new “predefined” knotlist, available in the System
Entities; this forces the fitting to use uniform knots, essentially
duplicating the function of BFitCurve.
3. You can choose a KnotList or KnotList2 object containing a suitable
specified list of knots, and two alternatives for each nc-specifier:
36
A. nc-specifier can be a positive integer (greater than type), specifying a
definite number of control points to use for the fit.
B. nc-specifier can be 0, which signifies “increase the number of control
points until tolerance is met”.
These can be used in any combination, except 3B. (nc-specifier = 0 can’t
be used in combination with a specified KnotList object, because of the
fixed number of knots.)
One common reason to use a NUBFitSurf prior to exporting IGES is to
control the characteristics and quality of NURBS approximation on a
surface-by-surface basis.
Another reason is to achieve exact matching along shared surface edges
in the exported file. For example, suppose we have ‘topside’ and
‘bottom’ surfaces on a single-chine hull, joining each other exactly along
the chine; suppose the u-direction is longitudinal and the v-direction is
transverse on each surface. If we just export the ‘topside’ and ‘bottom’
surfaces through IGES with a global tolerance, we are very likely to come
out with a different number of control points on the two exported
surfaces. For example, the ‘topside’ might meet the tolerance with 12 x 5
control points, while ‘bottom’ gets 10 x 7. Two NURBS curves with
different numbers of control points and different knots are identical only
under extremely unlikely conditions, so there will almost certainly be
gaps and overlaps between the two IGES surfaces that result.
However, if we first make two NUBFitSurfs ‘nub_topside’ and
‘nub_bottom’ using the following conditions:
• same ncu-specifier (say, 12 for this example)
• same knots for the u direction
• same number of u-divisions x u-subdivisions on the base surfaces
then we are guaranteed that the two fitted surfaces will join exactly
along their common edge. Since there is no further approximation in the
export of the NUBFitSurfs, the two exported surfaces will join exactly in
the IGES file, and in the receiving application.
Tools and NUBFit Surf

Tools/Clearance works with NUBFitSurf (as with BFitSurf). With a
single NUBFitSurf selected, it reports the number of control points used
and the goodness-of-fit RMS and worst-point statistics.
37
Surfaces
Foil Lofted Surfaces
Characteristic data curve1 ... curveN = master curves (t-orientation doesn’t matter)
type = type of airfoil family, 1 to 5; types 101 - 255 reserved for user-
defined foils
relabel = name of a Relabel object or ‘*’ for default labeling
Description In the Foil-Lofted Surface, the lofting curves are Foil Curves. This means
the Foil-Lofted Surface passes through its first (trailing edge), third
(leading edge), and fourth or fifth (trailing edge; if present) master
curves, but in general not the others, which control the thickness and
camber much as the second and fourth (if present) control points of a
Foil Curve do.
Similarly to a Foil Curve, the number of master curves (3, 4, or 5)
controls whether you generate a half or full profile.
Half-section surface
With 3 master curves, you get one half of a symmetric foil-lofted surface,
with the parameter v running from 0 at the trailing edge to 1 at the
leading edge. curve2 controls the half-thickness of the symmetric foil; this
is the distance of curve2 from the ruled surface consisting of all the chord
lines joining the leading and trailing edges.
Full-section surface
With 4 or 5 master curves, you get a full-section foil-lofted surface — that
is, if curve1 and the last master curve coincide, or are the same curve.
When curve1 and the last master curve are not the same, you still get a
more-or-less foil-shaped surface, but the trailing edge is open.
5 master curves. This creates a full section surface:
• symmetric — when curve2 and curve4 are symmetrically placed
(mirror images) with respect to the chord line
• cambered — when curve2 and curve4 are NOT symmetrically placed
with respect to the chord line
curve2 and curve4 act together in determining the thickness and camber.
The full-section foil will be cambered if curve2 and curve4 are not
symmetrically placed with respect to the chord line.
4 master curves. This creates a symmetric full section. A 4-point foil
behaves just like a 5-point foil in which the 4th parent is the mirror
image of the 2nd parent across the chord line.
u,v parameters. The u parameter on the surface is the same as the t
parameter for the first master curve. The v parameter is the parameter
along the lofting foils, running from 0 at the trailing edge (upper surface)
38
to 0.5 at the leading edge, then to 1.0 at the trailing edge (lower surface).
The lofting foils are the u = constant lines.
39
type:
type-1 NACA 4-digit series; max camber at 40%
type-5 NACA 0010-34; max camber at 40%
types 101-255 user-defined foils
relabel is used to control the labeling of the v-direction (the lofting

curves) of the surface. See Reference section “Relabeling Curves and
Snakes.”
Usage The Foil-Lofted Surface can be used to create complex wing, keel,
rudder, and hydrofoil geometries having true NACA sections or user-
defined foil sections.
Example 1 floft1.ms2 (half-section)
‘keel’ is a type-1 half-section foil-lofted surface defined by 3 master
curves: ‘trailing_edge’, ‘top’, and ‘leading_edge’. ‘keel’ passes through
‘trailing_edge’ and ‘leading_edge’ and uses ‘top’ to establish the half-
thickness of the foil surface.
half-section
foil-lofted surface
'top'
v
X
0,0 'leading_edge'
'trailing_edge'
'top'
u
v 0,0
X
'leading_edge'
'trailing_edge'
FLOFT1.MS2 at Lat -30, Lon 70 (left).
Example 2 floft2.ms2 (full-section)

FLOFT2.MS2 is a cambered elliptic wing made from 5 master curves that
are full foil sections: ‘trailing_edge’, ‘upper’, ‘leading_edge’, ‘lower’, and
‘trailing_edge’ (again). Note that ‘trailing_edge’ is named twice in order
to close the surface. In the model (but not the figure below), tickmarks
are turned on for the master curves so you can see them more easily.
Camber and thickness can be adjusted independently, by sliding
RelBeads ‘camber’ (cyan) and ‘thickness’ (red) respectively. (The model
is constructed so that the t-offset of RelBead ‘camber’ is the
40
camber/chord ratio, and the t-offset of RelBead ‘thickness’ is half the
thickness/chord ratio.)
'upper'
X 'thickness' Z
'trailing_edge'
'camber'
'lower'
'leading_edge'
FLOFT2.MS2
at Lat 30, Lon 60
Y (tickmarks turned off).
See also FCurve, FSnake
B-Fit Surface
Characteristic data utype, vtype = B-spline types for u and v directions: 1, 2, 3, etc.
ncu-specifier, ncv-specifier = integers; specify choices for number of control
points to use for fit (0 if free)
log-tolerance = log (base 10) of tolerance
surface = basis surface
Description A BFitSurf is a uniform B-spline surface, least-squares fitted to the
tabulated data of its basis surface. Its principal function is for exporting
surfaces to NURBS-based systems.
• you can see the NURBS approximation results right in MultiSurf,
prior to exporting IGES
• you can assure accurate joins between NURBS surfaces (by forcing
the same number of control points along adjoining surface edges)
• you can specify different tolerances for different surfaces
log-tolerance is the tolerance specified with a base-10 logarithm, e.g. -3 for
tolerance of 10-3. This notation permits a wide dynamic range. (Tolerance
itself is measured as the RMS (root-mean-squared) deviation of the fitted
surface from tabulated points of the basis surface.)
utype and vtype are the spline type for the u and v directions respectively:
1 = linear, 2 = quadratic, 3 = cubic, etc.
ncu-specifier and ncv-specifier are integer values which specify choices for
the number of control points to be used in the u- and v-directions. For
each specifier there are two choices:
41
When you specify a positive value (it must be greater than the
corresponding type), the program will use exactly this number of
control points for the fit.
When you specify zero, the program will start at type +1 control
points and cycle the number upward until it either achieves the
fit within the tolerance or reaches the limit of 128 control points.
If the BFitSurf does not meet the specified tolerance, the Error window
will pop up and let you know the problem. Clicking the <Edit> button
gives you quick access to the Edit dialog where you can try a different
set of values.
When the BFitSurf is displayed, you can see where the control points are
by turning on the "net" (visibility = net) for the surface − the control
points are located where the net lines meet.
If you would like MultiSurf to report information about the quality of fit,
select the BFitSurf and then choose Tools/ Clearance.
Example bfitsrf1.ms2
This is the demo model plus a BFitSurf. For the first run, we used the default values: u-type and v-type both
= 3, ncu-specifier and ncv-specifier both = 0 (which lets the program set the number of control points), and
log-tolerance = 0. The discrepancies between the XContours on the 2 surfaces visually shows that the fit is
not great (Fig. 2, left). Checking Tools/ Clearance with the BFitSurf selected, reports that MultiSurf used
ncu = 4 and ncv = 4 to achieve the fit.
Z
BFitSurf
Y
XContours on
CLoftSurf basis surface
X Fig. 1
With log-tolerance = -2 (the value it has in the final sample model file), the fit is better (Fig. 2, right).
Tools/ Clearance shows that MultiSurf used ncu = 6 and ncv = 4 to achieve this fit.
Fig. 2. Ship lines (body plan lifts) of 2 variations the BFitSurf: with log-
tolerance = 0.0 (left); with log-tolerance = -2 (right).
See also BFitCurve
42
B Surface
Characteristic data point1 ... pointMN = control points

utype, vtype = B-spline types for u and v directions: 1, 2, 3, etc.
ncu = no. of control pts. in u-direction
(ncv = in documentation below, no. of control points in v-direction)
Description BSurf is a uniform B-spline surface. The control points form a
quadrilateral net which guides or shapes the surface in much the same
way as the control points shape a B-spline curve. In general, the surface
does not interpolate any of its control points, except the four at the
corners of the net. You can think of the surface as being connected to the
interior control points by springs.
The edges are all B-spline curves using control points from the four
edges of the net:
v = 0 uses the first ncu control points
v = 1 uses the last ncu control points
A B-spline Surface is a simplified case of a NURB Surface — the knots
are uniform in both directions, and you don’t need to use weights
because they are all set to one. Technically, a B-spline should have at
least one more control point than its type, e.g. at least 2 for type-1, 3 for
type-2, etc. However, MultiSurf will automatically “demote” a spline to
lower type as necessary to fit the number of control points. For example,
a type-3 spline specified with only 2 points will be treated as type-1; with
3 points it will be treated as type-2; only with 4 or more points will it
actually be a cubic spline.
Example bsurf.ms2
In this example, ‘hull’ is a B-spline surface which uses the same data
points as the models:
CLFT5X4.MS2 in which ‘hull_c’ is a C-lofted surface (“CLoftSurf”
example)
BLFT5X4.MS2 in which ‘hull_b’ is a B-lofted surface (“BLoftSurf”
example 1)
Generally, a B-spline surface will lie inside a C-lofted surface defined by
the same set of control points. A B-spline surface and a B-lofted surface
defined by the same set of control points and using the same types of B-
splines will be identical, but the BLoftSurf entity is much more flexible
than the BSurf entity when it comes to refinement of surfaces.
For the surface ‘hull’ in this example, the u-direction (longitudinal, in
this case) B-splines are type-3 (cubic), the v-direction B-splines are type-2
(quadratic), and the number of control points in the u-direction is 5.
43
Z
Y
u 0,0
B-spline surface
X Z
control points
BSURF.MS2 at Lat -20, Lon 50 (top) and in <x> view (bottom).
See also BLoftSurf, CLoftSurf, NURBSurf
NURBS Surface
Characteristic data point11, point21 ... pointMN = control points

wt1, wt2 ... wtMN = corresponding weights
knotlist1 and knotlist2 = names of two KnotList objects or ‘*’ for uniform
knot spacing
utype, vtype = B-spline types for u and v directions: 1, 2, 3, etc.
ncu = number of control points in u-direction
(ncv = in documentation below, number of control points in v-direction)
Description The NURBSurf is a rational B-spline surface. The control points form a
quadrilateral net which guides or shapes the surface in much the same
way as the control points shape a B-spline Curve (Fig. 1). In general, the
surface does not interpolate any of its control points except the four at
the corners of the net.
44
Z
NURB surface Y 0,0

v
'net' for NURB surface u
Fig. 1
A NURBSurf and its net.
The edges are all rational B-spline curves using control points from the
four edges of the net:
v = 0 uses the first ncu control points
v = 1 uses the last ncu control points
wt1, wt2, etc. are the weights associated with each control point. If wt’s
are all 1 you get the plain vanilla B-spline Surface. Increasing wt on one
control point draws the surface (and the parameter lines) toward that
control point. In Figure 2, point ‘P32’ has been given a weight of 3, while
all others remain at 1. This pulls the surface (and the parameter lines)
toward point ‘P32’. Below the profile view, the body plan view compares
the X = 10 contour of the original, evenly-weighted model to that of the
model with ‘P32’ weight = 3.
knotlist1 and knotlist2 control knot spacing. If not default ‘∗’ (uniform
spacing) for each list, the knotlists should have (ncu - utype + 1) and (ncv
- vtype + 1) components respectively, and each list should begin with 0
and end with 1.
45
Z
NURB surface Y
'P32'
control points
X Z
X-contour at X=10
original model model with

'P32' wt=3
Fig. 2
A NURBSurf with the weight on one point (‘P32’) increased to 3.
Examples nurbs7a.ms2, nurbs7c.ms2.
‘hull’ is a NURBSurf formed by 28 points (‘P11’ to ‘P47’) with all weights
equal to 1. This is the same hull as in Figure 1 above, but to make that
figure, we used visibility = net plus u-constant and v-constant lines. The
u-direction B-splines are type-2 (quadratic) and the v-direction B-splines
are type-3 (cubic). The two default knotlists distribute the knots
uniformly in both the u and v directions. There are 4 control points in the
u-direction (and 7 in the v-direction).
Now let’s see what happens when we change knot spacing in the v-
direction. This means we need to add the definition of a knotlist to the
model. Since in the longitudinal direction on this hull surface we use
cubic splines (vtype = 3) with ncv = 7 columns of control points, then the
knotlist for the v-direction (which we named ‘knotsv’) needs 7 - 3 + 1 = 5
entries, for example { 0 .25 .5 .75 1 }.
This particular knotlist would actually produce uniform B-splines; i.e., it
would have the same effect as using the default ‘∗’ for knotlist2 in the
NURBSurf description (Fig. 3, top).
But using:
{ 0. .45 .50 .55 1.00 } would allow a region of high longitudinal
curvature near the middle (v = .5) of the hull
{ 0. .50 .50 .50 1.00 } would actually permit a slope discontinuity
(in this case, a crease or bend) to run around the hull at v = .5,
as in the model nurbs7c.ms2 (Fig. 3, bottom). Note that at most
type + 1 sequential knots can have identical values.
46
Y
v
u
X
0,0 u
uniform knot spacing
Y 3 knots at v=.5
v
u
X
0,0 u
non-uniform knot spacing
Fig. 3. Uniform (top; NURBS7A.MS2) and non-uniform (bottom;

NURBS7C.MS2) knot spacing on a NURBSurf; view here is Lat 90, Lon -90
orthographic.
See also BCurve, BSurf, BLoftSurf, CLoftSurf, KnotList.
Projected Surface
Characteristic data surface = basis surface

mirror = a plane, Line, or point object
Description A Projected Surface is a surface formed by projecting a specified surface
(called the “basis surface”) onto a mirror. Each point of surface is
projected normally (i.e., perpendicularly) onto mirror. The Projected
Surface therefore lies entirely in mirror. If the basis surface is later
changed, the Projected Surface will be automatically relocated to remain
in the perpendicularly-projected position.
mirror can be a plane, Line, or point object:
When mirror is a plane, each projection line is perpendicular to the
plane.
When mirror is a Line, each projection line is perpendicular to the
Line.
When mirror is a point, each projection line ends at that point.
(Note: When mirror is a point, each projection line passes through that
point. This Projected Surface would be doubly degenerate, i.e. the entire
surface would be at a single point. We do not know of any present utility
in making these degenerate surfaces.)
47
Example projsurf.ms2
‘outboard’ (the basis surface) is the outboard surface for the starboard
hull of a planing catamaran design. It is like half of a conventional
V-bottom hull. ‘inb_plane’ is a plane defined by three points: ‘p11’ at the
top of ‘MC1’ (the point of the stem), ‘p13’ at the bottom of ‘MC1’, and
‘p33’ at the bottom of ‘MC3’. (Since its three control points are all at Y =
10, ‘inb_plane’ currently is the same as the YPlane at Y = 0.)
‘inboard’ (the projected surface) is the inboard surface for the starboard
catamaran hull, obtained by projecting ‘outboard’ onto ‘inb_plane.
‘inboard’ is flat and parallel to the Y = 0 plane (the centerplane), at Y =
10.
Note that we could create the port hull for this catamaran by turning on
Y-symmetry (Settings/ Model), as we did in the “MirrSurf” entity
description.
Y Z
'p11'
projected surface 0,0
v
basis surface
u
'p13'
'inb_plane'
X
'p33' 'inb_plane'
Z
'p11'
projected surface
basis surface
Y
'p13'
'p33'
PROJSURF.MS2 at Lat -20, Lon 60 (top) and <x> view (bottom).
See also MirrSurf, ProjPoint, ProjCurve
Relative Surface
Characteristic data surface = basis surface

point1, ( point2, point3, point4 ) = offsets at corners of surface (1 or 4
required)
Description A Relative Surface is formed by offsetting the points of the basis surface,
in a way similar to making a Relative Curve from a basis curve. You
specify a Relative Surface by naming the basis surface and new positions
for either:
48
• one corner (point1) — Each point of the RelSurf is offset from the
corresponding point of the basis surface by a constant distance and
direction — the offset of point1 from the u=0, v=0 corner of the basis
surface. The u=0, v=0 corner of the RelSurf is at point1.
Note that a RelSurf using a constant offset is NOT the same as creating
an OffsetSurf (in which the offsets are applied perpendicularly to the
basis surface). The perpendicular distance between a RelSurf and its
basis surface would not in general be constant.
• each of the four corners (point1 to point4) — The surface is
stretched out to meet these four corner points. Intermediate points
are offset by amounts linearly related to their u, v coordinates.
The order of naming the four corner points is important. The corners are
identified as follows:
point1 is at u = 0, v = 0 point3 is at u = 1, v = 1
point2 is at u = 1, v = 0 point4 is at u = 0, v = 1
For most accurate results, give the RelSurf the same divisions as the basis
surface.
Example relsurf2.ms2
Z relative surface
0,0
u
'Q2'
v
'P13'
'Q3'
'Q4'
X
u 0,0
'r11
Y 'P23'
basis surface
RELSURF2.MS2 at Lat -20, Lon 60.
‘patch’ is the basis surface for the relative surface ‘relsurf’. ‘Q1’, ‘Q2’,
‘Q3’, and ‘Q4’ are the four corners of ‘relsurf’. They are RelPoints offset
from the four corner points of ‘patch’. ‘r11’ is an AbsMagnet at the u = 1,
v = 1 corner of ‘patch’. The translation surface has no point object
defined at that corner, yet one is needed in order to define the offset of
the relative surface from that corner, hence the need for ‘r11’.
Note that in this example the surface divisions are not coordinated:
‘patch’ has 6x3 5x2, while ‘relsurf’ has 5x3 4x2 — we did this just to
make the two surfaces easier to identify in black-and-white figures.
See also OffsetSurf, RelCurve, RelSnake
49
Tabulated Surface
Characteristic data filename = filename of .PAT file
patch number = number of patch in file
Description filename is the filename of a .PAT file. If no extension is given, MultiSurf
assumes .PAT.
A .PAT file (see sample contents and notes in example below) is a special
case of a .3DA file, in which one or more rectangular meshes (patches) is
represented by a set of nv + 1 polylines, each having nu + 1 points.
Individual patches are separated by an extra record having zero pen. A
polyline of nu + 1 points consists of one pen = 0 record followed by nu
records with nonzero pen. u-divisions × u-subdivisions is required to be
equal to nu, and v-divisions × v-subdivisions is required to be equal to nv,
or a geometry error occurs.
patch no. is the number of the patch to be taken from the .PAT file: 1 for
the first, 2 for the second, etc.
coordinates are added to the surface coordinates. This results in
a shift (but no rotation).
rotation.
The u-direction of the TabSurf is in the direction of the polylines.
Note: A TabSurf has less inherent smoothness than most native
MultiSurf surfaces. A TabSurf is a faceted model of the original surface
and cannot be further subdivided inside MultiSurf.
Example tabsurf.ms2 (needs fillet.pat)
FILLET.PAT is a patch file (see partial contents below) made from a
model having three surfaces: two plane patches, with a fillet surface
between them. The TabSurf ‘s1’ is the first of these patches, ‘s2’ is the
second, and ‘s3’ is the third. The three patches use for insert location
(frame/point) the Point ‘p1’ (at X=0, Y=0, Z=1), which has the effect of
adding 1.0000 to the Z-coordinate of each point read from the file. Since
frame/point is a point, there is no rotation of the patches, just a 1-unit shift
in Z. Moving ‘p1’ would shift the insert location of the entire set of
TabSurfs, again without rotating them.
The magnet ‘m1’ is added to show that TabSurfs, like other surfaces, can
parent magnets.
50
Z
tabulated surface 's2'
'p1'
'm1'
TABSURF.MS2 at Lat 30, Lon 160.
Partial contents of fillet.pat

0 0.000 0.000 0.000 pen=0 point (1st point) on 1st polyline
of 1st patch (‘s1’)
10 0.200 0.000 0.000
10 0.400 0.000 0.000
10 0.600 0.000 0.000
10 0.800 0.000 0.000
10 1.000 0.000 0.000 6th (last) point of 1st polyline; u-
divisions x u-subdivisions for ‘s1’ = 5; nu
+1=5+1=6
0 0.000 0.200 0.000 pen=0 point (1st point) on 2nd polyline
10 0.200 0.200 0.000
10 0.400 0.200 0.000
10 0.600 0.200 0.000
10 0.800 0.200 0.000
10 1.000 0.200 0.000
0 0.000 0.400 0.000 pen=0 point (1st point) on 3rd polyline
10 0.200 0.400 0.000
10 0.400 0.400 0.000
10 0.600 0.400 0.000
10 0.800 0.400 0.000
10 1.000 0.400 0.000
0 0.000 0.600 0.000 pen=0 point (1st point) on 4th polyline
10 0.200 0.600 0.000
10 0.400 0.600 0.000
10 0.600 0.600 0.000
10 0.800 0.600 0.000
10 1.000 0.600 0.000
51
0 0.000 0.800 0.000 pen=0 point (1st point) on 5th polyline
10 0.200 0.800 0.000
10 0.400 0.800 0.000
10 0.600 0.800 0.000
10 0.800 0.800 0.000
10 1.000 0.800 0.000
0 0.000 1.000 0.000 pen=0 point (1st point) on 6th (and last)
polyline; v-divisions x v-subdivisions for
‘s1’ = 6; nv + 1 = 5 + 1 = 6
10 0.200 1.000 0.000
10 0.400 1.000 0.000
10 0.600 1.000 0.000
10 0.800 1.000 0.000
10 1.000 1.000 0.000
0 0.000 0.000 0.000 extra pen=0 record = end of patch
0 0.342 0.121 0.000 pen=0 (first) point of 1st polyline on
2nd patch (‘s2’)
11 0.360 0.186 0.000
...
See also TabPoint, TabCurve, WireFrame
X-Spline Lofted Surface
Characteristic data curve1, curve2 ... curveN = master curves (t-orientation doesn’t matter)
type = direction of orientation of XLoftSurf
ecc = end condition code for the two directions other than the direction of
orientation (specified by type)
graph1, graph2, graph3, graph4 = names of BGraph objects or ‘*’ for
default graph = end condition data for the two directions other than the
direction of orientation (specified by type)
Description The XLoftSurf, like other lofted surfaces, is parented by two or more
master curves. For the XLoftSurf, the number of master curves is
arbitrary. The XLoftSurf is lofted with X-splines. Like the XCurve, the
XLoftSurf needs end condition data, which is supplied by the four
graphs.
type specifies the direction of orientation of the lofting X-splines:
1 for X orientation
2 for Y orientation
3 for Z orientation
Note: this usage of type has NOTHING to do with “spline” type as used
in BLoftSurf, CLoftSurf, etc. entity specifications. XLoftSurfs use cubic
splines.
ecc is the end condition code for the lofting X-splines, for the two
directions other than the direction of orientation (specified by type):
0 = moment at both edges
1 = slope at v=0 edge, moment at v=1 edge
52
2 = moment at v=0 edge, slope at v=1 edge
3 = slope at both edges
graph is the name of a BGraph object. The four BGraphs supply the end
condition data for the lofting X-splines, for the two directions other than
the direction of orientation (specified by type)
type-1 type-2 type-3

(X orientation) (Y orientation) (Z orientation)
graph1 Y’ or Y” Z’ or Z” X’ or X” at first MC
graph2 Z’ or Z” X’ or X” Y’ or Y” at first MC
graph3 Y’ or Y” Z’ or Z” X’ or X” at last MC
graph4 Z’ or Z” X’ or X” Y’ or Y” at last MC
relabel is used to control the labeling of the v-direction (the lofting

curves) of the surface. See Reference section “Relabeling Curves and
Snakes.”
The type-1 XLoftSurf can reproduce exactly any FAIRLINE/2 surface. In
this case, the BGraphs must be the same type as the end master curves
and must have one component for each master curve vertex. The slope or
moment columns in the FAIRLINE/2 Rep can be copied directly into the
BGraphs. To automatically perform the translation of a FAIRLINE/2 Rep
file into Points, BCurves, BGraphs, and an XLoftSurf, you can use the
utility program FL2MSF (see Appendix B).
Crossed Master Curves

Users of XLoftSurfs should be aware that they are vulnerable to a special
error condition that can cause unexpected and puzzling results: the
dreaded "Crossed Master Curves" error (Error 247 or 248). To locate a
point at parameters u,v on an XLoftSurf, MultiSurf first evaluates each
master curve (and each end condition graph) at t = u, then it passes an
XCurve through the points in order, then it evaluates the XCurve at t = v.
The problem arises in the second stage if the points obtained from the
master curves don't form a strictly ascending or descending sequence in
the coordinate corresponding to type, e.g. the X-coordinate for type-1.
The construction of the XCurve fails, and the surface is marked as in
error; any descendants of the XLoftSurf get error 284.
This error can occur even when the master curves don't actually cross
within the nominal u-range of 0 to 1. One way it happens is a
consequence of MultiSurf's extending the tabulation of curves and
surfaces for one subdivision beyond 0 and 1. Suppose you have specified
10 divisions and 1 subdivision in the u-direction; then MultiSurf actually
tabulates the surface from u = -0.1 to u = 1.1. If master curves are
approaching each other near the 0 (or 1) end of the range, their
extensions can actually cross a little ways beyond 0 (or 1), and this
"invisible" crossing will trigger error 248. Two workarounds are possible:
(1) try increasing the subdivisions for the u-direction, since a smaller
subdivision means the tabulated surface extends a shorter distance
beyond u = 0 (or 1)
53
(2) change the master curves so they are farther apart at u = 0 (or 1)
Another way this error can be triggered is by dragging a magnet near or
beyond the u = 0 or u = 1 edge of an XLoftSurf, where master curves are
converging. As the program performs its search for the u,v coordinates
on the surface corresponding to the visual position of the cursor, the
surface might be evaluated at points outside the range u = 0 to u = 1. If
the error occurs while dragging a magnet, you can usually clear it by
choosing Edit/ Model File, then just <Ok> without making any changes.
Likewise, the error can occur during the solution for position of a
ProjMagnet, IntMagnet, ProjSnake, or IntSnake on an XLoftSurf. It is
especially important on an XLoftSurf to use a magnet to specify the
starting position of one of these objects. This helps a great deal to limit
the area that the program has to search in, reducing the chance that it
will look beyond the u = 0 or u = 1 edges.
Example defaultx.ms2
Z
Y
v 0,0
X-lofted surface
u
'xMCC'
'xMCA'
'xMCB'
X
DEFAULTX.MS2 at Lat -30, Lon 60.

DEFAULTX.MS2 is the FL/2B Default Rep converted to a MultiSurf
model file by the utility program FL2MSF (for FL2MSF details, see
Appendix B). The MultiSurf model is formed by three type-2 B-spline
master curves and an XLoftSurf. You’ll probably want to use Edit/
Model File to follow the discussion. There are four BGraphs which
distribute bending moment along the end master curves of the
XLoftSurf. The BGraphs are type-2, the same type as the B-spline master
curves, and they have the same number of values as the number of
vertices in each of the master curves — 3 for xMCA and 4 for xMCC.
See also Xcurve
54
Parametric Solids
All of the solids you can form with MultiSurf are defined with reference to a trio of parameters — u, v, and
w — each of which, has a nominal range of 0 to 1. The rule that defines the solid in terms of its basis
surfaces and/or curves is written in terms of these three parameters. You can think of every point of a solid
as being labeled with a particular trio of values for u, v, and w. Generation of the solid is a continuous
mapping of points in the unit cube to points in 3D space. We refer to the unit cube as the “u-v-w parameter
space” of the solid. It is a 3D space, each of whose points correspond to a point in the solid.
At any point in a parametric solid there is a direction (called the “u-direction”) in which u increases while v
and w are constant. Similarly there is a v-direction in which v increases while u and w are constant and a
w-direction in which w increases while u and v are constant. The only reason you usually need to be
conscious of which direction is which is when deciding how finely the solid should be subdivided in the
three directions for tabulation and display.
A parametric solid nominally has six faces and eight corners. The eight corners have (u,v,w) values of (0.,
0., 0.), (0., 1., 0.), (0., 0., 1.), (0., 1., 1.), (1., 0., 0.), (1., 1., 0.), (1., 0., 1.), and (1., 1., 1.). Some of the faces
can be degenerate surfaces, i.e. curves or even points.
The distribution of the parameter t along the curves used to build surfaces supporting a solid affect the
shape of the solid. In most cases, the “default” labeling works just fine. For a discussion about changing the
labeling, see Reference topic “Relabeling Curves and Snakes”.
For some applications and examples of using solids, including how to export 3D meshes, see Reference
topic “Using Solids”.
Note: MultiSurf users, please understand that support of parametric

solids does NOT turn MultiSurf into a “solid modeler” comparable to the
design programs in that category such as Pro-Engineer, SolidWorks, and
Solid Edge. These solid modeling applications are based on B-rep
(“boundary representation”) solids, a very versatile way to represent
solid objects of arbitrary complexity. Parametric solids are a more
restricted solid representation in that they must have the basic 6-face
topology of a cube (just as parametric surfaces always have the 4-sided
topology of a square). On the other hand, the solid shapes you can make
with MultiSurf’s parametric solids are comparatively diverse because
MultiSurf’s surface modeling capability is much richer than that of most
solid modelers.
Additional Solid Attributes
divisions: u, v, and w divisions and subdivisions

All solids have u-divisions, u-subdivisions, v-divisions, v-subdivisions, w-divisions, w-subdivisions
attributes. The number of u (or v or w) divisions x subdivisions = the number of line segments the program
uses to represent the surface in the u (or v or w) direction for display and internal tabulation. u-divisions, u-
subdivisions, v-divisions, v-subdivisions, w-divisions, and w-subdivisions are all integers (1-255). The
greater the total number of u, v, and/or w divisions, the more closely the solid screen display will
approximate the solid itself (which is composed of an infinite number of points).
In addition, the number of u-divisions (or v- or w-divisions) establishes the spacing of the u-constant (or v-
constant or w-constant) lines drawn in a wireframe display.
55
v-direction lines
0,0,0 w-direction lines
u w
u-direction lines
Fig. 4. u, v, and w-direction lines on a solid.

For instance, if a solid is defined with 6x4 4x2 2x2 divisions, MultiSurf will tabulate the solid internally
with 24 segments in the u-direction, 8 segments in the v-direction, and 4 segments in the w-direction (Fig.
4).
Note: The visibility options for solids do not “parallel” those for surfaces.
Solid display options show the u-direction lines (not the u= constant
lines), etc. These display lines run in the same direction as the orientation
marks for u, v, and w.
Orientation
Like surfaces, solids have an orientation attribute which can be either 0 (normal) or 1 (reversed). Solid
orientation is different from surface orientation — currently its only effect is to reverse the sign of volume
and weight.
Using Solids
Some applications
• Accurate weight analysis and CG for sailboat ballast (see RuledSolid
examples 1 and 2)
• Tank volumes (for an example, look at the model Tanks.ms2
• Exterior volume meshes for CFD analysis (for an example, look at
the model Exterior-Grid.ms2.
• Interior volume meshes for finite element analysis (for an example,
look at the model Interior-Grid.ms2.
56
Discussion of tank and grid examples
Tanks.ms2
This is a model for an integral tank that can be dropped into almost any hull with a single-surface bottom. It
starts with ‘p0’, which sets the positions of the planes forming the flat forward, top, and inboard surfaces,
and ‘p1’ which controls the length of the tank. The line ‘l0’ between these points is projected horizontally
and vertically onto ‘hull’. Each of the resulting ProjSnakes gets a ring at each end (‘r00’, ‘r01’, ‘r10’, and
‘r11’). Lines between the rings are projected onto ‘hull’, then a SubSurf ‘s4’ is made between the four
ProjSnakes.
To make the RuledSolid ‘fwd_tank’, we need two surfaces. One is the SubSurf ‘s4’ and the other is a
degenerate RuledSurf between the line ‘l0’ and itself.
Tools/ Mass Properties gives the tank volume = 7.68 cubic ft.
Tools/ Weight Schedule gives the weight (and c.g.) of its contents = 958 lbs including the port-side image.
Exterior-Grid.ms2
‘mesh1’ is a BLoftSolid forming a body-fitted 3D mesh for CFD analysis of flow past a streamlined body.
The ‘body’ (bright green) is a BLoftSurf made from elliptic cross-sections. The BLoftSolid has three
control surfaces: ‘body’, ‘middle’ (bright blue; an OffsetSurf), and ‘outer’ (dark red; a cylindrical outer-
boundary surface). Using an OffsetSurf for ‘middle’ ensures that the mesh is orthogonal to ‘body’. The
mesh is also relabeled in the w direction, creating thin cells at the ‘body’ surface for viscous flow analysis.
‘mesh1’ has 8x4x8x4x8x2 = 16,384 cells in each quadrant.
Interior-Grid.ms2
This model is a volume grid for finite-element analysis of a pulley wheel. (Turn on symmetry or View/
Render to see all the images.) ‘grid’ (gray; on Layer 2) is a RuledSolid between the inner and outer
RevSurfs (bright green; on Layer 1).
If you turn layers 1 and 2 off and take the <y> view, you can make many parametric variations by dragging
points ‘p0’, ‘p1’, ‘e0’, and ‘p3’. The analysis grid automatically adapts.
Solid Entities
Boundary Solid
A Boundary Solid is parented by six surrounding surfaces, analogous to
the way a Tangent Boundary Surface is parented by four surrounding
curves. If the surfaces meet accurately along their junctions, the
Boundary Solid will exactly fill the volume they bound. In the language
of grid generation, Boundary Solid implements "transfinite
interpolation" between its six bounding surfaces, although our
implementation, using three relabels as blending functions, is more
general than the standard method.
57
Characteristic Data u-graph
v-graph
w-graph
Bounding surfaces
Description Exactly six bounding surfaces are required in all cases. Some of the
surfaces can be degenerate, i.e., collapsed to a curve or a single point.
This allows the Boundary Solid to have the topology of a prism (one face
degenerated to a curve) or a pyramid (one face degenerated to a point),
for example.
The u,v,w directions of the Boundary Solid depend entirely on which
surface parent is first in the list of "Bounding surfaces". The first surface
parent becomes the w = 0 face, and its u and v parameter directions
become the u and v parameter directions for the solid. A non-degenerate
surface should be chosen for the first parent, otherwise the orientation of
the solid may be unstable, or it may not fill the volume enclosed by the
other bounding surfaces.
Associated errors
433. Boundary Solid: Wrong number of bounding surfaces (6 required).
Sample files
Boundary Solid1.ms2 is a simple rectangular block whose faces are 6
BSurfs.
Boundary Solid-Prism.ms2 is a prismatic block with a degenerate BSurf
's5' as one of its faces. If you move 's5' to the top of the list of parents, the
resulting solid does not fill the volume enclosed by the surface parents.
BlockSolid
A BlockSolid is a very simple solid requiring minimal parents. It is a
rectangular solid, oriented by a frame and otherwise supported by just
two points at two opposite corners.
Characteristic Data
u-Relabel, v-Relabel, w-Relabel -- optional relabels for the 3 parametric

directions
Frame
Point1
Point2
Description The block is oriented parallel to the Frame coordinates. Its u-, v- and w-
directions are along the Frame's x-, y- and z-axes respectively. Its u=0,
v=0, w=0 corner is at Point1, and its u=1, v=1, w=1 corner is at Point2.
The three Relabels apply to the three coordinate directions.
58
Associated errors
None
Sample files
BlockSolid2.ms2 is the simplest example of a block. It uses the default
frame '*' , so its faces are parallel to the global coordinate system.
B-spline Lofted Solid

Characteristic data surfaces = control surfaces
type = type for the lofting B-splines; 1, 2, 3, etc.
relabel = name of Relabel object or ‘*’ for default labeling
Description The BLoftSolid is a solid between two or more supporting surfaces. It is
generated by evaluating the point on each of the surfaces at each given
u,v and constructing a B-spline curve from the points. The BLoftSolid
interpolates its first surface (surface1) and last surface (surfaceN), but in
general not the others, which have a “guiding” or “shaping” effect like
the interior control points of a B-spline Curve
Technically, a BLoftSolid should have at least one more control surface
than its type, e.g. at least 2 for type-1, 3 for type-2, etc; therefore, a
BLoftSolid should have at least one more supporting surface than its
type. However, MultiSurf will automatically “demote” a spline to lower
type as necessary to fit the number of control points. For example, a
type-3 BLoftSolid specified with only 2 surfaces will be treated as type-1;
with 3 surfaces it will be treated as type-2; only with 4 or more surfaces
will it actually be a cubic solid.
The supporting surfaces must have similar u,v orientation — the 0,0
corners need to be in a relatively similar place, and u and v must run in
similar directions (otherwise the solid could wind up with an
unexpected twist).
u and v for the solid run in the same directions as u and v for each
supporting surface. w is the parameter along the lofting B-splines.
relabel is used to control the labeling of the w-direction (the lofting
curves) of the solid. The default relabel ‘*’ produces the “natural”
labeling, in which the parameter w is concentrated in highly curved
areas formed by closely-spaced control surfaces. See Reference section
“Relabeling Curves and Snakes.”
Example bloftsolid1.ms2
This BLoftSolid is made from three supporting surfaces:
surface1 = ‘blendctrsurf’ (cyan), a Centerpoint Boundary Surface
surface2 = ‘bsurfmiddle’ (green), a flat BSurf
surface3 = ‘bsurfbottom’ (magenta), another flat BSurf
Each of the lofting B-splines has three control points, taken from the
point at the same u,v on each of the three surfaces. These lofting B-
splines are type-2. As you can see, the BLoftSolid passes through its first
and last supporting surfaces, but not its middle one. The w parameter for
59
the solid runs along these B-splines; w = 0 is at the first supporting
surface and w = 1 at the last one.
Z
BLoftSolid
'blendctrsurf' X
'bsurfmiddle'
Y 0,0
u 0,0
w
'bsurfbottom' v 0,0
BLOFTSOLID1.MS2 at Lat 6, Lon -10.
Bsolid
A BSolid (or B-spline solid) is constructed from control points in a way

that is very parallel to BCurve and BSurf entity types, but extended to
one more dimension.
Characteristic Data
u-Relabel, v-Relabel, w-Relabel -- optional relabels for the 3 parametric
directions
u-type, v-type, w-type -- B-spline types for the 3 parametric directions
ncu -- number of control points in the u direction
ncv -- number of control points in the v direction
Control points -- (ncu x ncv x ncw of them)
(BSolid has a "Net" visibility option, like the "Polygon" option for a
BCurve, or the "Net" option f or a BSurf.)
Associated errors
436 BSolid: ncu, ncv, ncw must each be 2 or greater .
437 BSolid: The number of control points is not evenly divisible by ncu x
ncv.
Sample models
BSolid1.ms2 is a minimal example of a BSolid, with just 8 control points.
Its B-spline type is specified as 3 (cubic) in all 3 directions, but because
there are only 2 control points in each of the 3 directions, all the B-splines
are automatically demotes to degree-1 (linear).
60
CopySolid
The CopySolid entity is the extension of the concepts of CopyPoint,

CopyCurve, and CopySurf to one higher dimension.
Characteristic Data
Solid
Frame0 -- source frame
Frame1 -- destination frame
x-scale, y-scale, z-scale -- scale factors for the 3 frame axes
Associated errors
None
Sample files
CopySolid1.ms2 is an example of a CopySolid 'o1' made as a mirror
image of another solid 'o0', reflected in the X=0 plane. RPYFrame 'F0' is
located at MirrPoint 'p9', is left-handed (Orientation = 1) and has Pitch of
180 degrees, so its three axes are all parallel to the global X,Y,Z axes, but
the x-axis is reversed.
Ruled Solid
Characteristic data surface1 and surface2 = surface parents
relabel = name of Relabel object or ‘*’ for default labeling
Description The RuledSolid is a solid between two supporting surfaces. It is
generated by taking the point on each of the surfaces at each given u,v
and constructing the line (“ruling”) between. w is the parameter along
the line.
The supporting surfaces must have similar u,v orientation — the 0,0
corners need to be in a relatively similar place, and u and v must run in
similar directions (otherwise the solid could wind up with an
unexpected twist).
u and v for the solid run in the same directions as u and v for each
supporting surface. w is along the rulings. from surface1 to surface2.
relabel is used to control the labeling of the w-direction of the solid. The
default relabel ‘*’ produces linear labeling with respect to w.
Example 1 ruledsolidballast1.ms2
This is the outside lead ballast casting for a trapezoidal fin keel. The lead
line is adjustable by dragging rings 'r0', 'r1'. You can have the weight
schedule open and watch weight and C.G. change as you adjust the lead
line.
The lead line at the keel surface is an IntSnake. The SubSurf 's0' below
the leadline is projected to centerplane (ProjSurf 's1'), then 's0' and 's1' are
the parents for the RuledSolid.
61
X
RuledSolid
'ballast_center'
0,0,0
w
u
v
'ballast_outer'
RULEDSOLIDBALLAST1.MS2 at Lat 25, Lon 52.
Example 2 ruledsolidballast2.ms2
Example 2: ruledsolidballast2.ms2
This is a bulb keel using solids to get weight and C.G. of the fin and bulb
combination. ‘bulb_solid’ is a RuledSolid between the bulb surface and a
deliberately degenerate RuledSurf that lies along the bulb axis
(projection of ‘meridian’ on axis line). ‘keel_solid’ is a RuledSolid
between the fin surface and a ProjSurf on the centerplane. Note that
‘keel_solid’ has orientation = 1 to give it a positive volume and weight.
The missing lead where the fin passes through the bulb (we assume here
the fin goes all the way through the lead) is accounted for by
‘subtr_solid’. The weight schedule gives the correct weight and C.G. for
the combination.
'keel_solid'
'bulb_solid'
'subtr_solid'
RULEDSOLIDBALLAST2.MS2 at Lat -30, Lon 50.
62
u
'bulb_ctrline' 0,0,0 w
w v
u v
0,0,0 'keel_ctr'
'bulb'
'keel'
RuledSoilds
u
0,0,0 w
v
'subtr_ctr'
'subtract'
The three solids in RULEDSOLIDBALLAST2.MS2. The w parameter on the keel and the
subtraction solids are transverse lines from ‘keel’ and ‘subtract’, respectively, to the centerplane.
See also BLoftSolid, RuledSurf
Triangle Mesh Entity Types

Surface Works 5.0 has partial support for a new class of entity types
called Triangle Meshes. A Triangle Mesh is a form of surface,
fundamentally consisting of an assembly of triangles that join edge-to-
edge. A Triangle Mesh can also be subjected to an optional
smoothing/refining operation known as subdivision, which produces
successively finer meshes. Points and curves can be embedded in a
Triangle Mesh, analogous to magnets, snakes and rings on a surface.
Triangle meshes are much less constrained topologically than parametric
surface patches. They can have holes, and any number of sides, or even
no sides -- seamless closed bodies are easy to make.
Triangle meshes are included in Surface Works 5.0 primarily for the
possibilities they offer in flattening with the add-on Flattener product.
An STL or RAW file can be imported, resulting in a Light Triangle Mesh.
So far, they have no other import and export options.
Triangle Mesh basics

The basic constituents of a Triangle Mesh are a set of points, called nodes,
and a set of triangles, each made from 3 distinct nodes. The nodes are
numbered with an index, 1 to nnodes. The triangles are numbered with
another index, 1 to ntris.
63
Each triangle has 3 edges -- lines connecting 2 nodes. Each edge can be
shared with at most one other triangle. An edge that is not shared with
another triangle is a boundary edge, and is part of the boundary of the
Triangle Mesh.
The triangles must be sufficiently connected by shared edges so you can
reach any triangle in the set from any other by stepping from one
triangle to the next, always crossing shared edges. For example, two
triangles sharing just a single node do not make a valid Triangle Mesh,
nor do two triangles that don't share any nodes.
Unlike a surface, which has an invisible extension outside the 0-to-1
parameter range, a Triangle Mesh has no extension; when you come to
the boundary, you're really at the boundary.
Recursive subdivision
Subdivision is an operation applied to a Triangle Mesh, that results in
another Triangle Mesh with 4 times the number of triangles. The nodes
for the subdivided mesh are derived from the original nodes by an
averaging or smoothing operation
In each stage of subdivision, a new node is generated near the middle of
each existing edge, and each original triangle is replaced by 4 triangles.
Recursive subdivision is the repeated application of the subdivision
operation. Any Triangle Mesh entity has a degree property that controls
its degree of recursive subdivision. degree is a non-negative whole
number. The default value is 0, which means "no subdivision". The
unsubdivided mesh is often referred to as the "coarse mesh".
Locations on a Triangle Mesh

A triangle mesh does not have an overall coordinate system like the u,v-
parameters of a parametric surface. However, within each triangle it has
a local coordinate system that can be used to locate points on the surface.
Locations in a triangle are specified by so-called barycentric coordinates u,
v, w which act as weights applied to the 3 corners of the triangle. In the
triangle interior each of u, v and w is in the range of 0 to 1, and they
satisfy the constraint: u + v + w = 1, or (equivalently) w = 1 - u - v.
Thus the combination {triangle index, u, v} is sufficient to uniquely
identify a point located on a Triangle Mesh, because w can be expressed
as a function of u and v.
Figure: Barycentric coordinates in a single triangle
Common data
Orientation. Any Triangle Mesh entity has a normal orientation
property with one of the two values: Normal and Reversed. This is very
similar to the orientation property of a surface; it just defines which
direction will be considered the positive normal.
64
Weight/unit area. A Triangle Mesh has a weight/unit area property that
is completely analogous to weight/unit area for a surface. It is multiplied
by the Triangle Mesh area for inclusion in the Weight Schedule.
Visibility options. The edges of component triangles can be made
visible or hidden.
Kind. A Triangle Mesh has a kind, which is not yet used for anything.
The only valid value for kind is 0.
Degree. The degree of subdivision, a nonnegative whole number.
Triangle Mesh
Characteristic data degree = degree of subdivision

Points = A list of points that are the nodes.
Triangles = A list of integers, which are node indices in groups of three;
each group of three is a triangle.
Breaks = A list of node indices that identify breakpoints and breaklines.
Description This is the most basic form of Triangle Mesh, where the nodes are
explicit point entities, and the connection into triangles is explicitly
specified.
A subdivided TriMesh (degree > 0) is related in shape to its coarse-mesh

nodes in a way that is similar to the relationship of a B-spline Surface or
NURBS surface to its net of control points. The surface generally doesn't
pass through its coarse-mesh nodes, but instead acts as if it is attracted to
them. TriMesh has a "net" visibility option which just shows the coarse
triangulation, to help visualize this relationship. A major difference is
that, while the control net of a B-Spline surface has to be organized into
orderly rows and columns of control points, the TriMesh control net has
a much freer topology -- it can be any valid triangle mesh.
Example TriMesh1.ms2
Triangle Mesh 'tm' is made from 5 points, named 'p1' to 'p5' (turn on
point nametags). The points are nodes 1 to 5. As you can see in the
Triangles data, there are 4 triangles, with node indices: 1, 2, 3; 2, 5, 3; 5, 4,
3; 4, 1, 3. The Degree is 0, i.e., no subdivision is being applied.
Joint.ms2
A Trimesh model of a human leg. Change the degree to get better
definition.
Light Triangle Mesh
Characteristic data degree = sub-division amount

Frame = Parent location
65
Scale Factors, X, Y, and Z
Nodes = A list of real values which are node coordinates in groups of
three; each group of three is the x, y, z coordinates of one node.
Triangles = A list of integers which are node indices in groups of three;
each group of three is a triangle.
Breaks = A list of node indices that identify breakpoints and breaklines.
Description This is another basic way to represent a Triangle Mesh as a list of nodes
and a list of triangles. Rather than depending on point entities for its
nodes, a LtTriMesh is a self-contained single entity. This makes it
relatively compact. It is implemented primarily for purposes of
importing large triangle meshes from other systems.
Also see: Triangle Mesh; Light NURBS Surface
Example LtTriMesh1.ms2
This is the same example as TriMesh1.ms2, but cast in the form of a Light
Triangle Mesh. Frame coordinates for the five supporting points are in
the Nodes section. The Triangles section is the same as in TriMesh1.ms2 -
- node indices for 4 triangles.
Surface Triangle Mesh

surface = Surface, Trimmed Surface or Composite Surface
nu = no. of triangles in u-direction
nv = no. of triangles in v-direction
style = 0 – Forward 1 – Reverse 2 - Alternating
Description The SurfaceTriangle Mesh entity type, parented by a parametric or

trimmed surface, or Composite Surface. This creates a Triangle Mesh
from a surface. It has 3 integer parameters: nu, nv, style. These
parameters have no effect when the parent is a TrimSurf or CompSurf --
in those cases, you just get the object's triangulation. When the parent is
a parametric surface, it is divided first into rectangles (in u,v parameter
space) along u=constant and v=constant parameter lines, then each
rectangle is split into two triangles.
nu controls the number of triangles in the u direction, with these options:
nu < 0 -- uniform subdivision into -nu strips, each covering 1/(-nu) of the
u parameter space.
nu = 0 -- similar uniform subdivision as nu<0, but taking nu from the
basis surface u-divisions x subdivisions
nu > 0 -- the surface is first divided at any degree-1 breaklines, then the
intervals between breaklines are uniformly subdivided into a total of nu
strips
nv has similar meaning for the v-direction subdivision
style selects forward diagonals, reverse diagonals, or alternating
diagonals for the division of each rectangle into two triangles.
66
Example SurfTriMesh1.ms2
‘patch’ (the basis surface) is a Translation Surface. Surface Triangle Mesh 'I0'
has nu = 8, nv = 4, and style = 0. It therefore divides 'patch' into 8 x 4 x 2 = 64
triangles. These are uniformly distributed in both u and v directions, because
'patch' has no degree-1 breaklines.
Copy TriMesh

Triangle mesh = Parent Triangle Mesh entity
Frame1 = Parent location
Frame2 = Copy Location
Scale Factors, X, Y, and Z
Description Copy entities takes a parent of the same class (the parent entity); two
frames; and three scale factors. The two “environments” are two frames,
designated Frame0 (the “source” frame) and Frame1 (the “destination”
frame). The process of constructing the Copy object has three steps:
1 transform the basis object into frame coordinates in Frame0;
2 apply the three stretching factors to the x, y, z coordinates in Frame0;
3 create the Copy at those same x, y, z coordinates in Frame1.
All this sounds like a lot of Frames. Although Frames are versatile and
not hard to create, remember that any point can serve as a frame; its x, y,
z axes are parallel to the global X, Y, Z axes respectively. In practice,
some 90% of Copy entities will use points for one or both frame parents.
Also see: Copy Curve, Copy Surface
Example CopyTriMesh1.ms2
‘tm’ (the parent Triangle Mesh) is a 5-node, 4-triangle Triangle Mesh
entity similar to TriMesh1.ms2. 'I0' is a Copy Triangle Mesh made from
'tm', with the two points 'p3' and 'p6' as the two frames. 'I0' has an x-
scale factor of 0.5, so it is only half as wide in the X-direction as its
parent.
PolyTriMesh
Characteristic data degree = sub-division amount

Log-tolerance = Base-10 logarithm of distance tolerance
Triangle meshes = Parent list
Description The Poly Triangle Mesh entity type is two or more Triangle Mesh objects
assembled into a single Triangle Mesh. The component Triangle Meshes
currently need to have compatible triangulations, with nodes agreeing
within a tolerance specified by the log-tolerance parameter.
67
Also see: Poly Surface
Example PolyTriMesh1.ms2
Poly Triangle Mesh ‘I0’ is assembled from two parent Triangular
Meshes: 'tm1' (6 nodes, 4 triangles) and 'tm2' (9 nodes, 8 triangles). Since
the two parents share 3 point entities as their common nodes, the value
of Log-tolerance is not significant in this example.
Triangle Mesh Magnet
Characteristic data Triangle mesh = Host Triangle Mesh entity

Triangle no. = A triangle index in the host Triangle Mesh
u,v = Barycentric coordinates in the selected triangle
Description A Triangle Mesh Magnet is a point located on a Triangle mesh by
{triangle index, u, v} location in the coarse mesh. It stays in the same
relative position when the host Triangle Mesh is subdivided.
Dragging is implemented in MultiSurf.
Example TMBSnake2.ms2
This model has 7 Triangle Mesh Magnets supporting two Triangle Mesh
B-spline Snakes.
Triangle Mesh Projected Magnet
Characteristic data Point = Point entity to project

TMMagnet/ TriMesh = designates the host Triangle Mesh entity that the
Triangle Mesh Projected Magnet will lie on.
Mirror/TriMesh. = Plane, line, point or Triangle Mesh entity.
Description A Triangle Mesh Projected Magnet is a point located on a Triangle mesh
by projection of another Point along a straight line.
When Mirror/TriMesh is a plane, the projection line is normal
(perpendicular) to the plane.
When Mirror/TriMesh is a line, the projection line is normal
(perpendicular) to the line.
When Mirror/TriMesh is a point, the projection line radiates from the
point.
When Mirror/TriMesh is a Triangle Mesh (it has to be the same Triangle
Mesh that the Triangle Mesh Projected Magnet is to lie on) plane, the
projection line is along the local normal direction to the Triangle Mesh;
the Triangle Mesh Projected Magnet will be at the closest point of the
Triangle Mesh to Point.
68
TMMagnet/TriMesh can be either a Triangle Mesh Magnet or Ring, or a
Triangle Mesh entity.
When TMMagnet/TriMesh is a magnet or ring, it serves two purposes:
(1) It specifies which Triangle Mesh the Triangle Mesh Projected Magnet
is to lie on -- the host Triangle Mesh of the magnet or ring.
(2) It gives a starting location for the iterative search for the projection
point. In the case of multiple intersections of the host Triangle Mesh with
the line of projection, the Triangle Mesh Projected Magnet will (usually)
be made at the intersection nearest to the starting location (in terms of
distance measured within the Triangle Mesh, not 3-D distance).
When TMMagnet/TriMesh is a Triangle Mesh, it specifies directly which
Triangle Mesh the Triangle Mesh Projected Magnet is to be hosted on. In
case of multiple intersections of the projection line, the starting location
for the search will be at triangle 1 of the host Triangle Mesh, and the
search will usually converge to the intersection closest to triangle 1.
Example TMProjMagnet1.ms2
Triangle Mesh Projected Magnet 'G1' (bright red) is the vertical
projection of Point 'P0' onto the Surface Triangle Mesh 'I0', using the
system plane *X=0 for the mirror.
Triangle Mesh Projected Magnet 'G2' (bright white) is the normal
projection of Point 'P0' onto the Surface Triangle Mesh 'I0', by using 'I0'
as the Mirror/TriMesh parent.
Triangle Mesh Ring
Characteristic data TMSnake = A Triangle Mesh Snake entity

t position = location along snake
Description A Triangle Mesh Ring is a point located by t parameter value along a

Triangle Mesh Snake. It can serve to mark a location along the snake; it
also serves as a location on the host Triangle Mesh.
Dragging of Triangle Mesh Rings is implemented in Surface Works.
Example TMBSubSnake1.ms2
This model has four Triangle Mesh Rings (bright blue) marking locations
on two different Triangle Mesh B-spline Snakes (magenta) for the ends of
SubSnakes (yellow and bright cyan).
Snakes on Triangle Meshes
69
There is a class of Triangle Mesh Snakes representing curves embedded in a Triangle Mesh. These are
highly analogous to snakes on surfaces. There is a variety of possible ways to specify them, represented by
5 different entity types.
NOTE: Triangle Mesh snakes are curves, but for technical reasons in Surface Works 5.0 they can't serve as
curve parents for other entities (for example, as master curves for a surface).
Triangle Mesh B-spline Snake
Characteristic data Type = B-spline degree: 1 = linear, 2 = quadratic, 3 = cubic

Periodic = Switch controlling periodic property; 0 or 1
TMMagnets/TMRings = Control points; Triangle mesh magnets or rings.
magnet1, magnet2,...magnetN
Description A Triangle Mesh B-spline Snake is a continuous curve embedded in a

Triangle Mesh entity and governed by a set of Triangle Mesh magnets as
control points. The magnets must all be hosted by the same Triangle
Mesh. Except for degree-1, the snake does not generally pass through its
control points, but instead behaves as if it's attracted to them. Moving
one control point produces a local modification of the curve shape that is
strongest in the vicinity of that control point.
At the time of Surface Works 5.0 release, there is no difference in the
behavior of degree 2 and degree 3 for this entity type.
Setting the periodic property makes the snake reuse its control points in
a cyclic fashion, so the snake is a closed curve (t = 0 and t = 1 are at the
same location), and in fact is periodic ( t and t+N are at the same
location, where N is a positive or negative integer). The t=0 location on a
periodic snake is generally close to magnet1, but the snake does not
generally pass through magnet1.
A non-periodic Triangle Mesh B-spline Snake does begin at magnet1 and
end at magnetN, similar to a B-spline Curve or Snake.
For details about B-spline degrees and how they behave, see "B-spline
entities -- Degrees".
See also: B-spline Curve, B-spline Snake
Example TMBSnake2.ms2
This model has a total of four Triangle Mesh B-spline snake examples.
The host Triangle Mesh 'tm' is similar to LtTriMesh1.ms2, but has a
subdivision degree of 2.
'N2' (yellow) is a degree-1 snake with 3 control magnets, 'G0', 'G1', and
'G2'. 'N0' (magenta) is the degree-2 snake made from the same 3
magnets.
'N3' (yellow) is a degree-1 periodic snake made from 4 control magnets
'G3' to 'G6'. 'N1' (magenta) is the degree-2 periodic snake made from the
same 4 magnets.
70
Triangle Mesh Edge Snake
Characteristic data Type = Counterclockwise (0) or Clockwise (1)

TMMagnet1 = Triangle Mesh Magnet/Ring for the t = 0 end
TMMagnet2 = Triangle Mesh Magnet/Ring for the t = 1 end
Description A Triangle Mesh Edge Snake is a curve that follows the boundary of a
Triangle Mesh, between two Triangle Mesh Magnets. The two magnets
must be hosted by the same Triangle Mesh entity; they must both lie on
the boundary; and if the Triangle Mesh has multiple boundaries, the two
magnets must be on the same boundary.
Two boundary magnets divide a boundary loop into two parts. Type
determines which of those parts becomes the Triangle Mesh Edge Snake.
Example TMEdgeSnake1.ms2
This model has two Triangle Mesh Edge Snake examples. The host
Triangle Mesh 'tm' is similar to the Trimesh1.ms2 example, but has a
subdivision degree of 2. The two Triangle Mesh magnets 'G0' and 'G1'
(bright red) lie on the boundary of 'tm'. Edge Snake 'N0' (magenta) goes
from 'G0' to 'G1' in the counterclockwise direction. Edge Snake 'N1'
(yellow) is made from the same two magnets, in the same order, but,
being type-1, goes in the clockwise direction.
Triangle Mesh Intersection Snake
Characteristic data Type = 0 to 1

TMMagnet = a Triangle Mesh magnet or ring
Mirror/Surface = plane, frame, surface, Line or point. Defines a family of
potential cutting planes or surfaces.
Point = point entity. Specifies the actual cutting plane or surface.
Description A Triangle Mesh Intersection Snake is a curve on a Triangle Mesh,

located where a real or virtual cutting plane or surface intersects the
Triangle Mesh.
The TMMagnet parent serves three roles:
It specifies which Triangle Mesh the Intersection Snake.is cut on (i.e.,
the Triangle Mesh that is host to the TMMagnet)
If there are multiple intersections, the one closest to the TMMagnet is
selected.
The end of the selected intersection that is nearest to the TMMagnet
becomes the t = 0 end of the Intersection Snake.
71
The surface doing the cutting can be the actual plane, frame (its x,y-
plane) or surface specified by Mirror/Surface, if Point lies in that plane or
surface; or it can be a virtual cutting surface offset from Mirror/Surface
and passing through Point. (For details, see "Specifying the cutting
surface".)
The Type property is significant only when the intersection is a closed
curve. Then it is used to determine the clockwise/ counterclockwise
orientation of the Intersection Snake.
Example TMIntSnake1.ms2
'tm' is a Triangle Mesh similar to TriMesh1.ms2. The parents of Triangle
Mesh Intersection Snake 'N0' are:
Triangle Mesh Magnet 'G0' (red)
System plane *Y=0
Bead 'e0' (bright white)
'G0' specifies the host Triangle Mesh for the snake ('tm', host of 'G0'), and
selects which end of the cut will be the t = 0 end of the snake. The cut is
with a plane parallel to *Y=0, passing through 'e0'. (2-point Plane 'a0' is
included in the model to show the cutting plane. but is not directly
involved in the Intersection Snake.)
Triangle Mesh Projected Snake
Characteristic data Curve = curve entity

TMMagnet/TriMesh = Triangle Mesh magnet or ring, or Triangle mesh
entity. Designates the Triangle Mesh on which the Projected Snake will
lie.
Mirror/TriMesh = plane, frame, line, point, or Triangle mesh entity
Draft Angle = Draft Angle (degrees) -- Not currently implemented
Description A Triangle Mesh Projected Snake is a projection of Curve onto a Triangle

Mesh. Each point of Curve is projected along a straight line to locate the
corresponding point on the Projected Snake. The projection can be done
with any valid type of mirror, or can be normal to the Triangle mesh
host:
When Mirror/TriMesh is a plane, the projection lines are normal
(perpendicular) to the plane.
When Mirror/TriMesh is a Line, the projection lines radiate from the Line
and are normal (Perpendicular) to the Line.
When Mirror/TriMesh is a point, the projection lines radiate from the
point.
When Mirror/TriMesh is a Triangle Mesh (it must be the same Triangle
Mesh the projected Snake is to lie on), the projection is along the local
normal direction to the Triangle Mesh. Each point of the Projected Snake
will be at the foot of a line from the corresponding point on Curve
72
dropped normally to the Triangle Mesh. (Usually this is the closest point
on the Triangle Mesh.)
TMMagnet/TriMesh specifies the Triangle Mesh the Projected Snake is to
lie on. When TMMagnet/TriMesh is a Triangle Mesh magnet or ring, it
serves two purposes:
It specifies which Triangle Mesh the Projected Snake is to lie on: the
Triangle Mesh that is host to TMMagnet.
It provides a starting location on the Triangle Mesh for the iterative
search that locates the first point (the t=0 end) of the Projected Snake.
This is especially important in case of multiple intersections of the
Triangle Mesh with the projection line; then the location of TMMagnet
will help to select the correct one of two or more possible Projected
Snakes.
Whenever you create a Triangle Mesh Projected Snake, we recommend
you use a magnet or ring for the TMMagnet/TriMesh parent, locating it
reasonably near where the Projected Snake is to begin.
When TMMagnet/TriMesh is a Triangle Mesh, it specifies directly the host
triangle mesh for the Projected Snake. The search for the intersection
point at the t = 0 end of the snake will start at the first triangle of the
Triangle Mesh.
Example TMProjSnake2.ms2
‘tm’ is a dome-shaped Triangle Mesh entity with 7 nodes, 6 triangles,
and 2 degrees of subdivision. Curve is a B-spline Snake 'n0' (magenta)
drawn on the B-spline Surface 's0'. Triangle Mesh Projected Snake 'N0'
(yellow) is the vertical projection of 'n0' onto 'tm', using the system plane
*Z=0 as mirror. Triangle Mesh Magnet 'G1' (bright red) identifies the
host Triangle Mesh for the Projected Snake (its host, 'tm'), and provides a
starting location for the projection of its t = 0 end.
Triangle Mesh SubSnake
Characteristic data Degree = blending function B-spline degree, 1 to --

Direction = Normal or Reversed
TMRings = ring1, ring2, ... ringN
Description A Triangle Mesh SubSnake is a portion of a parent Triangle Mesh snake,

between ring1 (t = 0 on the SubSnake) and ringN (t = 1 on the SubSnake).
In most cases, 2 TMRings will be sufficient (N = 2). When more than 2
rings are used as parents (N > 2), the intermediate rings control the
labeling of the SubSnake, using B-spline blending in a fashion similar to
SubCurve.
Notice that if ring1 is at a larger t-value on the parent snake than ringN,
the t parameter on the SubSnake will run in the opposite direction from
the t parameter on the parent snake.
73
The Triangle Mesh SubSnake is hosted on the Triangle Mesh that is host
to its parent TMRings. The TMRings must all be on the same Triangle
Mesh snake (the parent snake). Predefined rings *0 and *1 can be used
for any of the TMRings, except at least one of the TMRings has to be an
actual bead or ring, in order to specify the parent snake.
Degree is the B-spline degree for the blending functions: 1 = linear, 2 =
quadratic, 3 = cubic, etc. As usual, if the number of parents N is less than
Degree+1, the splines are automatically demoted to degree N+1.
The Direction property is active only when the parent snake is periodic.
Then there are two possible ways to get from ring1 to ringN -- clockwise
and counterclockwise. When Direction is Normal, the subsnake will use
the part that goes around in the host snake's positive t direction; when
Direction is Reverse, it will use the other part.
Example TMBSubSnake1.ms2
This model has three example Triangle Mesh Subsnakes.
'N2' (yellow) is between Triangle Mesh Rings 'W0' and 'W1' (bright blue)
on Triangle Mesh B-spline Snake 'N0' (magenta).
'N3' (yellow) is between Triangle Mesh Rings 'W2' and 'W3' in the
Normal direction, i.e. its t parameter runs in the same direction as the
periodic parent snake, Triangle Mesh B-spline Snake 'N1' (magenta).
'N4' (bright cyan) is also between Triangle Mesh Rings 'W2' and 'W3', but
is in the Reverse direction, i.e. its t parameter runs counter to the t
parameter of the parent snake 'N1'.
Formulas
MultiSurf 5.x has a profound new capability for generation of parametric families of geometric models, in a
new class of real-valued objects: the Variable and Formula entity types. This is a complex topic, full of
new information and rich with new possibilities.
Constants, variables and formulas

There are three ways to specify a real number in the syntactic definition of an object: with a constant
value, a variable or a formula. Variables and formulas are objects. A variable carries a value that lies within
a specified range; a formula involves an expression made of variables, constants, functions and formulas.
Constants
Constants are 1.23, -0.7, 12e-8, +98, etc. Although they have no units they are always considered
to carry the adequate unit needed at the particular location where they are used.
For instance in
Point P 1 1 / 1 2 3 ;
1, 2 and 3 are considered as having length units.
In
AbsMagnet M 1 1 / S 0.123 0.456 ;
74
the u and v parameter values 0.123 and 0.456 are considered to be unitless.
Variables
Please see the document “What’s New in MultiSurf 5.0” for complete Variable documentation.
Formulas
A formula expresses a computation from constants, variables and other formulas with operators
and functions; it carries units resulting from its expression.
Basic syntax
The simplest syntax for a formula is:
Formula name / { expression } ;
For instance
Formula c / { 177.69 } ;
Formula f / { x } ;
Formula s / { x + y } ;
Formula s / { x + 5 * y } ;
Expression
Operators
The expression of a formula can be made with the usual operators ‘+’ (addition), ‘-‘ (subtraction),
‘*’ (multiplication), ‘/’ division. Exponentiation is symbolized with ‘^’ and is evaluated right to left, i.e.
x^y^z is x^(y^z) and not (x^y)^z. Parentheses can be used to prioritize the evaluation.
Formula d / { b^2 – 4*a*c } ;
Formula P / { R*I^2 } ;
Formula V / { n * R * T / P } ;
Formula s / { (1-t^2)/(1+t^2) } ;
Functions
A set of mathematical functions and object accessors is available:
Name Argument(s) Result Synopsis

SIN 1, radian (unitless) Unitless Sine
SIND 1, degree (unitless) Unitless Sine
COS 1, radian (unitless) Unitless Cosine
COSD 1, degree (unitless) Unitless Cosine
TAN 1, radian (unitless) Unitless Tangent
75
TAND 1, degree (unitless) Unitless Tangent
ATN 1, unitless Radian (unitless) Arc tangent
ATND 1, unitless Degree (unitless) Arc tangent
ATN2 2, both with same any units Radian (unitless) Arc tangent(y/x)
ATN2D 2, both with same any units Degree (unitless) Arc tangent(y/x)
LOG 1, unitless Unitless Natural logarithm
LOG10 1, unitless Unitless Decimal logarithm
EXP 1, unitless Unitless Exponential
SQRT 1, unit dimensions multiple of 2 Units dimensions of argument Square root
divided by 2.
ABS 1, any units Same units as argument Absolute value
ROUND 1, any units Same units as argument Rounding
MIN 2, both with same any units Same units as arguments Minimum
MAX 2, both with same any units Same units as arguments Maximum
TPOS 1, bead Unitless T parameter
UPOS 1, magnet Unitless U parameter
VPOS 1, magnet Unitless V parameter
XPOS 1, point Length X coordinate
YPOS 1, point Length Y coordinate
ZPOS 1, point Length Z coordinate
DIST 2, point, point Length Distance between points
CLEAR 2, point, graphic object Length Clearance
ANGLE 3: point, point, point Unitless (degree) Angle of three points
ARCLEN 3: curve, unitless, unitless Length Arc distance along curve
GRAPH 2: graph, unitless Unitless Evaluation of graph
FRAMEPOS 3: point, frame, index (1-3, for Length Coordinates in frame
x,y,z coordinate)
AREA 2; surface, use_sym (0 or 1) Area = L^2 Area of surface
VOLUME 2; solid, use_sym (0 or 1) Volume = L^3 Volume of solid
CENTROID 3; object, use_sym (0 or 1), index Length Coordinates of centroid
(1-3, for X,Y,Z coordinate)
IF 3: any units Same as units of selected If arg1 <=0, arg2, else arg3
argument
Examples
Formula I / { A * SIN( omega * t + phi ) } ;
Formula x / { EXP( COS( phi ) ) } ;
76
Formula y / { SQRT( ABS( b^2 - 4 * a * c ) ) } ;
Formula z / { TAN( c * ANGLE( p1, p2, p3 ) + phi^2 ) / b }

;
Unit consistency
Rules
Variables or formulas cannot appear anywhere inside a formula: unit consistency must be
observed. The units of the result of a formula must be specified and must match the units generated by the
expression. The rules are:
1. Only variables and formulas of same units can be added or subtracted.
2. Exponentiation of an expression with a rational exponent (this includes the SQRT function) is
only possible if the result has integral unit dimensions, e.g SQRT( x^2 + y^2 ) is valid,
(x^3)^(1/4)
3. Constants are considered to have the same units as the other operand in a sum and no units in
a product, e.g .in 1 – x, 1 is considered to have the same units as x. This is offered as a
facility: no physically meaningful formula has unit-less additive constants, since the formula
would become inconsistent in case of unit scaling.
4. Function arguments and results have unit dimensions as specified in the preceding chapter.
Unit mismatch is an error. Unit consistency is checked at parse time when possible (the unit dimensions of
x^y cannot be determined without knowing y’s value). It is however completely checked at evaluation time.
Examples
Point V1 1 1 / 1 2 3 ;
Point V2 1 1 / 4 5 6 ;
Formula V1dotV2 / { XPOS(V1)*XPOS(V2)
+YPOS(V1)*YPOS(V2)
+ZPOS(V1)*ZPOS(V2)
} L^2 ;
Variable Intensity / 2.5 I (0,10) ;

Variable Resistance / 100 L^2 M T^-3 I^-2(0,500) ;
Formula DPotential / { Resistance * Intensity^2 } M T^-3
I^-2 ;
Evaluation logic
No simplification
Expressions are not simplified, therefore
Formula f / { x - x } ;
depends upon x (twice!), which gets evaluated when f needs to be.
No optimization
No optimization is performed which would prevent parents from being evaluated, e.g. in
Formula f / { x * y } ;
77
y will be evaluated even if x is 0.
Complete syntax
‘Formula’ name { attr, } { ‘R:1’ } ‘/’ ‘{’ expr ‘}’ { units } ‘;’
Due to the fact that numeric constants are needed in formulas, and that MS2 syntax allows object
names such as 12, 4e3, etc., objects with such names will not be recognized inside the expression of a
formula, between { and }; these symbols will be considered as numeric constants instead. Likewise, objects
having the same name as a function (SIN, SQRT, XPOS...) will not be recognized.
Variables and formulas as parents

Variables and formulas can of course be parents of formulas. But they can be parents of all objects
depending on a real value, and wherever an object regardless of its type can be a parent, such as in an
object list. Therefore they can be parents of points (as coordinates, offsets, angles, t-, u- and v-parameter),
curves and surfaces (via knot lists and weights), etc.
Variable A / 10 L ;
Variable B / 5 L ;
Variable C / 30 L ;
Variable D / 25 L ;
Point P1 14 1 / 0 0 0 ;
Point P2 14 1 / A 0 0 ;
Point P3 14 1 / A B B ;
Point P4 14 1 / C B B ;
Point P5 14 1 / C 0 0 ;
Point P6 14 1 / D 0 0 ;
Variable k0 / 0.25 ;
Variable k1 / 0.75 ;
KnotList kl / { 0 k0 0.5 k1 1.0 } ;
Variable w0 / 1 ;
Variable w1 / 3 ;
NURBCurve spline0 11 5 64x1 / * 2 kl { P1 1 P2 w0 P3 w1 P4 1 P5 1 P6 w0 }
;
Note that since a variable is defined by three real numbers, these can be replaced by variables or
formulas: you can for instance have a variable with a variable range:
Variable min / 0 ;
Variable max / 24 ;
Variable x / 7 (min,max) ;
In this case the objects used as value and range bounds must all carry the same units (as always,
constant are assigned the units that match).
78
User interface
Figure 2 shows the properties for the newly created Formula ‘f1’. Note the Unit dimensions are
specified as “Length”. Figure 3 is the Expressions dialog. The DIST expression was chosen from the
“Functions” drop-down and the two points were added from the “Entities” drop-down. The expression was
completed, by adding the closed parenthesis from the dialog, or your keyboard.
When you are editing a real value in the data for any object, you can use an appropriately
dimensioned real object (Variable or Formula) instead of a constant. For example, creating or editing a
RadiusArc, one required data element is the Radius. When you select this item a constant or a real object
can be entered. Figure 4 shows a Radius Arc with the Radius field active and a radius of 1. To enter a
Formula, go to the Surfer View and choose the appropriate Formula. (Figure 5) Figure 6 shows Formula
‘f1’ as a parent, which has a value of 0.500.
Fig.2
Fig. 3
79
Fig. 4 Fig. 5
Fig. 6
80
Example Application
The following syntax generates the profile of an I-beam given four
parameters: Height, Width, FlgThk, WebThk (note how FlgThk, WebThk are
upper-bounded).
Variable Height / 100.000 L ( 10.000, 200 ) ;

Variable Width / 70.000 L ( 10.000, 150 ) ;
Variable FlngThk "flange thickness" / 20.000 L ( 0.000, Height ) ;
Variable WebThk / 20.000 L ( 0.000, Width ) ;
Formula d1 / { ( Width + WebThk ) / 2 } L ;
Formula d2 / { Height - FlngThk } L ;
Formula d3 / { ( Width - WebThk ) / 2 } L ;
FrameAbsPt BotFlgLLC 14 1 / * 0.000 0.000 0.000 ;
FrameAbsPt BotFlgLRC 14 1 / * Width 0.000 0.000 ;
FrameAbsPt BotFlgURC 14 1 / * Width FlngThk 0.000 ;
FrameAbsPt WebLRC 14 1 / * d1 FlngThk 0.000 ;
FrameAbsPt WebURC 14 1 / * d1 d2 0.000 ;
FrameAbsPt TopFlgLRC 14 1 / * Width d2 0.000 ;
FrameAbsPt TopFlgURC 14 1 / * Width Height 0.000 ;
FrameAbsPt TopFlgULC 14 1 / * 0.000 Height 0.000 ;
FrameAbsPt TopFlgLLC 14 1 / * 0.000 d2 0.000 ;
FrameAbsPt WebULC 14 1 / * d3 d2 0.000 ;
FrameAbsPt WebLLC 14 1 / * d3 FlngThk 0.000 ;
FrameAbsPt BotFlgULC 14 1 / * 0.000 FlngThk 0.000 ;
BCurve I_profile 11 1 12x1 / * 1
{ BotFlgLLC BotFlgLRC BotFlgURC WebLRC WebURC TopFlgLRC
TopFlgURC
TopFlgULC TopFlgLLC WebULC WebLLC BotFlgULC BotFlgLLC } ;
81

Advanced Modeling in MultiSurf 6.0

Uploaded by

Copyright:

Available Formats

Advanced Modeling in MultiSurf 6.0

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Advanced Modeling in MultiSurf 6.0

Uploaded by

Copyright:

Available Formats

MultiSurf 6.

0 Reference: Advanced Modeling

ADVANCED MODELING .....................................................................................4

Proximity Points ......................................................................................................................................... 11

Curves and Snakes ..................................................................................................................................... 14

NUBFitCurve, NUBFitSnake, NUBFitSurf entities................................................................................. 34

Parametric Solids ....................................................................................................................................... 55

Solid Entities ............................................................................................................................................... 57

Triangle Mesh Entity Types ...................................................................................................................... 63

Snakes on Triangle Meshes........................................................................................................................ 69

© 2006 AeroHydro, Inc.

BLENDPT3.MS2 at Lat 0, Lon 90 orthographic.

Characteristic data filename = filename of .3DA or .PAT file

Sample files Tangent Magnet.ms2

On layer 1, an image of the u,v parameter space has been constructed

Surface or magnet designates the host surface for the XYZMagnet. If a

Frame specifies the coordinate frame in which to measure X, Y or Z. The

Location1, Location2 are the two coordinate offsets.

"Signed distance" is a measure of the position of a point in relation

ProxBead3.ms2 This model contains 3 ProxBeads hosted on a

ProxRing1.ms2 This model has 2 ProxRings hosted on a BSnake.

ProxMagnet1.ms2 This model has a single ProxMagnet 'm1', used to

Curves and Snakes

Characteristic data type = B-spline type

polygon for BFitCurve

BFitCurve and basis curve

See also Fitted Surface

Hyperbola (type-2 and type-3)

CONIC.MS2 in <Y> view; Conic types 1 - 3.

CATENARY.MS2 in <Y> view.

See also Arc, ArcSnake, BCurve, CCurve

Characteristic data point1 ... pointN = control points

'p2' wt=1 'p2' wt=2 'p2' wt=5

Characteristic data point1, poin2, point3 = control points

The program’s default relabel produces uniform labeling with respect to

RADIUSARC1.MS2 at Lat -20, Lon 30.

RADIUSARC3-POLYCURVE2.MS2 at Lat 10, Lon 100.

Characteristic data filename = filename of .3DA or .PAT file

Example tabcurve.ms2 (needs demomcs.3da)

This example produces Z

Characteristic data point1, point2 ... pointN = control points

type-1 type-2 type-3

XCURVE.MS2 at Lat -30, Lon 60.

Characteristic data magnet1, magnet2, magnet3 = magnet or ring control points

ARCSNAKE.MS2 in <Y> view.

See also Arc

Characteristic data magnet1 ... magnetN = control points

• The snake lies in the same surface as its magnets.

If ‘hull’ is changed by modification of one or more of the master curves,

Characteristic data curvature = a geodesic curvature value (defined below)

GeoSnakes 'm0' v 0,0

Characteristic data magnet1 ... magnetN = control points and weights

NURB snake v 'csm2'

'end2' & 'M1'

See also PolyCurve, SubSnake

NUBFitCurve, NUBFitSnake, NUBFitSurf entities

Tools and NUBFit Surf

relabel is used to control the labeling of the v-direction (the lofting

FLOFT1.MS2 at Lat -30, Lon 70 (left).