The Central Idea

A surface mesh is a collection of connected polygonal elements used to represent a surface.

The basic ingredients are:

The central geometric principle is:

\[ \boxed{\text{many simple local pieces can represent one complex surface}}. \]

Vertices

A mesh vertex is a point in space.

In three dimensions, a vertex may be stored as

\[ \boxed{\mathbf v_i=(x_i,y_i,z_i)}. \]

The full mesh is built by connecting these vertices according to a chosen combinatorial structure.

Edges

An edge joins two vertices.

If vertices are labeled

\[ v_i \]

and

\[ v_j, \]

an edge can be represented combinatorially by the pair

\[ \boxed{(i,j)}. \]

Faces

A face is a polygon whose boundary follows mesh edges.

A triangular face with vertices

\[ v_i,\quad v_j,\quad v_k \]

can be represented by the ordered triple

\[ \boxed{(i,j,k)}. \]

Triangle Meshes

A triangle mesh is a mesh whose faces are all triangles.

Triangle meshes are especially common because any polygon can be triangulated and three noncollinear points always determine a plane.

Why Triangles Are So Useful

Triangles are geometrically simple.

They have several useful properties:

Polygon Meshes

A mesh does not have to use triangles.

It may contain:

Such structures are often called polygon meshes.

Quadrilateral Meshes

A quad mesh uses primarily four-sided faces.

Quad meshes are useful when geometry has strong directional flow or when surface parameter lines are important.

Unlike triangles, four arbitrary spatial points need not lie in one plane, so nonplanar quads require interpretation or triangulation.

Tessellation

A tessellation covers a region using repeated shapes without gaps or unintended overlaps.

In the plane, familiar tessellations use:

On curved surfaces, tessellation generally uses polygonal patches that approximate or partition the surface.

Mesh Versus Tessellation

The two ideas overlap but emphasize different things.

A mesh emphasizes a connected data structure of vertices, edges, and faces.

A tessellation emphasizes the covering or subdivision of a region by pieces.

Adjacency

Adjacency records which mesh elements touch.

We may ask:

This connectivity is as important as the coordinates themselves.

Geometry and Topology of a Mesh

A mesh has two different kinds of information.

Geometry includes:

\[ (x_i,y_i,z_i). \]

Topology includes:

Same Topology, Different Geometry

Two meshes can have exactly the same connectivity but different vertex coordinates.

One may form a cube-like shape and another a distorted hexahedral form.

Their combinatorial structure is the same even though their metric geometry differs.

Face Orientation

The ordering of a face's vertices determines an orientation.

For a triangular face

\[ (A,B,C), \]

reversing the order to

\[ (A,C,B) \]

reverses its normal direction.

Face Normal

For triangle \(ABC\), define

\[ \mathbf u=\overrightarrow{AB}, \qquad \mathbf v=\overrightarrow{AC}. \]

Then

\[ \boxed{\mathbf n=\mathbf u\times\mathbf v} \]

is perpendicular to the face.

Unit Normal

If

\[ \mathbf n\neq\mathbf0, \]

then the unit normal is

\[ \boxed{ \widehat{\mathbf n} = \frac{\mathbf n}{\|\mathbf n\|} }. \]

Unit normals are useful for orientation, shading, geometric analysis, and comparison of neighboring faces.

Triangle Area

The area of triangle \(ABC\) is

\[ \boxed{ A= \frac12 \|\overrightarrow{AB}\times\overrightarrow{AC}\| }. \]

This formula works directly in three-dimensional coordinates.

Total Mesh Area

If a triangle mesh has face areas

\[ A_1,A_2,\ldots,A_m, \]

then its total polygonal surface area is

\[ \boxed{ A_{\text{mesh}} = \sum_{i=1}^{m}A_i }. \]

Approximating a Curved Surface

A smooth surface can be approximated with many small flat faces.

The approximation usually improves when:

Resolution

Mesh resolution describes how finely a surface is subdivided.

Higher resolution usually means:

More Faces Do Not Automatically Mean Better Geometry

A large number of poorly shaped or poorly placed elements can still give a poor approximation.

Mesh quality depends on both:

\[ \boxed{\text{resolution}} \]

and

\[ \boxed{\text{element quality and placement}}. \]

Refinement

Refinement increases local mesh resolution.

One simple method subdivides a triangle into smaller triangles.

Refinement may be:

Uniform Refinement

Uniform refinement applies the same subdivision rule throughout the mesh.

If every triangular face is split into four smaller triangles, then after one refinement step

\[ F_{\text{new}}=4F_{\text{old}} \]

when considering face count alone.

Adaptive Refinement

Adaptive refinement concentrates additional elements where they are most useful.

More resolution may be added near:

Subdivision Surfaces

A subdivision surface is generated by repeatedly refining a coarse control mesh according to a rule.

The limiting shape can become smooth even though each finite stage is polygonal.

This connects recursive geometry with smooth surface modeling.

Control Mesh and Limit Surface

The initial polygonal structure is the control mesh.

Repeated subdivision produces finer meshes.

The mathematical limiting object is the limit surface.

Tessellating a Sphere

A sphere cannot be tiled exactly by finitely many flat polygons without replacing the smooth surface by a polyhedral approximation.

One strategy is:

  1. start with an inscribed polyhedron such as an icosahedron;
  2. subdivide its triangular faces;
  3. project new vertices outward onto the sphere.

Why an Icosahedron Is Useful

An icosahedron has \(20\) triangular faces distributed relatively evenly around a sphere.

Subdivision of those triangles produces a convenient nearly uniform spherical mesh.

Projection Back to the Sphere

Suppose a new vertex is at

\[ \mathbf p. \]

To move it onto a sphere of radius \(R\) centered at the origin, use

\[ \boxed{ \mathbf p_{\text{sphere}} = R\frac{\mathbf p}{\|\mathbf p\|} }. \]

Approximation Error

A flat mesh face does not exactly match a curved surface patch.

The difference between the polygonal mesh and the target surface is an approximation error.

Error can be measured in several ways, including:

Chord Versus Arc

A mesh edge approximating a curved path acts like a chord replacing an arc.

For sufficiently small segments, the chord and arc become close.

This is the local geometric reason refinement can improve approximation.

Curvature and Mesh Density

High-curvature regions bend more sharply.

They often require smaller mesh elements to achieve the same geometric accuracy as low-curvature regions.

This connects mesh design directly to the earlier study of curvature.

Boundary Edges

In an open surface mesh, a boundary edge belongs to only one face.

Boundary edges connect to form one or more boundary curves.

Interior Edges

In an ordinary manifold-like triangle mesh, an interior edge belongs to exactly two faces.

Those two faces meet along the shared edge.

Nonmanifold Edges

If more than two faces meet along one edge, the mesh is nonmanifold there.

Such structures can be useful in special models but do not locally behave like an ordinary smooth surface near that edge.

Vertex Neighborhood

The faces and edges adjacent to a vertex form its local neighborhood.

In a well-formed interior manifold vertex, neighboring faces arrange cyclically around the vertex.

Valence

The valence of a mesh vertex is the number of incident edges.

In a regular triangular tessellation of a flat plane, an interior vertex commonly has valence

\[ \boxed{6}. \]

Why Six Equilateral Triangles Fit Around a Flat Vertex

Each equilateral triangle has angle

\[ 60^\circ. \]

Therefore,

\[ 6(60^\circ)=360^\circ. \]

The triangles fill the neighborhood without angular gap or overlap.

Angle Defect

For a polyhedral surface, the angle defect at an interior vertex can be defined as

\[ \boxed{ \delta = 2\pi-\sum_i\theta_i }. \]

Here the \(\theta_i\) are the face angles meeting at that vertex.

Interpreting Angle Defect

If

\[ \delta>0, \]

the local geometry is often positively curved in a discrete sense.

If

\[ \delta<0, \]

the local configuration has saddle-like excess angle.

If

\[ \delta=0, \]

the local angular geometry is flat.

Discrete Curvature

Meshes replace smooth curvature fields with discrete geometric approximations.

Angle defects, changing face normals, and local fitting methods provide several ways to approximate curvature.

Dihedral Angle

Two adjacent faces meet along an edge.

The angle between their planes is related to the dihedral angle.

Large changes between neighboring face normals indicate sharper local bending across that edge.

Normals and Shading

A flat triangle has one constant face normal.

To create the visual impression of a smoother surface, computer graphics may interpolate vertex normals across the face.

The underlying geometry is still polygonal even when the shading appears smooth.

Flat Shading Versus Smooth Shading

Flat shading emphasizes each polygonal face separately.

Smooth shading interpolates normal information to soften visible transitions.

Shading changes appearance, not the actual mesh geometry.

Regular Tessellations of the Plane

Using one kind of regular polygon, only three regular tessellations of the Euclidean plane exist:

Why Regular Pentagons Do Not Tile the Plane Alone

A regular pentagon has interior angle

\[ 108^\circ. \]

No integer multiple of \(108^\circ\) equals

\[ 360^\circ. \]

Therefore, regular pentagons alone cannot meet around a vertex to form a regular Euclidean tessellation without gaps or overlap.

Semiregular Tessellations

Tessellations may use more than one type of regular polygon while keeping a consistent vertex arrangement.

These are often called semiregular or Archimedean tessellations.

Tessellations on Curved Surfaces

Curved surfaces impose different angular constraints from the plane.

A spherical tessellation, for example, can use arrangements impossible in a flat Euclidean plane.

Geodesic-Like Meshes

Subdivided polyhedra projected onto spheres create geodesic-like spherical meshes.

Their edges are generally straight chords in the polyhedral model rather than true spherical geodesic arcs unless specifically interpreted that way.

Mesh Simplification

Mesh simplification reduces the number of elements while attempting to preserve important shape features.

It is the opposite goal of refinement:

\[ \boxed{\text{fewer elements with acceptable geometric error}}. \]

Why Simplify a Mesh?

A lower-resolution mesh can reduce:

Feature Preservation

Simplification should preserve important geometry whenever possible.

Features may include:

Remeshing

Remeshing changes the mesh structure while trying to preserve the represented surface.

Goals can include:

Mesh Quality

A triangle can have the correct position but still be numerically or geometrically undesirable if it is extremely thin.

Good meshes often avoid:

Degenerate Triangles

A triangular face is degenerate when its three vertices are collinear or repeated so that its area is zero.

Algebraically,

\[ \overrightarrow{AB}\times\overrightarrow{AC} = \mathbf0. \]

Closed Meshes

A closed surface mesh has no boundary edges.

Every ordinary edge belongs to exactly two faces.

Closed meshes can represent boundaries of solids.

Open Meshes

An open mesh has one or more boundary curves.

Examples include:

Euler Characteristic of a Mesh

For a polyhedral surface,

\[ \boxed{\chi=V-E+F}. \]

For a closed sphere-like mesh,

\[ \chi=2. \]

Different topologies can have different Euler characteristics.

Mesh Topology Can Change

Adding a handle, creating a hole, closing a boundary, or splitting a component can change the topology of a mesh.

Such operations may change

\[ V-E+F \]

in systematic ways.

Triangulation Preserves Surface Structure

Dividing one polygonal face by a diagonal adds:

Therefore,

\[ \Delta(V-E+F)=0-1+1=0. \]

The Euler characteristic is unchanged.

Meshes and Parametric Surfaces

A parametric surface

\[ \mathbf r(u,v) \]

can be sampled on a parameter grid.

Nearby sample points can then be connected into triangles or quadrilaterals to form a mesh approximation.

Parameter-Space Tessellation

Suppose the \((u,v)\)-domain is divided into small rectangles.

Each rectangle can be mapped through

\[ \mathbf r(u,v) \]

to a curved patch and approximated by one or more polygons in space.

Meshes and Implicit Surfaces

An implicit surface has the form

\[ F(x,y,z)=0. \]

Computational methods can approximate such a surface with polygons by locating where the equation changes sign or crosses the target level inside a spatial grid.

Iso-Surfaces

More generally,

\[ F(x,y,z)=c \]

is a level surface.

A polygonal approximation to such a level surface is often called an iso-surface mesh.

Meshes and Fractals

Recursive refinement can generate increasingly intricate meshes.

Fractal-like surface models may add detail at multiple scales, connecting tessellation with the previous lesson on self-similar form.

Meshes and Numerical Methods

Meshes are essential in computational mathematics.

They divide complicated domains into manageable elements for methods such as:

Common Error: Assuming a Mesh Is the Exact Smooth Surface

A polygon mesh usually approximates a smooth target surface.

The two coincide exactly only in special cases or by definition.

Common Error: More Faces Always Means Better

Increasing face count can improve resolution, but poor element placement or bad shapes can still produce poor quality.

Resolution and quality must be considered together.

Common Error: Smooth Shading Means Smooth Geometry

Smooth shading can make a coarse mesh look visually smooth.

The actual geometry remains polygonal unless the vertices or surface representation are changed.

Common Error: Every Edge in a Closed Mesh Is a Boundary Edge

In a well-formed closed mesh, ordinary edges are interior to the surface structure and belong to two faces.

Boundary edges occur only when the surface itself has a boundary.

Common Error: Every Polygonal Face Is Automatically Planar

Triangles are always planar.

A polygon with four or more vertices can have noncoplanar vertices in three dimensions.

Such a face may need triangulation or another geometric convention.

Common Error: Ignoring Face Orientation

Inconsistent vertex ordering can cause neighboring normals to point in conflicting directions.

Consistent orientation is important for closed surfaces, rendering, and geometric computations.

Common Error: Confusing Tessellation with Arbitrary Overlap

A tessellation is intended to cover a region without gaps or unintended overlaps.

Arbitrarily scattered polygons do not automatically form a tessellation.

A Reliable Procedure for Analyzing a Surface Mesh

  1. Identify the vertices, edges, and faces.
  2. Determine whether faces are triangles, quads, or mixed polygons.
  3. Check adjacency and connectivity.
  4. Identify boundary edges and components.
  5. Check face orientation.
  6. Compute normals and areas when needed.
  7. Check for degenerate or poorly shaped elements.
  8. Compare mesh resolution with target curvature and feature scale.
  9. Check topology using quantities such as \(V-E+F\).
  10. Decide whether refinement, simplification, or remeshing is appropriate.

Example: Two-Triangle Square Mesh

Let the square have vertices

\[ A=(0,0,0), \quad B=(1,0,0), \quad C=(1,1,0), \quad D=(0,1,0). \]

Triangulate using diagonal \(AC\).

The faces are

\[ \triangle ABC \]

and

\[ \triangle ACD. \]

Then

\[ V=4,\qquad E=5,\qquad F=2. \]

Boundary of the Square Mesh

The outer edges

\[ AB,\quad BC,\quad CD,\quad DA \]

each belong to one face and are boundary edges.

The diagonal

\[ AC \]

belongs to both triangles and is an interior edge.

Practice 1: Triangle Area

Let

\[ A=(0,0,0), \quad B=(2,0,0), \quad C=(0,3,0). \]

Find the area of triangle \(ABC\).

Practice 2: Face Normal

Using the same triangle, find a normal vector.

Practice 3: Boundary Edge

In an ordinary open triangle mesh, how many incident faces does a boundary edge have?

Practice 4: Interior Edge

In an ordinary manifold-like triangle mesh, how many incident faces does an interior edge have?

Practice 5: Uniform Refinement

A mesh has \(40\) triangular faces.

Each face is subdivided into \(4\) smaller triangles.

How many triangular faces result?

Practice 6: Flat Triangular Valence

How many equilateral triangles meet around an interior vertex in a regular flat triangular tessellation?

Practice 7: Angle Defect

Five equilateral triangles meet at a vertex.

Find the angle defect.

Practice 8: Euler Characteristic

A closed mesh has

\[ V=12,\qquad E=30,\qquad F=20. \]

Find

\[ \chi=V-E+F. \]

Practice 9: Degenerate Triangle

What happens to the triangle area when its three vertices are collinear?

Practice 10: Sphere Projection

A point

\[ \mathbf p=(1,2,2) \]

is projected radially onto a sphere of radius \(6\) centered at the origin. Find the projected point.

Practice Answers

Practice 1

We have

\[ \overrightarrow{AB}=\langle2,0,0\rangle, \] \[ \overrightarrow{AC}=\langle0,3,0\rangle. \]

Their cross product is

\[ \langle0,0,6\rangle. \]

Therefore,

\[ A= \frac12(6) = \boxed{3}. \]

Practice 2

One normal vector is

\[ \boxed{\langle0,0,6\rangle}. \]

Any nonzero scalar multiple is also normal to the face.

Practice 3

A boundary edge belongs to

\[ \boxed{1} \]

face.

Practice 4

An ordinary interior edge belongs to

\[ \boxed{2} \]

faces.

Practice 5

\[ 40\cdot4=\boxed{160}. \]

Practice 6

Since

\[ 6(60^\circ)=360^\circ, \]

the answer is

\[ \boxed{6}. \]

Practice 7

The angle sum is

\[ 5\left(\frac{\pi}{3}\right) = \frac{5\pi}{3}. \]

Therefore,

\[ \delta = 2\pi-\frac{5\pi}{3} = \boxed{\frac{\pi}{3}}. \]

Practice 8

\[ \chi = 12-30+20 = \boxed{2}. \]

Practice 9

The cross product of two edge vectors is zero, so the triangle has

\[ \boxed{\text{area }0}. \]

Practice 10

First,

\[ \|\mathbf p\| = \sqrt{1^2+2^2+2^2} = 3. \]

Therefore,

\[ \mathbf p_{\text{sphere}} = 6\frac{(1,2,2)}{3} = \boxed{(2,4,4)}. \]

What to Remember

The central habit is: separate the surface's continuous geometry from the discrete mesh used to represent it, then analyze how resolution, connectivity, and local element shape affect that representation.

Where We Go Next

Meshes and tessellations provide discrete ways to represent and analyze spatial form. The final lesson in this sequence asks how different curves and surfaces can be organized according to the geometric properties they preserve or exhibit.

In Classification through Geometric Properties, we will compare forms through dimension, curvature, symmetry, boundedness, connectedness, smoothness, orientability, convexity, and other structural invariants.