CS3VR16 : Graphics

Lecture 5 – Aut 22
CS3VR16 – Virtual Reality – Graphics - 5
p1 RJM 17/08/23
© Prof Richard Mitchell 2023
 Components for 3D
Graphics
Remaining Topics
 Representation
• Planar Surfaces
• Constructive Solid Geometry
• Solids/Voxel modelling
 Camera Models/Scene Views
• Perspective
 Shading and Lighting
• Shading models

CS3VR16 – Graphics - 5
p2 RJM 17/08/23
© Prof Richard Mitchell 2023
 Representation
A scene can contain different type of objects
Various representation models are available
 Polygonal surfaces and quadrics
 Spline surfaces
 Solid modeling
 Volumetric models
 Procedural models
 Physics based modeling

CS3VR16 – Graphics - 5
p3 RJM 17/08/23
© Prof Richard Mitchell 2023
 Representation
Many ways to represent a surface
Each has advantages/disadvantages
3D
Models
Non-Polygonal Polygonal
Models Models

Constructive Parametric
Structured Polygon
Solid Surfaces
Soups
Geometry
Implicit
Surfaces Convex Non Convex

CS3VR16 – Graphics - 5
p4 RJM 17/08/23
© Prof Richard Mitchell 2023
Planar Surface Modelling: Polygon

Polygon mesh:
 vertex, edges and polygons
 each edge shared (max 2
polygons)
• vertex: point (x,y,z)
• edge: line joining two vertices
• polygon: closed seq. of edges

CS3VR16 – Graphics - 5
p5 RJM 17/08/23
© Prof Richard Mitchell 2023
 PSM - Polygonal
 These can be represented as
• Explicit
• Pointers to list of vertices
• Pointers to list of edges
 Criteria to evaluate different
representations
• Time
• Space
• Topological Information

CS3VR16 – Graphics - 5
p6 RJM 17/08/23
© Prof Richard Mitchell 2023
 PSM - Explicit
Explicit Representation  Advantages
 Each P list of vertex coords • Efficient rep. for
each polygon
𝑃 1 = { ( 𝑥1 , 𝑦 1 , 𝑧 1 ) ,... , ( 𝑥 𝑛 , 𝑦 𝑛 , 𝑧 𝑛 ) }

𝑃 2 = { ( 𝑥3 , 𝑦 3 , 𝑧 3 ) , ..., ( 𝑥 𝑚 , 𝑦 𝑚 , 𝑧 𝑚 ) }
 Disadvantages
 Vertices stored in order • High storage
 Shared vertices cost
duplicated • To move
 No explicit representation vertex,
traverse all Ps
for shared vertices /edges
• Shared edges
drawn twice
CS3VR16 – Graphics - 5
p7 RJM 17/08/23
© Prof Richard Mitchell 2023
 PSM – Pointers to vertex
list
Pointers to list of  Advantages
vertices • Each V stored
 Each V stored once in once
list • Coords easily
changed
 P = list of V lists  Disadvantages
• Difficult to find
𝑃 1 = { 𝑉 1 ,𝑉 2 ,𝑉 4 }
Ps that share an
𝑃 2 = { 𝑉 4 , 𝑉 2 ,𝑉 3 } V2 edge
V1 P 1 V3 • Shared edges
P2 drawn twice
V4
CS3VR16 – Graphics - 5
p8 RJM 17/08/23
© Prof Richard Mitchell 2023
 PSM Pointers to Edge list
Pointers to list of edges  Advantages
 Again, a list of vertices • Each V stored
 P list of indexes to E once
• Shared edges
list
drawn once
 Edges point to vertices • V coords easily
and polygons belong changed
to  Disadvantages
V2 • Diff. to find
E1 edges which
E2
V1
P1 P2 V3 share a vertex
E5 E4
E3
V4
CS3VR16 – Graphics - 5
p9 RJM 17/08/23
© Prof Richard Mitchell 2023
 PSM – Polygonal Meshes
 Polygonal Meshes – types of
 Triangle Strip
• n vertices,
• (n-2) connected triangles
 Triangle Fan
• n vertices,
• (n-2) connected triangles
 Mesh of Quadrilaterals
• (n-1)  (m-1) quadrilaterals
• for n  m vertices
CS3VR16 – Graphics - 5
p10 RJM 17/08/23
© Prof Richard Mitchell 2023
 PSM Polygons - Characteristics
 Many polygons, often triangles / quads
 Concave or Convex Shapes
 Varying levels of topological information:
none (soups) to complex connectivity
graphs

Convex Concave

CS3VR16 – Graphics - 5
p11 RJM 17/08/23
© Prof Richard Mitchell 2023
 PSM P - Characteristics
Characteristics
 Re modelling solid shapes..
 Poor for smooth curves,
 need huge numbers of polygons

CS3VR16 – Graphics - 5
p12 RJM 17/08/23
© Prof Richard Mitchell 2023
 PSM – P - Issues
Issues to consider with models
• Which order to draw
• Do polygons collide?
• Relevant in animation
• How make algorithms efficient
• Concave / Convex affect tests
Convex objects quicker

CS3VR16 – Graphics - 5
p13 RJM 17/08/23
© Prof Richard Mitchell 2023
 PSM – Convex Polygons
Convex Polygons
 An object is convex iff
 “A line segment between any two points on an
object is on or within its boundary”

 Two convex objects can only ever touch at one

point or in one plane – very useful

CS3VR16 – Graphics - 5
p14 RJM 17/08/23
© Prof Richard Mitchell 2023
 PSM – Polygon Collision
Polygons – collision test
 Separating Axis Theorem (SAT)

 → Highly optimized algorithms for convex polys

 Non-convex models much harder
 No separating axis, but still no contact

CS3VR16 – Graphics - 5
p15 RJM 17/08/23
© Prof Richard Mitchell 2023
 PSM – Polygonal
Non-Convex Polygons
 So use convex decomposition

 Heuristic search methods may not work

 ‘Brute-Force’ methods can be slow
 Geometric coherence can speed up ..
 Ensure maintained as objects moved

CS3VR16 – Graphics - 5
p16 RJM 17/08/23
© Prof Richard Mitchell 2023
 GC – 2 phases re checks
 Broad phase – put bounding boxes round
objects
 (smallest box in which shape fits)
 Narrow phase – do test only if
bounding boxes over lap
 Objects comprising many elements
divide into tree of BBs

CS3VR16 – Graphics - 5
p17 RJM 17/08/23
© Prof Richard Mitchell 2023
 Representation
More ways to represent a surface
3D
Models
Non-Polygonal Polygonal
Models Models
Constructive Parametric
Solid Surfaces Structured Polygon
Geometry Soups
Implicit
Surfaces Convex Non Convex

CS3VR16 – Graphics - 5
p18 RJM 17/08/23
© Prof Richard Mitchell 2023
Non-Polygonal Models – Implicit
Surfaces
 Non-polygonal techniques to represent shapes
 smoother,
 less memory consuming
 But distance and intersection much harder
 An implicit surface defined by locus of points at
 f(x,y,z) = 0
 f is smooth
 and has a non zero gradient at any point

CS3VR16 – Graphics - 5
p19 RJM 17/08/23
© Prof Richard Mitchell 2023
NPM – Implicit Surfaces - Sphere
 Example – sphere
 A sphere is a function in x, y and z such that:

What is its origin and radius?

 If put x,y,z in LHS, test sign of result
 -ve : inside sphere; +ve outside
 If f low degree polynomial
 then collision detection is often simple.
 Higher degree functions much more complex

CS3VR16 – Graphics - 5
p20 RJM 17/08/23
© Prof Richard Mitchell 2023
 NPM – ISE - another
 Implicit Surface Example

[x 2
+ y + z − a k ] − b [ ( z −k ) − 2x ][ ( z+ k ) − 2 y ] = 0
2 2 2 2 2 2 2

CS3VR16 – Graphics - 5
p21 RJM 17/08/23
© Prof Richard Mitchell 2023
 NPM – Parametric Surfaces
 2D planar shape mapped to 3D coordinates
 → a parametric surface (PS)
 PS need not be a topological or closed surface
 Generally used to define a surface boundary
 Hence, parametric objects are usually made up
of many surface ‘patches’

CS3VR16 – Graphics - 5
p22 RJM 17/08/23
© Prof Richard Mitchell 2023
 NPM – Parametric Surfaces
 Non-Uniform Rational B-Splines widely used
 Finding intersection between two not trivial
 Precision affected by processing time
 Often not suitable for real-time simulation
 Tessellate into polygons, then collision detect

CS3VR16 – Graphics - 5
p23 RJM 17/08/23
© Prof Richard Mitchell 2023
NPM – Parametric Surfaces Example

 Parametric Surface Example

CS3VR16 – Graphics - 5
p24 RJM 17/08/23
© Prof Richard Mitchell 2023
 Solid Modelling - CSG
Constructive Solid Geometry
 Start with set of primitives
 Blocks, cones, cylinders, spheres, etc
 These can be scaled/rotated
 Create new solid using logical
operations
 Resultant shape can be scaled, rotated

CS3VR16 – Graphics - 5
p25 RJM 17/08/23
© Prof Richard Mitchell 2023
 Solid Modelling – CSG
Constructive Solid Geometry
 Use binary tree to define

Union Intersection Difference

CS3VR16 – Graphics - 5
p26 RJM 17/08/23
© Prof Richard Mitchell 2023
 Solid Modelling – CSG
 Union () … Intersection () … and Difference
()

Union Intersection Difference

  -
Logical OR Logical AND Logical NOT

CS3VR16 – Graphics - 5
p27 RJM 17/08/23
© Prof Richard Mitchell 2023
 Solid Modelling – CSG – eg
1
 Example 1

CS3VR16 – Graphics - 5
p28 RJM 17/08/23
© Prof Richard Mitchell 2023
 Solid Modelling – CSG – eg
2
 Example 2

CS3VR16 – Graphics - 5
p29 RJM 17/08/23
© Prof Richard Mitchell 2023
 Solid Modelling – CSG – eg
3
 Example 3

CS3VR16 – Graphics - 5
p30 RJM 17/08/23
© Prof Richard Mitchell 2023
 Solid Modelling – CSG
Advantages
Powerful, high level commands
Easy to construct solid model
Concise description – less memory
Easy to assure objects solid or ‘water tight’
Disadvantages
Only Boolean operations : restricts possible
shapes
Comp. expensive to find boundaries, faces,
edges
Care ensure generate
p31 RJM 17/08/23 valid
© Prof shapes
CS3VR16 – Graphics - 5
Richard Mitchell 2023
 Solid Modelling – CSG
Applications
 CSG has a number of practical uses:
• Where simple geometric objects are desired
• Where mathematical accuracy is important
• Many graphics engines use CSG
• There are also libraries that add CSG to Unity
 CSG popular
 Important in manufacturing / engineering

CS3VR16 – Graphics - 5
p32 RJM 17/08/23
© Prof Richard Mitchell 2023
 Solid Modelling – CSG Workflow
 Different rules …

CS3VR16 – Graphics - 5
p33 RJM 17/08/23
© Prof Richard Mitchell 2023
 Solid Modelling – CSG Code
 PovRay – CSG models from code
 Example:
cylinder { <0,-4,0> <0,4,0> 0.2 pigment { Yellow } }
difference {
box { <-1.5,-0.5,-0.5> <1.5,0.5,0> 0.1 }
cylinder { <-0.8,0,-3> <-0.8,0,3> 0.1 }
cylinder { <0.8,0,-3> <0.8,0,3> 0.1 }
pigment { Red }
}
 Persistence Of Vision Raytracer – www.povray.com

CS3VR16 – Graphics - 5
p34 RJM 17/08/23
© Prof Richard Mitchell 2023
Solid Modelling - Extrusion
 Extrusion
• Useful if rotational or translational
symmetries
• Use 2D primitive and path in 3D Primitiv
space e

Linear
Extrusio
n

Circular
Extrusion
CS3VR16 – Graphics - 5
p35 RJM 17/08/23
© Prof Richard Mitchell 2023
 Solid Modelling – Octtrees
 Spatial Numbering (Octtrees)/Voxel
Modelling
 Hierarchical tree structure, each node is 3D region
 These regions are divided in octants (cubes),
 8 elements are stored in each node of the tree
 Nodes can be of 3 types:
 white (out the object);
 black (inside the object)
 Grays (neither inside nor outside)
 Individual elements of 3D space are called voxels
CS3VR16 – Graphics - 5
p36 RJM 17/08/23
© Prof Richard Mitchell 2023
 Solid Modelling – Octtrees
 Divide vol into smaller vols where need more
detail
 Example

CS3VR16 – Graphics - 5
p37 RJM 17/08/23
© Prof Richard Mitchell 2023
 Summary
In this lecture we have considered different ways of
representing surfaces and models
• Planar Surfaces
• Constructive Solid Geometry
• Solids/Voxel modelling
We can now model 3D different scenes
But we need to draw them on a 2D screen
So we need to think about perspective …

CS3VR16 – Graphics - 5
p38 RJM 17/08/23
© Prof Richard Mitchell 2023