1. Lecture 3. Virtual Worlds : Representation,Creation and Simulation ( II )
고려대학교 그래픽스 연구실

2. Contents Representation of Virtual Worlds Creation of virtual Worlds
Simulation of Virtual Worlds
Simulation of Virtual Agents

3. Creation of Virtual Worlds
{geometry, physics, behavior}
1. 3D Modeling and Layouting
2. Vision-based Techniques
3. Algorithmic Creation
4. Multimedia and Virtual World Authorings

4. 1. Three-D modeling Methods for modeling 3D object geometry
graphics library
tedious, requires skill
CAD package
issue: CAD file format --> VR system’s representation
commercial 3D database
issue: polygonal mesh --> parts/objects
3D digitizer
issue: sparse set of 3D points --> polygonal mesh
VR editor

5. 2. Vision-based techniques
Various ‘Shape-from-X’ techniques
Passive methods
stereo vision
model-based vision
multi-view analysis
Active methods
laser range finding
structured lights

6. 4. Virtual world authoring
Multimedia authoring
CD-Rom titles
Web pages
Animation authoring
Game authoring
Virtual world authoring

7. Virtual world authoring environment
single authoring environment vs. a collection of tools
3D model builder
audio source and editor
device drivers / library
runtime module
simulation
rendering
application builder

8. Application builder similar in spirit to the many 3D GUI builder.
some useful features
built-in, extensible object type system and browser
direct scene graph manipulation
object property editor
script language (interpreted, but compilerable)
access to external codes
importing a variety of CAD formats
support for a wide variety of devices
What can we learn from the 2D GUI builder?

9. Simulation of Virtual Worlds
H-sensor
perception
cognition
motion control
H-effector
P-effector
P-sensor
L-effector
L-sensor
sensing
action
Human
Virtual environment
avatar
virtual
agent
object
V-effector
V-sensor

10. 1. Kinematics Simulation
(e.g., collision detection)
2. Dynamics Simulation
(e.g., collision response)
3. Behavioral Simulation
4. Management of the Distributed Virtual Worlds

11. Simulation is problem-dependent. Common issue: “N-body problem”
given n objects, it takes O(n(n-1)) computations.
Approaches
(1) Distributed processing
(2) “Multi-modeling” or Dynamic LoD

12. Dynamics LoD Select a simulation model as the function of … example
dynamic states of simulation
distance to the user, viewing condition
distance between objects, etc.
example
rigid body --> kinematics/dynamics simulation --> point-mass
Automatic generation of dynamic models in multiple levels is still elusive.

13. Collision detection Introduction Basic Algorithm and Its Problems
Categorizing of previous methods
State of the Art

14. Simulation of Virtual Agents
1. Virtual Body
2. Motion Control Techniques
3. Generation of Facial Expression
4. Case Study: Existing Systems
Improv
Jack

15. 1. Virtual Body Representations Geometric representations
appearance (geometry)
kinematics
dynamics
Geometric representations
skeleton
muscles
skin
hair

16. Example - Humanoid (Thalmann)
3 levels of detail (2,000 ~ 40,000 polys)
skeleton - meta-ball - body contours - skin with polys
74 DOFs (body) + 30x2 DOFs (hand)

17. 2. Motion Control Techniques
Key-frame animation
Inverse kinematics
Forward dynamics
Inverse dynamics
Motion capture

18. 3. Generation of Facial Expression
Brief History
70’s
Gouraud (1971)
demonstrated shading technique using his wife’s head model
Chernoff (1971)
computer generated 2D face drawings
Gillenson (1973)
interactive system for 2D line drawn facial images
Ekman (1977)
developed FACS(Facial Action Coding System)
Parke ( )
Ivan Surtherland’s CG course at Univ of Utah, M.S. thesis.
developed 3D facial animation system

19. 80’s and later Parke (1980) Platt (1981) Brennan (1982) Waters (1987)
developed the 3D parameterized facial animation system
Platt (1981)
developed the physically based muscle controlled facial model
Brennan (1982)
2D facial caricatures
Waters (1987)
developed the muscle model
Terzopoulos (1990)
developed the dynamic model using lattice network
Wang (1994)
developed the hierarchical layered spline model
etc.

20. FACS What is FACS? Definition of some facial knowledge
46 AUs (Action Units)
qualitative motion of skin which makes a unique facial action
6 primary emotional expressions
happy, sad, surprise, fear, anger, disgust
widely used for facial animation systems

21. Action Units (AU)
Example of AUs

22. Modeling methods Libraries of expression and interpolation
Parametric models
using expression parameters
Physically-based models
using dynamic equation
Anatomy-based models
Integration of additional features
texture map, hair modeling, etc.

23. Parametrized Model [Parke 80]
Approach
face model includes “parameter set”
Basic operation
procedural construction
interpolation
rotation/scaling
position offset

24. Muscle Model [Waters 87] Approach Muscle
not the exact simulation of skin, muscles, etc.
mimic the primary biomechanical characteristics.
Muscle
can be grouped by orientation of muscle fibers
linear / parallel
Elliptical / circular
Sheet / flat

25. Dynamic model [Terzopoulos 90]
using lattice network (volume network)
three layers : skin, fat, muscle(& bone)

26. Spline Model [Wang 94] Hierarchical layered spline modelling
local area surface deformation
Level 2
Level 1
Level 3
Level 4

27. 4. Case Study: Existing Systems
Humanoid (Thalmann)
Improv (Perlin)
Jack (Badler)
Hierarchical Virtual Characters (Bruderlin)