interactive walkthrough in virtual cities and its applications in urban design
DESCRIPTION
Interactive Walkthrough in Virtual Cities and its Applications in Urban Design - Master's Thesis Defense Presentation - 2009 November 9 - Khaled Mohamed Ahmed Abd El Gawad - University of Alexandria - Institute of Graduate Studies and Research - Department of Information TechnologyTRANSCRIPT
University of Alexandria
Institute of Graduate Studies and Research
Department of Information Technology
Interactive Walkthrough in Virtual Cities
and its Applications in Urban Design
Master's Thesis Defense Presentation2009 November 92009 November 9
Khaled Mohamed Ahmed Abd El Gawad
Examination Committee:
Prof.Dr. Ayman El-Dessouky Ibrahim
Prof. Dr. Shawkat Kamal Guirguis
Dr. Walaa Mohamed Sheta
Supervisors:
Dr. Walaa Mohamed Sheta
Dr. Sahar El-sayed Bayoumi
Outline
� Motivation
� Objective
� Contribution
� literature review
5/28/2010 2
� Methodology
� Experiments and Results
� Future Work
Motivation
� 3D Virtual City Applications diversity.
� Applications in urban design.
� Egyptian non-structured cities.
5/28/2010 3
� Increasing size of 3D datasets makes them impossible
to display in real time with classical approaches.
Objective
� Improve visibility culling performance in non-structured
virtual city
�Study performance for the developed visibility
algorithm
5/28/2010 4
Main Contribution
� Developing potentially visible set (PVS) algorithm on
non-structured virtual city
� Providing a new technique to calculate offline 2d
visibility for virtual city based on user predefined
budget (accuracy, time and storage media size)
5/28/2010 5
budget (accuracy, time and storage media size)
� (PVS) algorithm over perform partitioning algorithm on
non-structured virtual city
Literature Review
� Virtual Bath (CASA group)
�Visibility culling (back-face, view-frustum)
�Occlusion culling algorithms
5/28/2010 6
View frustum
eye
View frustum
back
face
Occlusion
Literature Review cont.
�Potentially visible set Algorithm
� Run-time / pre-process calculating
� Indoor Scenes Cell and Portal
� Outdoor Scenes
5/28/2010 7
� Outdoor Scenes
Methodology
1. Building a virtual walkthrough prototype system for
both acceleration algorithms (PVS - partitioning)
2. Applying both algorithms on two virtual cities with
different urban patterns (non structured – well
planned)
5/28/2010 8
planned)
3. Measuring performance
4. Analyzing results and outcomes
Prototype Implementation
� System Architecture Overview
PVS
module
Partitioning
module
Make Partitions
5/28/2010 9
Make Partitions
Prototype Implementation cont.
� System Architecture Overview
PVS
module
Partitioning
module
Make Partitions
5/28/2010 10
Make Partitions
� 2D View Cell Analyzer
� View Cell Manipulator
Prototype Implementation: PVS Module
5/28/2010 11
Prototype Implementation: PVS Module
� 2D View Cell Analyzer� Non-structured city.
5/28/2010 12
Prototype Implementation: PVS Module
� 2D View Cell Analyzer� Well-planned city.
5/28/2010 13
� 2D View Cell Analyzer
� View Cell Manipulator
Prototype Implementation: PVS Module
5/28/2010 14
� 2D View Cell Analyzer
� PVS Calculating
Prototype Implementation: PVS Module
5/28/2010 15
Prototype Implementation: PVS Module
5/28/2010 16
� 2D View Cell Analyzer
� PVS Calculating
Prototype Implementation: PVS Module
5/28/2010 17
Prototype Implementation cont.
� System Architecture Overview
PVS
module
Partitioning
module
Make Partitions
5/28/2010 18
Make Partitions
� Walkthrough Generator 1
Prototype Implementation: PVS Module
5/28/2010 19
� Walkthrough Generator 1
Prototype Implementation: PVS Module
5/28/2010 20
� Walkthrough Generator 1� Routing Map
Prototype Implementation: PVS Module
5/28/2010 21
When the user
enters a view cell
Prototype Implementation cont.
� System Architecture Overview
PVS
module
Partitioning
module
Make Partitions
5/28/2010 22
Make Partitions
� Partitioning Tool
� Find Polygon/Partition Center
Prototype Implementation: Partitioning
Make Partitions
5/28/2010 23
� Partitioning Tool
� Find Polygon/Partition Center
Prototype Implementation: Partitioning
Make Partitions
5/28/2010 24
Prototype Implementation: Partitioning
Make Partitions
� Partitioning Tool
� Create Partitions
5/28/2010 25
Prototype Implementation: Partitioning
Make Partitions
5/28/2010 26
Prototype Implementation: Partitioning
Make Partitions
5/28/2010 27
Prototype Implementation cont.
� System Architecture Overview
PVS
module
Partitioning
module
Make Partitions
5/28/2010 28
Make Partitions
� Walkthrough Generator 2
Prototype Implementation: Partitioning
5/28/2010 29
Prototype Implementation cont.
� System Architecture Overview
PVS
module
Partitioning
module
Make Partitions
5/28/2010 30
Make Partitions
Outline so far:
� Motivation
� Objective
� Contribution
� literature review
5/28/2010 31
� Methodology
� Experiments and Results
� Future Work
Experiments and Results
� Dataset description
5/28/2010 32
Well planned : Al Ahram Garden
277 building
Non structured: Wasat El Delta
277 building
� Navigation Patterns
Experiments and Results cont.
5/28/2010 33
Random Walk
(RW)
Spiral Pattern
(SP)
Changed Circular Pattern
(CCP)
� Experiment #1
Sub Experiment 1-1 1-2 1-3 1-4 1-5
The view cell size against the number of loaded objects in the
non-structured urban pattern city when applying the PVS
algorithm
Experiments and Results cont.
5/28/2010 34
View Cell No. 2 4 8 16 32
View cell size 800 X 300 400 X 300 400 X 150 200 X 150 200 X 75
1-1 1-2 1-3 1-4 1-5
132.75134
182 201
161
150
200
250
Nu
mb
er
Of
Loa
de
d O
bje
cts
Non-Structured City – PVS Algorithm
Experiments and Results: Experiment #1
5/28/2010 35
112
132.75
116.875
81.4375
55.96875
9078
58
2626
112
0
50
100
800 X 300 400 X 300 400 X 150 200 X 150 200 X 75
Average No Of Objects
Min No. Of Objects
Max No. Of Objects
View Cell Size
Nu
mb
er
Of
Loa
de
d
� Experiment #2
The view cell size against the number of loaded objects in the
well planned urban pattern city when applying the PVS
algorithm
Experiments and Results cont.
Sub Experiment 2-1 2-2 2-3 2-4 2-5
5/28/2010 36
2-1 2-2 2-3 2-4 2-5
View Cell No. 2 4 8 16 32
View cell size 800 X 350 400 X 350 400 X 175 200 X 175 200 X 75
203
172.25160
246 246
219
206
162
200
250
300
Nu
mb
er
Of
Loa
de
d O
bje
cts
Well Planned City – PVS Algorithm
Experiments and Results: Experiment #2
5/28/2010 37
150.75
124.5625
98.3125
160
121
83
42
46
162
0
50
100
150
800 X 350 400 X 350 400 X 175 200 X 175 200 X 75
Average No. Of Objects
Min No. Of Objects
Max No. Of Objects
View Cell Size
Nu
mb
er
Of
Loa
de
d
250
300
No
. o
f lo
ad
ed
Ob
ject
s
Well planned Max No of objects
Non structured Max No of objects
Experiments and Results cont.
� Experiments #1 and #2 conclusions
5/28/2010 38
0
50
100
150
200
280k 240k 120K 140k 70k 60K 35k 30K 15k
No
. o
f lo
ad
ed
Ob
ject
s
View Cell Size
200
250
No
. o
f lo
ad
ed
Ob
ject
s
Well planned Avg no of objects
Non structured Avg no of objects
Experiments and Results cont.
� Experiments #1 and #2 conclusions
5/28/2010 39
0
50
100
150
200
280k 240k 120K 140k 70k 60K 35k 30K 15k
No
. o
f lo
ad
ed
Ob
ject
s
View Cell Size
140
160
180
No
. o
f Lo
ad
ed
Ob
ject
s
Well planned Min No of objects
Non structured Min No of objects
Experiments and Results cont.
� Experiments #1 and #2 conclusions
5/28/2010 40
0
20
40
60
80
100
120
140
280k 240k 120K 140k 70k 60K 35k 30K 15k
No
. o
f Lo
ad
ed
Ob
ject
s
View Cell Size
� Experiment #3
The view cell size against the frame rate in the well planned
urban pattern city when applying the PVS algorithm.
Experiments and Results cont.
Sub Experiment 2-1 2-2 2-3 2-4 2-5
5/28/2010 41
View Cell No. 2 4 8 16 32
View cell size 800 X 350 400 X 350 400 X 175 200 X 175 200 X 75
29.4907818129.96435728
30.13162235
28.68523507
29.84108181
24.93212933
22.7490917522.97267238
23.20343061
24
26
28
30F
ram
e R
ate
Experiments and Results: Experiment #3
5/28/2010 42
19.1571897719.63045629
18.46098643
19.20655074
16.91969015
21.15675006
22.7490917522.97267238
23.20343061
16
18
20
22
800 X 350 400 X 350 400 X 175 200 X 175 200 X 75
RW
CCP
SP
View Cell Size
Fra
me
Ra
te
Well planned - PVS
� Experiment #4
The view cell size against the frame rate in non structured
urban pattern city when applying the PVS algorithm
Experiments and Results cont.
Sub Experiment 1-1 1-2 1-3 1-4 1-5
5/28/2010 43
View Cell No. 2 4 8 16 32
View cell size 800 X 300 400 X 300 400 X 150 200 X 150 200 X 75
28.0897360727.52872754 27.50527764
22.78121455
29.24600028
29.7310093
28.76899778
30.99649029
23
28
33F
ram
e R
ate
Experiments and Results: Experiment #4
5/28/2010 44
16.73338993
14.44401665
17.54001673
19.89686771 19.8479581620.64370651
20.61034316
22.78121455
13
18
23
800X300 400X300 400X150 200X150 200X75
RW
CCP
SP
View Cell Size
Fra
me
Ra
te
Non structured- PVS
� Experiment #5
The partition size against the frame rate in well planned city
when applying the partitioning algorithm
Experiments and Results cont.
Sub Exp. 5-1 5-2 5-3 5-4 5-5 5-6
5/28/2010 45
Partitions No. 2 4 8 16 32 64
Partitions size 400X610 400X305 200X305 200X155 100X155 100X77
16.04837325
14.63680569
15.690212716.33234097
20.197540420.67608491
15.10331206
17.66874392
15
20
25F
ram
e R
ate
Experiments and Results: Experiment #5
5/28/2010 46
6.6358462647.16564891
7.9255450758.594371213 9.045524881 9.250304366
11.79780791
10.05555103
10.44367175
11.45429424
0
5
10
2 par 400X610 4 par 400X305 8 par 200X305 16 par 200X155 32 par 100X155 64 par 100X77
RW
CCP
SP
Partition Size
Fra
me
Ra
te
Well planned - Partitioning
� Experiment #6
The partition size and the frame rate in non structured urban
pattern city when applying the partitioning algorithm
Experiments and Results cont.
Sub Exp. 6-1 6-2 6-3 6-4 6-5 6-6
5/28/2010 47
Partitions No. 2 4 8 16 32 64
Partitions size 375X500 375X250 190X250 190X145 100X145 100X75
22.69119622
24.62832275
22.73839676
27.53329429
30.6624535
33.0433291633.47852866
27
32
Fra
me
Ra
te
Experiments and Results: Experiment #6
5/28/2010 48
11.43958488
9.332535195
12.97570919
16.03335947
22.69119622
16.3419815.31591728
18.54151373
20.94712329
22.27447554
23.2409737721.71500904
22.73839676
7
12
17
22
2 par 375X500 4 par 375X250 8 par 190X250 16 par 190X145 32 par 100X145 64 par 100X75
RW
CCP
SP
Partition Size
Fra
me
Ra
te
Non structured - Partitioning
City
TypePath type
Partitioning Algorithm Potential Visible Set (PVS)
2 par 4 par 8 par 16 par 32 par 64 par 2 Sens 4 Sens 8 Sens 16 Sens 32 Sens
structured RW 11.43958488 9.33254 12.9757 16.0334 22.6912 24.6283 16.7334 14.444 17.54 19.8969 19.84796
Experiments and Results cont.
� Results Matrix
5/28/2010 49
Non
structured
CCP 16.34198 15.3159 18.5415 20.9471 22.2745 23.241 20.6437 20.6103 28.0897 27.5287 27.50528
SP 21.71500904 22.7384 27.5333 30.6625 33.0433 33.4785 22.7812 29.246 29.731 28.769 30.99649
2 par 4 par 8 par 16 par 32 par 64 par 2 Sens 4 Sens 8 Sens 16 Sens 32 Sens
Well planned
RW 6.635846264 7.16565 7.92555 8.59437 9.04552 9.2503 19.1572 19.6305 18.461 19.2066 16.91969
CCP 16.04837325 14.6368 15.6902 16.3323 20.1975 20.6761 29.4908 29.9644 30.1316 28.6852 29.84108
SP 11.79780791 10.0556 10.4437 15.1033 11.4543 17.6687 24.9321 21.1568 22.7491 22.9727 23.20343
Conclusion and Discussion� The highest number of occluded objects and frame
rate can be obtained when applying PVS algorithm on
non structure city
� PVS algorithm over perform partitioning algorithm in
non structured city
5/28/2010 50
Future Work� Providing a 3d visibility technique based on user
predefined budget
� Studying performance for the developed algorithm
compared to other algorithms
� Enhance the developed algorithm to be used in virtual
5/28/2010 51
� Enhance the developed algorithm to be used in virtual
cities with dynamic objects
Thank You
Discussion
5/28/2010 52
Discussion
University of Alexandria
Institute of Graduate Studies and Research
Department of Information Technology
Interactive Walkthrough in Virtual Cities
and its Applications in Urban Design
Master's Thesis Defense Presentation2009 November 92009 November 9
Khaled Mohamed Ahmed Abd El Gawad
Examination Committee:
Prof.Dr. Ayman El-Dessouky Ibrahim
Prof. Dr. Shawkat Kamal Guirguis
Dr. Walaa Mohamed Sheta
Prototype Implementation
� Software Used
Modeling Phase
AutoCAD 3DS MAX Photoshop CS3
5/28/2010 54
AutoCAD 3DS MAX
Simulation Phase
Flash CS4
AS 3
VRML97 Vrmlpad Cortona player Scripting
Photoshop CS3
For further information
5/28/2010 55
For further information
see the following slides
Prototype Implementation cont.
� Interactive Walkthrough (well planned) PVS
5/28/2010 56
Prototype Implementation cont.
� Interactive Walkthrough (non structured) PVS
5/28/2010 57
Prototype Implementation cont.
� Interactive Walkthrough (Log file structure) PVS
5/28/2010 58
Tracking Loaded
View Cells
� Experiments environment
Operating System Windows XP Professional (5.1,
Build 2600) Service Pack 2
(2600.xpsp_sp2_gdr.070227-
Experiments and Results cont.
5/28/2010 59
(2600.xpsp_sp2_gdr.070227-
2254)
Processor Intel(R) Celeron(R) CPU
2.53GHz
Memory 478MB RAM
Display Devices SiS 661FX
Display Memory 32.0 MB
� Partitioning Tool� Non-structured city.
Prototype Implementation: Partitioning
Make Partitions
5/28/2010 60
� Partitioning Tool� Well-planned city.
Prototype Implementation: Partitioning
Make Partitions
5/28/2010 61
Conclusion and Discussion
15
20
25
Fra
me
Ra
teNon structured city (RW)
PVS
5/28/2010 62
0
5
10
280K 144K 70K 35K 15K
Fra
me
Ra
te
Partitioning
Conclusion and Discussion
PVS
20
25
30
Fra
me
Ra
teNon structured city (CCP)
5/28/2010 63
Partitioning
0
5
10
15
280K 144K 70K 35K 15K
Fra
me
Ra
te
Conclusion and Discussion
Partitioning
PVS
20
25
30
35
Fra
me
Ra
teNon structured City (SP)
5/28/2010 64
Partitioning
0
5
10
15
20
280K 144K 70K 35K 15K
Fra
me
Ra
te