Full text: Proceedings, XXth congress (Part 5)

   
RRR RNR 
one major disadvantage: creates stressful stereo pairs due to 
vertical parallax. The off-axis projection is more difficult to be 
implemented since it requires a non symmetric frustum which is 
not supported by all rendering packages but produces less 
stressful stereo pairs (fig. 2a). By default, OpenView uses the off- 
axis projection. But it is also possible to use the toe-in projection, 
too. 
Offaxis projection (Top view) Toe-in projection (Top view) 
Projection plane 
à 1 TE 7 
À 
/ 
  
  
L 1 ; 
1 1 1 
L ^ + 
VN ' ' ' 
A 1 1 i 
NN UM 
\ \ ) 3 1 1 t 
\ \ : | ^ f 
\ N 1 1 d / 
(Nan ens 
NN o f d j 
MEN AS ¢ 
\ vu A ; 
Y / / 
\ A y / 
' FN 
vi 
  
= 
> 
BA E 
— 
Tone 
mare 
npo te 
Left Center 
eue eue 
Left Center 
eue eue 
     
Figure 2: Off-Axis (correct) and Toe-in (incorrect) projections 
3.8 Creating 3D videos 
OpenView can be used in order to create 3d videos of the object 
or the scene. All OpenView's "head" cameras can save, on 
demand, single images or avi files of the rendered scene. Using 
this ability, creation of 3d videos is an easy task: OpenView is 
used to produce left and right images of the scene and saves them 
as image or avi files. Then an external application (3D Video 
Creator) imports them and produces the video. 
Additionally, the position and the rotation of every camera can be 
saved into a file and then stored as metadata to the video stream, 
in order to create geo-referenced 3d videos (Sechidis et al, 2001). 
4. PERFORMANCE TESTS 
In order to test OpenView's performance, several tests have been 
made, using two different CPU and graphics card combination 
setups and different virtual worlds. Next table shows the results 
of these tests, while additional details about tests follow: 
  
  
  
  
  
  
  
Speed (fps) for each display 
Setup ! Setup 2 
Loading test 1 20 (sec) 7 (sec) 
Test 1 3 6 
Test 2 150 200 
Test 3 130 170 
Test 4 20-30 55-60 
  
  
  
  
  
Tablel: Performance results for several tests 
Setup 1: Pentium 4. 1.8 GHz, 512M RAM, Matrox P750, 64Mb 
RAM 
Setup 2: Pentium 4 2.8 GHz, 1G RAM, Nvidia 5600 XT, 256 Mb 
RAM 
  
   
  
  
  
  
     
  
   
    
   
  
  
   
    
  
   
    
   
   
    
   
   
   
  
   
    
  
   
  
   
  
  
    
  
  
   
   
   
    
   
   
   
  
  
  
   
  
    
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004 
Testl: One surface (453 x 526 points grid) with a 3392 x 3935 
pixels, RGB ortho image of Thessaloniki (as shown in Fig. 1). 
Test 2: An archaeological site comprised of about one thousand 
polygons, using texture materials. 
Test 3: The site of the test2, loaded 10 times in different positions 
Test 4: A 3ds file with 67392 vertices (58928 faces) having 1214 
objects and 15 different textures. 
Both left and right displays were 1024 x 768 pixels wide, while 
Control room was about 400 x 300 pixels wide in all tests. 
As shown above, OpenView's performance was good enough for 
most of the tests, except the first one. In that test, the one big 
surface is responsible for low performance since it had to always 
be rendered, no matter of the viewer's position and view angle. 
The solution to this (which is under construction) is to split the 
one big surface to a lot of smaller ones. 
5. CONCLUSIONS - FURTHER WORK 
OpenView is a generic tool for stereoscopic representations of 3D 
objects or VR scenes. It is common sense that a specialized tool 
for presentations of archaeological sites has different 
requirements from an equivalent of medical applications or from 
a tool that will present VR worlds for entertainment. However, 
OpenView can be used with success in all above applications 
even if it has the minimal requirements that these applications 
require. 
Performance tests have shown that it can be used for real time 
presentations, even with the usage of low cost CPU systems. 
The further growth of OpenView will continue be focused in the 
implementation of capabilities that will have generic character. 
Thus, in future versions they will be added animation, 3D (or 5.1) 
sound, internet abilities (client - server), ability of creating scenes 
using two computers, possibility of measurements and motion 
that will be based on physics (ODE). For more specialized 
requirements, the team of OpenView hopes in the help of its 
users. Since the source code will be available, is given the 
occasion in each one to face and to resolve each own problem. All 
the new solutions will be incorporated in the OpenView so as to it 
becomes more powerful, useful and functional. 
Additionally OpenView, coded in Delphi and using public 
domain or shareware components, depends on GLScene library, 
an active open-source project; therefore any improvement on 
GLScene or other component performance will improve 
OpenView, too. 
References 
Ogleby, C., 1996, A reconstruction of the ancient city of 
Ayutthaya using modern Photogrammetric techniques, IAPRS, 
Vol. XXXI, Part B5, Com. V, Vienna, 1996, pp. 416-425. 
Ogleby, C., 2001a, *The ancient city of Ayutthaya-Explorations 
in virtual reality and multi-media", Proc. of International 
Inte 
Work 
of Cu 
Ogle 
Olym 
Work 
of Cu 
Sabry 
E. Pa 
Textu 
Realit 
Sechi 
Video 
Herita 
ISSN 
Refer 
Burke 
http:// 
er/ (ac 
Lischl 
http:// 
script‘ 
http:// 
Autho 
comm 
* [SIS 
State
	        
Waiting...

Note to user

Dear user,

In response to current developments in the web technology used by the Goobi viewer, the software no longer supports your browser.

Please use one of the following browsers to display this page correctly.

Thank you.