PRACTICAL DESIGN AND IMPLEMENTATION OF A CAVE

SYSTEM

High Quality Four Walls CAVE Howto

Achille Peternier, Sylvain Cardin

Virtual Reality Laboratory (VRLab), Ecole Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland

Frédéric Vexo, Daniel Thalmann

Virtual Reality Laboratory (VRLab), Ecole Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland

Keywords: CAVE, low cost, immersive Virtual Reality, calibration, tracking, stereographic rendering.

Abstract: CAVE systems are nowadays one of the best Virtual Reality (VR) immersive devices available for render-

ing virtual environments. Unfortunately, such kind of hardware is extremely expensive, complex and cum-

bersome, thus limited in its spread. Several cheaper solutions already exist, but they implement usually only

a subset of features of a professional CAVE. In this paper we describe how we have built a low cost CAVE

with four screens (three walls and a floor), stereographic rendering and user tracking by only using hard-

ware commonly available on the market and free software, we show the different solutions and work-around

we implemented to solve the problems we encountered and we conclude with an evaluation of our system

by using two applications we developed with it.

1 INTRODUCTION

Immersive environments giving the illusion of being

surrounded by a fictive world are a key feature re-

quired by many Virtual Reality applications and are

extremely difficult to simulate. Visual immersion

needs specific, expensive and cumbersome hard-

ware, such as head-worn displays (HMDs), large

displays or CAVE systems (Sutcliffe et al., 2006).

HMDs offer a good level of immersion, but often

suffer of a small field of view and isolate the user

and his/her body both from the real and the virtual

world (Czernuszenko et al., 1997). Spatially Immer-

sive Displays (SIDs), like wall-displays and CAVEs,

have the advantage of being multi-user, to allow

persons to be physically within the virtual environ-

ment and feature a wide field of view. Many studies

(Tyndiuk et al., 2005) (Buxton et Fitzmaurice, 1998)

showed that devices based on large displays offer a

better immersion and cognitive interaction with a

virtual 3D environment. Due to the high cost of pro-

fessional solutions and their complexity, the prolif-

eration of such kind of environments is limited to

institutes or organizations able to pay and manage

such structures.

In this paper we describe how we have built a

four sides CAVE (three walls, one floor) by using

standard market products and internally developed

software, without requiring any third part or high-

end professional product. Despite of this, we

achieved to manage a very flexible, robust, high

quality and fast CAVE environment, featuring

stereographic rendering, a good calibration system

for walls and sensors, head-tracking and last genera-

tion graphics comparable to most recent video

games. We expose how we managed every aspect

and solved all the accuracy and practical problems

of this project in order to show how home-made

SIDs can be created with a minimal effort and in a

relatively cheap way, without sacrificing the quality.

We compare our results with other similar frame-

work and establish a fair evaluation. Finally, we

illustrate some applications we developed with our

system to show what it is possible to handle with our

environment.

2 RELATED WORK

The CAVE Automatic Virtual Environment (or sim-

ply CAVE) has been originally conceived in 1992

129

Peternier A., Cardin S., Vexo F. and Thalmann D. (2007).

PRACTICAL DESIGN AND IMPLEMENTATION OF A CAVE SYSTEM - High Quality Four Walls CAVE Howto.

In Proceedings of the Second International Conference on Computer Graphics Theory and Applications - AS/IE, pages 129-136

DOI: 10.5220/0002080701290136

 SciTePress

by Thomas Defanti and Dan Sandin and imple-

mented by Carolina Cruz-Neira at the University of

Illinois at Chicago (Cruz-Neira et al., 1992). The

idea behind this project was to create a VR system

without the common limitations of previous VR so-

lutions, like poor image resolution, inability to share

the experience directly with other users and the iso-

lation from the real world. A head tracking system is

used to produce the correct stereo perspective: this

allows the user to see his/her entire environment

from the correct viewpoint, thus creating a compel-

ling illusion of reality. Real and virtual objects are

blended in the same space and the user can see

his/her body interacting with the environment.

Over the past decade, interest and development

of CAVEs and SIDs have become increasingly sig-

nificant, some of them becoming also commercial

products like Barco (www.barco.com) and VRCO

(www.vrco.com). Despite of the amount of devel-

opment in this direction, both professional and cus-

tomized solutions are still extremely expensive, be-

cause they are often close to prototypes requiring

specific cares, like the CAVE-like display developed

by Gross et al. (Gross et al., 2003) to project and

acquire 3D video content.

There also exist reduced versions of CAVE sys-

tems, with fewer walls or even transportable. One of

them is the V-CAVE (made with only two walls,

hence the name V-CAVE, because the two screens

form a letter “vee”). Two digital projectors point

into the corner of a room, avoiding the requirement

of dedicated screens (Jacobson, 2003). This system

(Jacobson et Hwang, 2002) is also based on the top

of a game engine which offers good quality graphics

on personal computers but sacrifices the versatility

of their approach for context other than a walk-

through of static pre-processed models, thus reduc-

ing the use of their software in contexts with ex-

tremely dynamic graphics. We aimed at a more

complex system with four walls and a more generic

engine in order to be used in contexts also requiring

extremely dynamic geometries or modifications to

the scene graph.

Different approaches have been studied for track-

ing the user position inside the CAVE. The recurrent

problematic depends on the physical characteristics

of the system. The most used indoor tracking sys-

tems are based on active vision algorithms or on

magnetic tracking. Both of these solutions are costly

and have specific drawbacks. The vision based ap-

proach with markers, like the Vicon system, uses

video acquisition from cameras and is sensible to

obstructions. In a CAVE framework, the field of

view of the cameras has to be wide enough to cover

a large area of interaction, thus multiple cameras are

necessary to improve tracking accuracy and in order

to cover a larger area. There also exist cheaper im-

plementations using passive vision based systems

with passive markers and standard cameras, running

on free library such as ARtoolkit and good enough

to produce accurate results in stereo acquisition

(Koo et al, 2004).

Magnetic tracking, using systems such as the

MotionStar

one (www.ascensiontech.com), is

widely used for full body motion tracking. The main

disadvantage in using this technology is that most of

the CAVEs are built on iron frames and metallic

masses may alter the magnetic field measured by the

sensors. Research at the Illinois University (M.

Ghazisaedy et al., 1995) offers a good illustration of

the magnetic field distortions. They also present a

method to calibrate the system to improve the accu-

racy, by correlating the measurements with ultra-

sonic sensors.

Sauter described a low-cost CAVE solution

based on generic Windows and Macintosh com-

puters to make this technology more accessible

(Sauter P. M., 2003). We adopted a similar but up-

dated architecture for our framework. We also ex-

tended the low-cost idea to the tracking system and

by improving calibration tools to solve common

problems of home-made virtual devices. Based on

these experiences, the next part will describe our

system and its different implementations.

3 SYSTEM

In this section we describe our solution, first with a

brief overview of the whole thing then by explaining

every aspect in detail.

3.1 System Overview

Our CAVE features three walls and a floor. We used

eight off the shelf LCD beamers (Canon LV-7210)

connected to four personal computers (Intel Dual

Xeon 3 GHz, with dual DVI output NVidia Geforce

7800), generating and projecting images on a cinema

screen folded in the form of a box (for the walls) and

on a white wooden panel (on the ground).

Back-projection has been used for the walls, di-

rect one for the floor. A fifth master PC leads the

four clients, through a server-client architecture con-

nected via a private 1 Gigabit LAN. The server PC

also manages a 5.1 audio system to bring spatial

sounds in the environment.

The system works both in mono and stereo-

graphic rendering, either active (shutter glasses) or

passive (red and blue glasses). The CAVE is 2.20

meters high, 2.50 meters large and 1.80 meters

GRAPP 2007 - International Conference on Computer Graphics Theory and Applications

130

depth. Up to three users can comfortably stay within

the framework.

Software side, our environment uses a graphic

engine specifically developed in our lab: server and

clients run local instances of the same engine syn-

chronized through the server. Synchronization in-

cludes models, scene graph, textures, shaders, etc.

thus obtaining a full dynamic environment.

Figure 1: Hardware setup overview.

3.2 Graphic and Audio System

Our CAVE is running a modified version of an in-

ternally developed graphic engine first created for

pedagogical purposes and called MVisio (Peternier,

2006).

MVisio (Mental Vision) is a lightweight, robust

and user-friendly 2D/3D graphic engine. The goals

of MVisio are several: offering an extremely easy to

use and intuitive interface to 2D/3D graphics, being

able to run on almost every available desktop PC or

laptop by automatically adapting the rendering qual-

ity and settings, being also able to work on hetero-

geneous devices (PC, PocketPC) or operative sys-

tems (Windows and Linux), being compact in sizes

and system requirements and finally being fast and

modern (featuring advanced rendering techniques

such as soft shadows and post-processing effects like

bloom lighting through OpenGL and OpenGL|ES).

Because of the flexibility and already tested robust-

ness of this software, we decided to add CAVE ren-

dering to its features.

MVisio for CAVE uses a server-client architec-

ture based on low-level TCP/IP communication be-

tween wall and floor client PCs and the main server

computer. Thanks to the speed of the local Gigabit

LAN the latency generated by the network is ex-

tremely low. Client machines run on a minimal in-

stallation of Windows XP. A service runs perpetu-

ally waiting for connections from an MVisio for

CAVE server. MVisio server sends a wake-up re-

quest to all the services running on the different cli-

ent machines, the service starts an MVisio local cli-

ent which is an interpreter of high-level commands

(move, rotate, display, etc.) sent from the server.

Every operation effectuated on the server PC is for-

warded to the clients. This include data loading (tex-

tures and geometries are synchronized at startup),

shaders, modifications to the scene-graph, etc. Run-

ning MVisio on a local PC or in the CAVE just re-

quires for the end-user to modify a couple of lines of

code during initialization of the graphic engine: eve-

rything else is automatically handled by our soft-

ware. This task can also be skipped by using con-

figuration files, in order to allow to exactly the same

source code to switch from the single PC version to

the CAVE version of MVisio in a completely trans-

parent way, by just passing a different configuration

file as argument. MVisio CAVE clients can be con-

sidered as copies of the MVisio engine running re-

motely and directly manipulated by the user: infor-

mation is synchronized through the network connec-

tivity between the machines. The different frusta are

computed server side and independently forwarded

to the specific clients. The architecture is completely

versatile and can handle from one remote client up

to any arbitrary number of displays. It is interesting

to mention that our architecture can also be used as

remote rendering system to display real-time images

on a remote PC somewhere in the world and not

being a part of the CAVE private network. This fea-

ture also theoretically allows data synchronization

between two or more remote CAVEs.

The audio engine has also been developed in our

laboratory. It features spatial audio positioning

through the OpenAL API (www.openal.org), audio

streaming (from .WAV and .OGG files) as well as

DSP effects through Creative EAX 2.0 (devel-

oper.creative.com). The main advantage of using our

own audio engine is that it perfectly fits with MVi-

sio, so adding audio sources to a scene graph is

straightforward and robust. Contrarily to the graphic

subsystem, the audio engine doesn’t require to be

instanced on each CAVE client and runs only on the

server PC that is connected to the five Dolby 5.1

loudspeakers (see figure 1). Audio and images are

synchronized by using the server timer.

3.3 CAVE Display Calibration

We used a market level flat home-cinema screen as

display for the three walls and a white painted

wooden panel as floor. To fold the home-cinema

screen in order to create a box surrounding the user

we used a transparent nylon wire stretched along the

corners of an iron frame. This solution caused a ma-

jor drawback: the screen folded with a curve around

the corners, assuming the shape of a parenthesis.

PRACTICAL DESIGN AND IMPLEMENTATION OF A CAVE SYSTEM - High Quality Four Walls CAVE Howto

131

Also projecting images from two separate projectors

generates an unavoidable misalignment, despite an

accurate calibration of their convergence.

Figure 2: Non calibrated walls, notice the glitch between

the two displays (left red box) and the grid not aligned for

stereographic rendering (right red box).

Figure 3: Walls after calibration.

We managed to solve these problems by mean-

ings of calibration software, allowing the user to

manually draw a shape accurately following the

screen borders and by using this shape as a polygon

to render the images into. Basically, the CAVE per-

forms a high resolution (1024x1024 pixels) render-

to-texture using the OpenGL frame-buffer object

extension. This texture is then mapped onto the

polygon previously adapted to fit the display shape.

To improve image superposition for stereo render-

ing, we also used a grid with moveable control

points. By matching the control points of the grids

projected from the two projectors, we avoided de-

formations caused by beamers being physically pro-

jecting from two points slightly different in the

space. An accurate calibration showed per-pixel ac-

curacy and almost entirely removed the corner ef-

fect.

We developed the calibration software by reus-

ing parts of the MVisio for CAVE engine, thus cre-

ating again a server-client architecture. To simplify

and accelerate the calibration procedure, a user just

has to run this software and access the different grids

and control points by using a wireless joystick. This

way, the four walls can be calibrated at the same

time by one user from within the CAVE, instead of

having to login on every CAVE computer separately

and adjust the grids with a mouse, one by one.

Figure 4: Dual-projector assembly with shutters.

This calibration solution also solved another

problem. Our projectors have a maximal resolution

of 1024x768 pixels. In order to cover the side wall

according to the space available, we had to place

them vertically, thus having different pixel resolu-

tions between the main and floor screens (768 pixels

vertically) and the right and left ones (1024 pixels).

Moreover, pixel sizes on the CAVE walls are not the

same on each wall, because of the different distances

between the projectors and the displays. Anti-

aliasing combined with an accurate match of the

render-to shapes made this problem almost not no-

ticeable. By using these techniques, we managed to

achieve a continuous image on the different walls

which creates a great immersion. To improve it a

step further, in the next part we describe how we

added depth information to the virtual scene through

stereoscopic rendering.

3.4 Stereographic Rendering

Stereographic rendering requires two images to be

generated and displayed at the same time, one com-

puted for the right and one for the left eye. Top-level

solutions use a high refresh rate CRT projector fea-

turing up to 120 Hz as refresh frequency. Through

the use of shutter glasses, such systems can display a

maximum of 60 images per second per eye. The

video projector has to be driven by a professional

graphic card which includes a synchronization out-

GRAPP 2007 - International Conference on Computer Graphics Theory and Applications

132

put for the shutter glasses. This output changes the

polarity of the shutter glasses according to the verti-

cal synchronization of the display. Professional

graphic cards and high quality CRT projectors are

very expensive and require specific periodical cares

to correct convergence.

We used a different approach in our system by

adopting two standard LCD projectors (one per eye)

with a built-in refresh rate of about 60 Hz (through a

DVI cable). To achieve proper superposition of the

two rendered images we built home made supports

for the projectors. These supports allow fine adjust-

ment of the position and orientation of the beamers.

Figure 5: Blue & red stereo in the CAVE.

The idea is to let the two LCD beamers continu-

ously showing the images and synchronizing left and

right eyes by putting ferroelectric shutters in front of

the LCD lenses. This way, user shutter glasses syn-

chronized with the ferroelectric shutters in front of

the projectors allow a fixed number of images per

second per eye, independently from the refresh rate

of the display devices. In our configuration we used

a switching between left and right eyes at 70 Hz

which is the limit of the shutter glasses we adopted.

We also implemented an old-style blue & red

glasses stereographic system to be used during pub-

lic demonstrations when a large number of visitors

access the CAVE at the same time, because of the

low amount and fragility of the shutter glasses. Fi-

nally, blue & red glasses can also be used for stereo-

graphic testing on the virtual CAVE (see 3.6) on a

single computer.

Stereographic rendering is improving immersion

by adding depth information to the surrounding im-

age. But while the user is moving, the perspective

referential should be accordingly displaced to avoid

break in presence. The next part will present our

solutions to correct this effect by tracking the user

head.

3.5 Head Tracking, Environment

Walkthrough

Head tracking is a key feature in a CAVE system

required to correctly generate the illusion of a sur-

rounding space around a specific user. Head-

tracking is used to determine the position of the user

within the CAVE physical space, in order to know

his/her distance from the walls and compute a cor-

rect projection.

Figure 6: The cheapest solution: blue & red glasses with a

head-wearable lantern. Detail picture on the left shows the

lantern in the darkness.

For testing purposes we implemented three dif-

ferent methods to track the user’s head position. The

first method we studied was to use a camera and a

vision based algorithm to track markers located on

the user’s head. The open source library ARtoolkit

provided all the necessary functions to treat the

video information in real time and extract 3D posi-

tion and orientation of the predefined markers. We

used a standard Panasonic video camera for the ac-

quisition. The main issue in the setup is the weak-

ness of the vision based algorithms to luminosity

changes. At first we had very poor results due to the

main source of illumination on the markers coming

from the image displayed on the CAVE itself. This

illumination of the markers was changing frequently

and degrades the tracking accuracy. To solve this

problem, a light bulb has been placed inside a small

cube covered with markers fixed on the user head,

creating some kind of head-worn lantern with semi-

transparent markers on the sides, to keep their lumi-

nosity constant and independent from the brightness

of the images rendered on the CAVE walls. This

system provided low tracking resolution at low re-

fresh rate. With proper calibration and filtering it

provides nice results to adjust to small user move-

ments.

PRACTICAL DESIGN AND IMPLEMENTATION OF A CAVE SYSTEM - High Quality Four Walls CAVE Howto

133

As second tracking method we used magnetic

trackers. These sensors are composed by three in-

ductances which measure the magnetic field induced

by a referential field generator. The whole system is

built by Ascension technologies under the name Mo-

tion Star system. This system provides 6 degree of

freedom tracking with a really high resolution for

the orientation and around 1 cm of accuracy for the

positioning. The main advantage of this system is

the high refresh rate with a correct accuracy.

We recently acquired an optical tracking system

with active markers using multiple viewpoints. This

system is composed by a set of linear video cameras

which keep track of different LEDs which blinks

according to their identifier. A complete dedicated

hardware is in charge of the acquisition from multi-

ple view points and computing of the positions of

the markers. This system provides great accuracy

around a few millimetres in positioning, which gives

perfect results for our kind of applications.

To simplify calibration of sensors within the

CAVE space and generalize the conversion between

the values returned and the CAVE coordinates, we

developed a simply and fast method using reference

images projected on the CAVE walls and three laser

pointers. We built a three orthogonal axis rigid sup-

port to put three laser pointers onto with a place

where to lock the sensors on its origin. By aligning

the three laser dots on a reference grid displayed on

the walls and the floor, it is possible to easily match

the reference points for converting sensor spatial

coordinates to CAVE coordinates.

3.6 Virtual CAVE

Developing for a CAVE environment often requires

the programmers to be near the hardware and to test

directly on the device. This solution isn’t always

practical, mainly when different concurrent projects

share the same CAVE at the same time.

Figure 7: Bar scene displayed in the CAVE simulator. On

the left, the white sphere represents the user’s head, on the

right images as viewed from the user’s position.

To make development of CAVE-based applica-

tions independent from the physical device, we de-

veloped a software CAVE simulator, running ex-

actly like the real one but in a system window. The

virtual CAVE also features blue & red stereographic

rendering, thus enabling a very accurate reproduc-

tion of the stereographic images that will be gener-

ated on the real device.

Virtual Cave aims at providing an easy testing

tool for developers, for draft evaluation of the sys-

tem. In order to provide a global fair evaluation, the

next part will discuss the test and results of our sys-

tem.

4 EVALUATION

We based this evaluation on some concrete applica-

tions we developed in the laboratory with our frame-

work as case of study.

4.1 Applications

The first application is a videogame developed as

semester projects by students in our laboratory. The

goal of this project was to create software accessing

all the functionalities offered by our platform. Stu-

dents developed a first person shooting game where

the user holds a position tracked toy gun as input

device for aiming at enemies. This game has also

been used as demo to show to guests and visitors

coming in our lab.

The second application used our framework to dis-

play virtual humans animated in real-time through

an inverse kinematics algorithm. Virtual humans

reacted by imitating the postures of a user staying

within the CAVE and wearing led-based optical

markers.

Figure 8: Virtual human miming the movement effectu-

ated by the user through an IK algorithm.

GRAPP 2007 - International Conference on Computer Graphics Theory and Applications

134

4.2 User Feedbacks

The first application gave us a large amount of feed-

back from a very heterogeneous amount of users,

ranging from primary school students to Computer

Graphics specialists.

Every user has been so far surprised by the qual-

ity both of the rendering and displaying of the im-

ages and found the illusion of being surrounded by a

virtual environment convincing. Some of them com-

plained about the black border around the cinema

screen and cutting continuity between the wall dis-

plays and the floor (see figures with CAVE images).

We are trying to raise the floor panel in order to

avoid this break in presence.

Users disliked long sessions with blue & red

glasses because of the strong ghost effect showing

part of the right image on the left eye and vice versa.

4.3 Discussion

Our system showed a good global luminosity even

when used with ferroelectric shutters and shutter

glasses. Despite the significant amount of luminosity

lost through the shutters and the retro projection, the

final images feature all the details, even if slightly

dimmed. User tests also reported that bloom lighting

improved the sensation of luminosity, without

changing anything at the hardware level.

Our calibration software (either for walls and

sensors) offered a robust workaround to manage the

irregularities of the screen display and the glitch

between projected images. Our approach also

showed is usefulness to quickly correct some mis-

matching that may occurs due to dilatation of the

support according to CAVE room temperature varia-

tions. Very small modifications on the beamer sup-

ports may rapidly become a few pixels misalignment

when projected on the display: our software allowed

a user to correct them in a couple of minutes.

We used relatively small ferroelectric shutters

which caused a significant drawback: overheating.

When closed, shutters block a high amount of light

which raises their temperature. When the CAVE is

active for long-time sessions (more than one hour, as

we noticed during public demos with the first appli-

cation), special cares have to be taken in the account,

like some kind of screen-saver or heating monitor-

ing. Alternatively, blue & red glasses can be used

and the shutters removed. We mounted them on a

magnetic support to rapidly switch between the two

modalities (see fig. 4: red disks are magnets).

All the different tracking systems we tested are

suitable for application but each one with some spe-

cific limitations. ARtoolkit markers offer the cheap-

est solution but the lack of dedicated hardware, the

sensitivity to the variable illumination, limitation of

the field of view and occlusions due to the use of a

single camera make the resultant tracking a bit un-

stable and a refresh rate around a few Hz. With

proper software filtering this solution offers accept-

able results in a restricted area.

The magnetic tracking measurement is very ac-

curate in orientation but the positioning is fairly poor

in our case. In fact these types of sensors use the

attenuation of the magnetic field to measure the dis-

tance to the referential generator. The main problem

is coming from the fact that magnetic fields are per-

turbed by metallic masses. Since the CAVE frame

responsible of the screen support is made of metal,

the usable area of this system is limited as a meter

cube around head’s high in the center of the CAVE

with an accuracy of about a couple of centimetres.

Figure 9: Stereographic rendering.

The multiple camera optical tracking system of-

fers an amazing accuracy around a few millimetres

after a proper calibration and a fair range of effec-

tiveness. The main disadvantage is that it requires

camera located on the floor corner and multiple ac-

tive markers on the user head but nothing compared

to the bulky ARToolKit lantern.

On the graphic engine side, generating two im-

ages (left and right) per frame on a single PC may

seem computational expensive, but it wasn’t the

case. Modern graphic cards can process geometry

extremely quickly and thanks to extensions like the

frame buffer object, render to texture come almost

for free. Moreover, we were limited by the relatively

low projector resolution of 1024x768, thus sparing

some filling rate to be used for anti-aliasing or image

post-processing (Gaussian filtering, bloom lighting).

The most expensive feature we implemented is soft-

shadowing. We used a shadow map-based algorithm

which requires an additional pass for every light

source, thus making two additional passes when in

PRACTICAL DESIGN AND IMPLEMENTATION OF A CAVE SYSTEM - High Quality Four Walls CAVE Howto

135

stereographic mode. We also used very high resolu-

tion light maps (2048x2048 texels) to improve video

quality. All these graphic improvements may stress

the hardware and rapidly killing the framerate. De-

spite of this, the bar scene used in our test had about

15.000 triangles, high resolution textures and still

ran between around 25 fps in stereographic mode,

with one light source casting soft shadows and

bloom lighting activated. It is important to mention

that our engine is entirely dynamic, so that you can

completely change the scene at run time, but nulli-

fies different optimizations that could be used to

significantly speed up the rendering procedure (con-

trarily to most gaming engines which often pre-

generate an optimized environment but static). Our

engine also features an almost direct use of models

exported from 3D Studio MAX though a proprietary

plug-in, without requiring any special care or con-

version: other CAVE solutions based on videogames

graphic engines, even if faster, put usually more

constraints about that.

5 CONCLUSION

In this paper we exposed the different approaches

we applied during the development of a high quality

stereographic CAVE from the scratch. We showed

how it is possible to build a very good system with-

out requiring professional equipment, thanks to the

high standard of quality of recent common market

level products and a bit of practical sense. Software

side, we created our framework by readapting por-

tions of code from an existing graphic engine, rap-

idly getting a robust, performing and complete solu-

tion in a relatively short time.

Our system has shown its versatility and quality

on real applications and user tests, and also goes in

the direction evoked by Hibbard (Hibbard, 2000)

about three major constraints limiting fruition of VR

immersive devices: our framework offers a rela-

tively cheap solution, fast sensor response and easi-

ness of use on real contexts.

In the next phase we will refine our system and

use it in a wider range of both scientific and busi-

ness-oriented projects, in order to extend the fruition

of this framework to other areas requiring immersive

content visualization.

REFERENCES

Sutcliffe, A., Gault, B., Fernando, T., and Tan, K. 2006.

Investigating interaction in CAVE virtual environ-

ments. ACM Trans. Comput.-Hum. Interact. 13, 2

(Jun. 2006), 235-267

Tyndiuk, F., Thomas, G., Lespinet-Najib, V., and Schlick,

C. 2005. Cognitive comparison of 3D interaction in

front of large vs. small displays. In Proceedings of the

ACM Symposium on Virtual Reality Software and

Technology, Monterey, CA, USA (Nov. 2005).

Buxton, B. and Fitzmaurice, G. W. 1998. HMDs, Caves &

chameleon: a human-centric analysis of interaction in

virtual space. SIGGRAPH Comput. Graph. 32, 4 (Nov.

1998), 69-74.

Czernuszenko, M., Pape, D., Sandin, D., DeFanti, T.,

Dawe, G. L., and Brown, M. D. 1997. The Immer-

saDesk and Infinity Wall projection-based virtual real-

ity displays. SIGGRAPH Comput. Graph. 31, 2 (May.

1997), 46-49.

Cruz-Neira, C., Sandin, D.J., DeFanti, T.A., Kenyon, R.,

and Hart, J.C. 1992. The CAVE, Audio Visual Experi-

ence Automatic Virtual Environment. Communica-

tions of the ACM, June 1992, pp. 64-72.

Sauter, P. M. 2003. VR

Go™: a new method for virtual

reality development. SIGGRAPH Comput. Graph. 37,

1 (Feb. 2003), 19-24.

Gross, M., Würmlin, S., Naef, M., Lamboray, E., Spagno,

C., Kunz, A., Koller-Meier, E., Svoboda, T., Van

Gool, L., Lang, S., Strehlke, K., Moere, A. V., and

Staadt, O. 2003. Blue-c: a spatially immersive display

and 3D video portal for telepresence. ACM Trans.

Graph. 22, 3 (Jul. 2003), 819-827.

Jacobson, J. 2003. Using “CAVEUT” to Build Immersive

Displays With the Unreal Tournament Engine and a

PC Cluster, http://citeseer.ist.psu.edu/639123.html

Jacobson, J., Hwang, Z. 2002. Unreal Tournament for

Immersive Interactive Theater. Communications of

the. ACM. 45, 1 (2002), 39-42.

Peternier, A., Thalmann, D., Vexo, F., Mental Vision: a

computer graphics teaching platform, In Lecture Notes

in Computer Science, Springer-Verlag Berlin, 2006

Jae Phil Koo, Sang Chul Ahn, Hyoung-Gon Kim, Ig-Jae

Kim, "Active IR Stereo vision based tracking system

for immersive displays," ACCV 2004, Jan. 2004.

M. Ghazisaedy, D. Adamczyk, D.J. Sandin, R.V. Kenyon,

T.A. DeFanti, "Ultrasonic calibration of a magnetic

tracker in a virtual reality space," vrais, p. 179, Vir-

tual Reality Annual International Symposium

(VRAIS'95), 1995.

Hibbard, B. 2000. Visualization spaces. SIGGRAPH Com-

put. Graph. 34, 4 (Nov. 2000), 8-10.

GRAPP 2007 - International Conference on Computer Graphics Theory and Applications

136