Immersive Virtual Reality on Commodity Hardware

1
Immersive Virtual Reality on Commodity Hardware

• Robert G. Belleman
• Section Computational Science
• Universiteit van Amsterdam
• Email robbel_at_science.uva.nl

2
Overview

• Background
• Project goals
• Design options
• Results
• Conclusions

3
Background SARA

• Architecture, engineering, biotechnology,
  psychology, medicine.

FOR MORE INFO...

• SARA homepagehttp//www.sara.nl/

4
Background UvA/SCS

• Interactive Dynamic Exploration Environments

FOR MORE INFO...

• Section Computational Science homepagehttp//www
  .science.uva.nl/research/scs/

5
Commercial VR solutions

• De facto standard in VR SGI/Irix systems
• High performance graphics
• Expensive, little flexibility, difficult to scale

6
Project Goals

• Multi user, immersive VR
• back-projection screens
• stereoscopic
• interaction hand/head tracking
• High performance, flexible, scalable
• state of the art hardware
• readily available

7
Project Goals (continued)

• Low porting effort
• availability of (or alternatives to) existing
  software and libraries
• well established operating system
• Cost effective

8
Multi user stereoscopic display

• Projection technology
• active stereo
• one projector
• low demands on screen material
• expensive glasses
• interference from environment lighting

9
Multi user stereoscopic display

• Projection technology
• passive stereo
• two projectors or Z screen
• high demands on screen material
• cheap glasses
• viewer must keep head aligned with polarization

10
3D graphics adapters

• API OpenGL
• De facto standard developed by SGI
• Available on most OS
• Hardware acceleration
• Now influenced by gaming industry
• Cheap, readily available
• Very usable for scientific visualization
• Important features missing for VR

11
Stereoscopic 3D using X11

• Quad buffered stereo
• Left/right, front/back
• Not always supported by hardware or drivers
• X11

12
Stereoscopic 3D using X11

• Two hosts, each with a graphics cards, connected
  by a network
• Two hosts, two displays
• Synchronization issues

13
Stereoscopic 3D using X11

• One host, two graphics cards
• One hosts, two displays
• On PCs only one AGP slot!

14
Stereoscopic 3D using X11

• One host, multi-headed card
• One host, two screens
• Requires OS support

15
Stereoscopic 3D using X11

• One host, one graphics card
• One host, two windows
• V-sync doubler (vsync in, 2 x vsync out)
• Halves vertical resolution

16
Tracking devices

• Head/hand tracking
• Motion parallax
• Magnetic, optical, acoustic, inertial
• Ascension, Polhemus, Isense
• Supported by most VR APIs
• Still very expensive

17
Software availability
18
Linux Immersive Environment

• Dual P-III architecture
• Electrohome projector
• nVidia GeForce graphics
• Polhemus/Ascension tracking
• Vsync doubler (top/bottom stereo)

19
Results

• Performance
• Visual quality is good
• Framerate depends on application
• Price/performance ratio is excellent
• Experiences
• Two built, one under construction
• Porting effort
• Reasonable

20
Conclusions

• COTS VR shows great promise
• Solutions for deficits
• Better solutions will come RSN
• Tracking remains biggest problem
• Compared to commercial solutions
• Competitive performance
• Better price/performance ratio