I am currently in the first year of my PhD research, entitled High Fidelity Perception of Dynamic Scenes, in the Department of Computer Science, University of Bristol. My research builds upon the techniques of computer graphics, visualisation and virtual reality which I learnt as an undergraduate in the department, graduating in July 2000 with a 2:1 BSc Honours degree.

My main field of interest throughout my undergraduate training was that of computer graphics: for my undergraduate dissertation I chose to work on an ongoing project in the department, that of producing an aid for the visually impaired [1 & 2]. This project was of great interest to me as it also demonstrated the need for interdisciplinary research, i.e. it linked psychology with computer science, and my PhD continues with this interdisciplinary theme as it will utilise perceptual research.

My research will contribute to the significant interdisciplinary research being conducted in The Digital Media Research Laboratory [3] at the University of Bristol in realistic three-dimensional media content acquisition, processing and display. This research will be based on the use of the Explorer/1 simulator [4 & 5]. This simulator offers a new challenging delivery platform for investigating the high fidelity perception of scenes as the user is undergoing motion.

The major challenge in Virtual Reality is to achieve realism whilst keeping the computational time small and keeping up with real time control movements made by the user. With the Explorer/1 it is possible to obtain realistic motion and display 'blocky graphics' in real time. However, as soon as the rendering of these graphics is improved the rendering lags behind the control movements made by the user, thus destroying the effect of realism.

Some examples of the 'blocky graphics' achieved on the visual displays in the simulator with motion.

Here are some still Radiance images [6] which show some 'realistic' graphics, what we would like the display visuals to look like with motion in the simulator.

The main area therefore for my PhD research is to improve computer graphics in dynamic scenes for the motion simulator without simultaneously increasing the computational time taken to render the present graphics. One obvious way to solve this is by using a far more powerful system, but the economic constraints often preclude this. I think that perceptual research will provide the information to help me accomplish this task.

Currently I am looking into the use of the flaws in the human visual system as possible advantages in producing realistic computer graphics, for the human eye is good but it isn't perfect! It is interesting to consider how the eye's inability to perceive certain objects in images might be used to reduce the level of rendering detail, without the change being perceptible to the viewer. The final goal, therefore, would be to use these flaws to save computational time rendering the 3D world we have created.

What I would like to learn from this campfire

Coming from a Computer Science background I understand how to create the graphics and program the simulator but my knowledge is lacking when it comes to the psychological and visualisation experiment techniques I will need to know and understand to perform the experiments. As I have just started my PhD most of my work has been researching into what people have already accomplished in this field rather than producing any results of my own. Thus I would be interested if anyone has done similar work in this field, how they accomplished it, what difficulties they came across, and any suggestions of how I should set up my own experiments.

Thus the main thing I wish to take advantage of at the campfire is the knowledge of other attendees.

The issues which I would like to discuss

• How important is it that the computer graphics in simulators are extremely realistic?
• Is it more important that in a simulator you have realistic motion and graphics that tie up with that motion than just realistic graphics?
• If we flood the senses, i.e. use motion, visuals and sound, can we get away with a lower level of rendering detail in the visuals, than if we just use the visuals?
• What sense of motion can you get just from the visuals? e.g. In a real life example, if a lorry starts rolling backwards when you are stationary in another vehicle next to it, you feel like you are moving forwards. Is it possible to simulate this effect?
• How can information about change blindness & blindsight be used to make sure that the user perceives important objects/icons in the case of user interfaces?

References

[2] http://www.caterk.freeserve.co.uk/WebPage.html - Image Processing using Neural Networks - Producing an aid for the Visually impaired.

[3] http://www.cs.bris.ac.uk/Research/Digitalmedia/ - The Digital Media Research Group, University of Bristol.

[4] http://www.motionbase.com - The motionbase website.

[5] http://www.cs.bris.ac.uk/~cater/MotionSim.html - Some pictures showing the motion platform.

[6] http://radsite.lbl.gov/radiance/gallery/gallery.html - A gallery of images and animations produced using Radiance.

© Copyright is held by the author, Kirsten Cater, 2001