A REVIEW OF 3-D ACCELERATOR TECHNOLOGY FOR GAMES NATHAN CHIA, RICHARD CANT, DAVID AL-DABASS Department of Computing and Mathematics The Nottingham Trent University Nottingham NG1 4BU. Email: richard.cant/david.al-dabass@ntu.ac.uk KEYWORDS OpenGL, DirectX, anti-aliasing, animation graphics. ABSTRACT In this paper we attempt to review the current technology of 3 -D accelerators for animating graphics used in games and visual simulation systems, together with associated techniques and software architectures. Topics covered include OpenGl, DirectX, anti-aliasing, motion blur and depth of field. A summary of work in progress is give in the conclusions section. INTRODUCTION Ever since 3dfx released their first Voodoo card (the first 3D hardware accelerator for the consumer market), it had always been a 5 horse race between the top video card makers: nVidia, 3dfx, ATI, Matrox and S3. It was a race to see who can hit the highest screen resolution and who can draw the most polygons in a period of time. It was rarely about making the real-time renders closer to real-life. 3dfx came out with a simplified version of the accumulative buffer to gain more speed from the available version. It was named the ‘T-buffer’ and it was introduced as the solution for drawing effects like motion blur, depth-of-field and multi-sampling anti- aliasing in real-time. Now that 3dfx has faded away and nVidia is bringing out with their own version of multi- sampling anti-aliasing named ‘Quincunx AA’, it seems everyone had forgotten about the things that the T- buffer had put out to do. In this paper we will first review the development of 3- D graphics technology and the current state of the art. We will then discuss some effects that remain to be incorporated in 3-D rendering systems. The problems facing 3D accelerators today are mainly spatial aliasing, depthless rendering, temporal aliased animation and images looking 'too perfect'. One might question the motive for this by asking isn’t it better to have sharper, perfect renders rather than distorting them? However, including these effects will give a greater feel of realism and improve game play by modifying the difficulty of task. Hopefully these effects can be included in ways that exploit the features of current hardware such as texture mapping support. This will give software and game makers more freedom to create 3D renders with their very own style in them. The consumers will benefit from this greatly as more real-time graphics is pushed towards reality. THE HISTORY OF COMPUTER GRAPHICS Computer Graphics first started in 1945 when one of the earliest electronic computers, the ENIAC (Electronic Numerical Intergrator And Computer), was built at the University of Pennsylvania’s Moore School of Engineers. By the 1950s, they were powerful enough to deal with computer graphics. To handle the task of drawing lines, special vector display devices were designed to be interfaced to the computer. Vector display can produce wire-frame images (Figure 1), which needed only a minimal amount of storage. As computer memory is extremely expensive in those days, it was only capable of drawing a list of segments. As the 60s grew nearer, General Motors and IBM jointly designed the DAC-1 (Design Augmented by Computer). It allowed the user to enter geometric specification of a wire-frame object and view it from different angles (Figure 2). Figure 1 Vector Display Figure 2 Wire-frame Object In the 1960s, Digital Equipment Corporation (DEC) produced the PDP-1 computer. MIT bought one of the machines and a group of its students created the first video game – Spacewars. The game was so popular