Glossary of Terms

16- 24- and 32-bit Colour
Each pixel is represented by a colour. The number of colours able to be represented depends solely on this measure: 16-bit mode can produce 65,536 different colours and 24-bit mode 16.7 million different colours. 32-bit mode has the same amount of colours as 24-bit but this is only a technicality. Manipulating 32-bit graphics is usually a significantly faster process than 24-bit graphics. However, 32-bit mode takes up 25 percent extra memory.

Since human beings cannot see more than ten million different colours, 24- and 32-bit modes are considered to be photographic quality. In comparison, common 8-bit modes have only 256 different colours.

AGP (Accelerated Graphics Port)
This is a new standard which enables games to hold large texture maps in main memory and very quickly transfer them across the AGP bus to the AGP graphics card. In order to use AGP, a compatible bus motherboard and card are needed.

Alpha Blending/Transparency
Alpha Blending is a feature which allows objects to appear in front of each other, with varying degrees of transparency. Objects can now appear “glass-like”. For example, this effect can be used to simulate water or “see-through” explosions — a very important feature in computer graphics.

Benchmark
A program devised to test the performance of a device.

Bi-linear Texture Filtering
Used to reduce blockiness in textures. Bi-linear filtering calculates a weighted average of four surrounding texture pixels and therefore produces smoother images.

Bitmap
A series of coloured pixels which make up a graphical image.

Blit Rate
The process of copying an array of data from the main memory of a computer, to a video card’s memory, is called Blitting. The speed of this operation is the “blit rate”. Blitting can also be performed within the video card’s memory and this naturally produces higher numbers.

Cache
A utilisation of memory designed to increase performance by holding (locally) recent pieces of information for fast retrieval. The size of the cache can greatly affect the overall speed of many applications.

CPU Extreme mode
In order to ensure that future processors work hard enough, CPU extreme mode is a technique for causing the processor to work four times harder. To prevent outdating, in the case of a FAST processor, all polygon counts are quadrupled.

Depth Fog
This functionality allows close objects to be seen clearly, while distance objects can be made to look faded. This allows objects to disappear into the clouds, for instance. It is generally used to create atmospheric effects.

Diagnostic mode
Final Reality’s™ special mode allows the real-time toggling of 3D image enhancements such as bi-linear texture filtering. Turning these options on or off will change the speed at which the image runs and help diagnose problems in graphic card drivers.

Direct3D™ or D3D
A 3D standard designed by Microsoft. In practice, all 3D accelerators have drivers for D3D. Using Direct3D, however, usually causes performance losses of up to 50 percent compared to native support.

DirectX™
A set of Windows APIs designed to make the task of game and multimedia development easier for coders by focusing on one standard implementation (or way of doing things). It provides a high degree of compatibility.

DMA mode (AGP)
In DMA mode, the graphics accelerator needs to transfer data from the main memory to it’s own memory before processing it.

Double Buffering
While a system is rendering to one buffer, another is displayed. When the rendering is complete, these buffers are swapped. Without this technique, polygons appear on the screen one at a time, and this causes flicker. Double buffering needs twice the amount of display memory.

Execute mode (AGP)
Allows a graphics accelerator to use AGP to access system memory as if it were local video memory.

Fill Rate
A measurement of how many pixels a 3D card can draw in one second.

Flat Shading
Whole polygons are shaded with constant colour.

FPU (Floating Point Unit)
A co-processor which handles all irrational number calculations so, for example, 1.23456 x 2.345678 can be computed accurately. By comparison, an integer unit is of very limited accuracy and so the example above could not be calculated accurately. 3D programs use FPU in their geometrical calculations. FPU cannot be used simultaneously with MMX™ instructions.

Gouraud Shading
A separate colour is calculated for all vertices and interpolated along the polygon. Also referred as “smooth shading”.

Mapping
All objects in 3D are made up of polygons. Mapping is the technique by which a bitmap is applied to the polygon to provide realistic surface details. Usually referred to as “texture mapping”.

MIP-Mapping
This technique improves the image quality of distant 3D objects. It is achieved by using different resolutions of the textures: if the object becomes small enough, a smaller texture is applied. This makes it possible to prevent a chaotic-looking pixel mess in distant objects. Texture details are much better preserved when this technique is used.

MMX™
These are additional instructions added to the processor to help it perform multimedia tasks more efficiently. MMX™ can improve the speed of some applications as much as 200 percent. The biggest advantage can be seen in special image-processing applications. Essentially, programs need to have MMX™ code in them in order to benefit form the MMX™ chip technology. This technology does not currently help 3D hardware-equipped systems to run games significantly faster.

Perspective Correction
“Perspective correction” also applies perspective projection inside polygons. To achieve full realism, both textures and gradients (colours) should be perspective-corrected. The alternative is to perform projection on the object vertices only, which causes textures to bend and wobble.

Phong Shading
Instead of interpolating the actual colour along the polygon, (see Gouraud Shading) phong shading interpolates the polygon normal vector along the surface and calculates new shading colour for each pixel. True implementation of phong shading is rather unlikely to appear in the near future due to its complexity. Phong shading is, however, easily approximated with a texture mapping-based technique.

Pixel
The basic building block of images/textures.

Rendering
The overall process of creating a 3D image with the chosen features e.g. fog, alpha blending, textures, lighting and so on.

SVGA (Super Video Graphics Adapter) resolution
The term applied to a compatible video device operating in 800 x 600 resolution.

Tracker
A piece of code which plays special tracker files. The tracker file originated on the Amiga demo scene and originally consisted of four tracks, each able to play an instrument file (a raw sample file without any header). Typically, each track plays different samples and by changing the note values allows complex music patterns to be built up. The advantages of using a tracker is that the quality of the music is very high. Instruments can be anything you sample and the code is small because only the information from the original sample is required. Nowadays, trackers are more complex and allow compositions of up to 32 tracks. Final Reality™ uses a variant of ScreamTracker (coded by Future Crew) embedded in the core code of the benchmark in order to play the music. Other popular trackers are FastTracker II from Triton and Impulse Tracker.

Tri-Linear Texture Filtering
Calculates a weighted average of two bi-linearly filtered (see Bilinear Texture Filtering) MIP-Map levels (see MIP-Mapping). It further improves image quality, mainly in distant objects.

Triple Buffering
An increase in performance may be gained over double buffering. It is rarely seen, as the process requires an extra buffer and so consumes memory unnecessarily.

VGA (Video Graphic Adapter) resolution
The term applied to a compatible video device operating in 640 x 480 resolution.

Z-Buffering
Instead of storing just the spatial (x,y) position of a pixel, z-buffering technique also stores the depth. A buffer containing these depth values is called Z-buffer. When a new pixel is processed, its depth is compared with the depth that has been stored for current position. If the current pixel is behind the older one, it is not drawn. Essentially, this provides nearly perfect visibility-checking. Further different z-buffer depths indicate how many possible depths are present. As a rule of thumb, 16-bits is the absolute minimum. In general, to achieve perfect results, more depth is needed.