This dictionary section contains the terms that should be known by
a person working with computer graphics.
The terms were chosen to be specific to computer graphics, in the sense of
image generation, rather than as used by humancomputer interfaces or
computation in general.
We excluded terms that are either generic (e.g. color) or
too uncommonly used.
These excellent resources can be consulted for further details about
the terms defined here:
 2D
 Twodimensional
 3D
 Threedimensional
 Abuffering

An antialiased extension to
Zbuffering. An Abuffer identifies visible segments within a subpixel area
which are represented with bit masks and area sampled for pixel intensity.
The technique employs logical operations on the bit masks and thus avoids
floating point geometry calculations.
 Achromatic

Light without color.
The quantity of light is the only attribute associated with achromatic
light. In physical terms this is the intensity or
luminance or in the psychological sense it is the perceived intensity in
which case the term brightness is used.
In the YIQ or YUV representations, this is the Y component.
In the HSV representations, it is the V (value) component.
In the HSL representations, it is the L (lightness or intensity) component.
 Adaptive forward differencing

An efficient way to evaluate parametric functions describing
curves or surfaces. Each value of the function is determined as the
sum of the previous value and a difference term. The distance between
points at which the function is evaluated is adapted to the flatness
of the function.
The value can be a vector as well as a scalar, and this is useful for
calculating Bsplines.
 Adaptive sampling

Adaptive sampling is a method of reducing aliasing artifacts when
rendering by adapting the sampling rate in response to the local
characteristic of the object being rendered. This technique is often
useful to reduce the jagged edges at the edges of objects (or jaggies).
 Adaptive subdivision

A paradigm for representing data in a hierarchical
manner by repeatedly dividing and classifying it until no further
definition is necessary, given an error tolerance. E.g.
see octree and quadtree.
 Additive color model

In an additive color model, colors are defined as a
sum of contributions from primary colors. The most commonly used additive
color model is the RedGreenBlue model.
 Affine map/transform

A geometrical transformation from an affine space
to another. In computer graphics it is used for 2D spaces (image) or
3D spaces (3D image). It is defined by :
p' = M p,
where M is a real matrix and p is a real vector.
Affine transformations preserve parallelism.
 Algebraic surface

A surface defined by the set of points for which an algebraic function is
equal to a constant value. For an algebraic function
at point p and a
constant value c the surface S can be formally defined as:
 Alpha blending/alphachannel compositing

A technique for computing the color of a pixel when multiple
structures contribute to the pixel ( e.g. at a region
boundary where we want to avoid aliasing problems arising from partial pixel coverage or
transparency).
Alpha is the percent of a pixel covered by a given structure,
and can be used as part of the color description of every pixel
associated with a structure. Computing the resulting color of a
combined pixel uses the alpha values of the source pixels, plus
other information, such as any relation between the surfaces.
 Alpha channel

The collection of alpha values associated with an image where each
alpha value represents the coverage of each pixel in the image. The alpha
values are used in the process of alpha blending.
 Ambient lighting

A global (artificial)
illumination level representing infinite diffuse reflections from all
surfaces within a scene ensuring that all surfaces are visible (lit)
particularly those without direct illumination.
Ambient lighting is usually treated as a constant in local shading
functions but is simulated directly in radiosity calculations.
 Anaglyph

A stereoscopic picture consisting of two images of the same object,
taken from slightly different angles, in two complementary colors. When
viewed through colored spectacles, the images merge to produce a
stereoscopic sensation.
 Animation

(1) A medium that provides the illusion of a moving scene using a
sequence of still images.
(2) Techniques used in the production of animated films. In computer
graphics this primarily concerns controlling the motion of computer models
and the camera.
 Anisotropic filtering

Image filtering that produces different amounts filtering
(e.g. smoothing filtering in different directions at each pixel
in an image.
Two uses of anisotropic filtering in graphics are to: 1) produce textures
with different spatial frequency distributions in different directions
and 2) to reduce aliasing effects along edges without blurring the
edges as much.
Anisotropic filtering can be done in either the image or the frequency
domains.
 Antialiasing

Antialiasing is a method of reducing or preventing aliasing
artifacts when rendering by using color information to simulate higher
screen resolutions. In one technique, blurred pixels are introduced by filtering the image,
or individual elements, to remove spatial frequencies that are greater than
the pixel sample rate by convolution. If high frequencies remain they may
cause other visual artifacts such as Moiré patterns.
An alternative and often preferable technique is supersampling,
where many samples per pixel are estimated and combined.
 Artifacts/Artefact

A classifiable visual error. E.g., a loss of resolution
when zooming into an image or
incorrect depth sorting due to the painter's algorithm.
 Atmosphere effects

Atmospheric effects arise because light is affected by the properties
of the medium through which it passes. The main effects are
attenuation, where distant objects get lower contrast (see
depth cueing) and blurring, such as might occur with
dust, fog or haze, which scatters the light.
 Attenuation

1. Atmospheric attenuation: the simulation of the
atmospheric attenuation from the object to the viewer which affects
both the illumination strength and color. The attenuated
illumination is computed by
where s is a scale factor ranging from 0 to 1,
is the
illumination and is the
depthcue color.
2. Light source attenuation: a factor in the illumination equation
used to simulate surface illumination depending on how
far the surface is from the light source. It is defined by:
where is the distance
between the source light and the surface,
and
and are user
defined constants associated with the light source.
 Augmented reality

The idea that an observer's experience of an environment can be augmented
with computer generated information. Usually this refers to a system in
which computer graphics are overlaid onto a live video picture or projected
onto a transparent screen as in a headup display.
 Bspline

A multisegment spline curve representation based on local
polynomials having
continuity of curve orientation and curvature at the points (knots)
where different segments join.
Cubic bsplines are popular, but linear, quadratic, quartic, etc. splines are also used.
The B in Bspline stands for basis, because the bspline segments
are formed from the weighted sum of four local basis functions.
The local shape of the spline segment is controlled by four control
points; in the case of bsplines, these control points do not lie
on the curve itself ( i.e. bsplines are not interpolating).
One important advantage of bsplines is that the movement of a
control point affects only four segments of the curve.
Bspline surfaces can be defined from bspline curves lying in
both directions on the surface. Here, 16 noninterpolated
control points are needed per patch, but each patch then has
tangent and curvature continuity where it joins a neighboring
patch.
The bspline is defined over a uniform parameter domain, and
is evaluated as a simple polynomial function.
More complex forms, such as NURBS, relax these assumptions.
 Backfacing polygons

Polygons whose surface normals point away from the camera
position, which can be easily tested by the dot product of the polygon
surface normal n and the ray v from the viewer to the
polygon. A polygon is backfacing if .
For closed objects there is no need to draw backfacing
polygons as they are always occluded by nonbackfacing polygons.
 Background color

The intensity level of pixels
which are not intersected by any of the displayed surfaces.
 Backwards ray tracing

Backwards ray tracing is used to render a scene on a view plane by tracing
imaginary "eye rays" from the viewer's eye to the surface of the objects
in a scene, to determine the objects' visibility. A grid on the view plane is
used to cast eye rays from the center of projection (the viewer's eye). It
is convenient for the grid to correspond to the pixels of the display
screen. For every pixel on the view plane, an eye ray is cast from the
center of projection, through the center of the pixel and into the scene.
The pixel's color is determined by the eye ray's point of first intersection
with an object in the scene.
The basic backwards ray tracing algorithm can be extended to render shadows
in a scene. This extension involves firing an additional ray from the first
point of intersection to each of the scene's light sources. If the ray
intersects with an object on its path to the light source, then the point
of first intersection is in shadow for that light source. The combination of
the effect of each ray to the light source determines the first intersection
point's color.
 Basis spline

A spline curve or surface that can be formulated as a weighted sum of
polynomial basis functions. Commonly known as a BSpline.
 Beam tracing

Beam tracing is a method of rendering similar to raytracing
but using an arbitrarily shaped projection, commonly a polygonal cone,
rather than a single ray. It is an improvement on raytracing since it
reduces the CPU overhead and reduces aliasing artifacts by taking
advantage of known spatial coherence in the beam.
 Bézier curve

A spline curve that (in the usual case of a cubic
Bézier curve) is represented by four control points defining a cubic
polynomial.
 Bicubic surface

A type of parametric two variable polynomial surface patch
where the polynomials are cubic in both parameters.
 Bilinear filtering

An averaging technique applied to the color values of
adjacent pixels
so that textures look smooth rather than blocky. It aims to make
the texture looks more realistic
 Bilinear interpolation

An algorithm for interpolating image data in order to estimate the intensity
or color of the image in between pixel centers. The interpolated
value is calculated as a weighted sum of the neighboring pixel values.
 Binary space partition tree

A method for representing a polyhedron that explicitly uses the planes that
bound the polyhedron. The technique represents the object
as a binary tree, with planes at each nonleaf node. The planes
bound a face of the polyhedron, and divide space into two subregions,
which in turn can be further bounded by the two children at a node.
Leaf nodes (at the edge of the tree) are either
completely object or completely free space.
A similar idea can be used in 2D for representing polygons.
This representation is useful for hidden surface removal and
point classification (determining whether a point is inside or
outside the object).
 BitBlt/RasterOp

An abbreviation of bit block transfer. This is an efficient technique
for copying rectangular arrays of pixels that exploits the fact that
computer memory is organized into multibit words.
 Bitmap

Strictly a onebitperpixel
representation for a defined area of a display.
 Bits per pixel

The number of bits used to describe the color or intensity of a pixel.
For example, using 8 bits for to store a value from the RGB color
model would permit 3 bits to be used for both red and green values and 2
bits for the blue value. Blue gets a smaller range because the human eye
contains less blue cones and is thus is less sensitive to blue variations.
 Blend surface

A surface added to two or more others surfaces to provide a
continuous join between them.
 Blitter

A blitter is a specialpurpose chip or hardware system used for fast
implementations of bitmapped graphics.
Blitters are used to copy sections of video memory from one place to
another.
During the copy operation several source areas may be used and logical
operations may be performed on them.
One application of blitters is the provision of fast animated graphics,
known as sprites.
 Boundary representation/Brep

A paradigm for representing graphical
data in terms of the boundaries of the objects involved. E.g.,
representing a cube as a collection of bounding faces, or a
polygon by its edges.
 Bounding box/volume

The smallest regular shaped box that encloses an object, usually
rectangular in shape. Bounding boxes are used to accelerate tests such as
visibility or rayobject intersection by providing a pretest which can
eliminate many cases.
 Bresenham's algorithm

A technique developed in the framework of raster graphics for
generating lines and circles. These algorithms use only
integer arithmetic, avoid rounding and perform an iterative
computation of the primitive points by approximating
the distance to the nearest pixel center along
either the x or y axis.
These characteristics make for efficient algorithms.
 Brightness

The perceived intensity of a radiating object.
 Bump mapping

A technique used to increase the realism of a surface by changing
how light reflects from that surface. Usually, the surface normal
at a given point on a surface is used in the calculation of the
brightness of the surface at that point. In bump mapping, the true
surface normal n is perturbed a small amount
n
as a function of position
on the surface. The perturbation can be regular, so as to give a
regular textured shape to the surface, or it can be random, so as
to increase the natural appearance of the surface. Part of what
gives this techniques its appeal is that the original surface
maintains its original (usually smooth) shape, and the bumpmapping
distortion is specified by a compact function of shape. This is
usually much simpler and more compact than specifying the surface
texture by explicitly representing the textured surface.
 CAD

Abbreviation of Computer Aided Design. In the context of graphics, CAD refers
to the use of computer based models of objects for visualization or testing
as an aid in the design process.
 Camera

A virtual viewpoint in world space with
position and view direction to provide a view of a scene in the same way as a
photographer would position a camera.
 Candela

Derived from candle and denoted by the symbol "cd", it is the basic
SI unit of luminous intensity. It is defined as the radiation intensity, in a
perpendicular direction, of a surface of 1/600000 square meter of a
black body at the temperature of freezing platinum under a pressure of
101325 newtons per square meter.
 Canvas

A twodimensional region of graphics information. The canvas may
be displayed on screen or be recorded in offscreen display memory.
 Cartesian coordinates

A common system of representing a point in two or more
dimensions using an ordered set corresponding to its projection on a
spanning orthogonal base set. Commonly encountered Cartesian
coordinate systems are the XY plane 2D coordinate system, (row,column)
2D image coordinate system and XYZ 3D scene coordinate system.
3D coordinates in graphics are usually specified with x and y being aligned
with x and y on the screen, +x is to the right, +y is upwards, and +z goes
into the space 'behind' the screen. This is a lefthanded coordinate system
with the property that most zvalues are thereby positive. It's why
zbuffers are called zbuffers when they are
actually depthvalue buffers.
 CAVE

An immersive virtual environment
where the viewer stands inside a room upon whose walls
are projected images. The images may be in stereo requiring
stereo shutter glasses to be worn.
The name CAVE comes from Computer Augmented Virtual Environment.
 Caustic

The effect given when light is transmitted through a specular surface and
then strikes a diffuse surface. If the specular surface is of high
curvature the light will be tend to be focused. When this effect is taken
into account, rendered scenes involving liquids or glass are much more
photorealistic.
Caustics can also arise when light is reflected from a specular surface.
The classic example is the caustic on the surface of a liquid.
Refraction may make it happen too.
The caustic shape is the envelope of the reflected rays.
 Center of projection/viewpoint

Part of the model representing the the projection from a 3D space (the world)
to 2D planar space (the image). It is the point of intersection of all the
straight projection rays emanating from the object points in the 3D space
and intersecting the projection plane to form the projection.
 Chroma

1) A characterization of how much a color differs from both the
pure color and the grey of the same intensity. Also called saturation.
2) The color component of a composite video signal.
 Chromaticity coordinates/tristimulus coordinates

Chromaticity coordinates are based on the
Commission Internationale de l'Éclairage (CIE) color scheme,
which uses three standard (but physically unrealizable) primary colors called X, Y and Z.
(These are different from red, green and blue, and are chosen to
represent human color matching performance.)
Any visible color c can be expressed as a weighted sum of these
primary colors: .
The weights are called the tristimulus values
and are a way of objectively encoding all visible colors.
(Actually, each set of weights represent an infinite set of colors which are
indistinguishable.)
Normalizing the colors by:
generates the chromaticity coordinates , which are independent
of the brightness of the color. Note that
z = 1  x  y, so we can recover z, but we have lost the absolute
brightness of the color.
 Chrominance

Information describing hue, or the color components
orthogonal to the brightness. YUV and YIQ are
chrominance/luminance color models.
 Clipping

The selective removal of an object
disjoint with the display area or the nonvisible parts of an object that
does intersect the display area. Parts of an object intersecting the
display area may lie
outside of the display area or be partially or fully obscured by another
intersecting object.
 Collision detection

Collision detection is used in a virtual environment
to monitor the relative
locations of solid objects. If the virtual environment manager detects that the
proximity of two or more objects is sufficiently close, a collision event
occurs. As a result of this event the object's movement can be controlled so
their surfaces do not intersect. In an environment which models a natural
system, the kinetic energy of a moving object is (partially) transferred to
the object it collides with, making the second object move.
 Color keying/chroma keying

Using the pixel color of one image to designate that pixel
data from another image should replace the first pixel's color.
The first image might be a binary image, which would select
regions of interest from the second image.
Another use is in bluescreening, where an actor works against
a blue background. In the output image, the blue pixels get
replaced by another image. For example, a weather map can be placed
behind the weather presenter who is actually standing in front of a
blue screen.
 Color models

A color model is a method of specifying a color (position) in color space,
often using a coordinate system. Examples include RGB and the
Munsell Color System.
 Color space

A mathematical space defining a range and encoding of colors.
E.g. see RGB, LUV, HSV,
HSL, YIQ,
YUV and XYZ.
 Compositing

The process of combining multiple images into a single image. Usually
this is performed in films to make a computer graphics generated
character appear on a previously filmed background.
The term is also used in traditional photographic manipulation
to refer to the process by which cel
animation is recorded onto film under a rostrum camera.
In film the 'mechanical' process is usually called matte photography
(see color keying), and the process, when
used in film sequences is ambiguously called traveling matte.
 Concave/convex polygon

A concave polygon has the property that some points
within its area can be joined by a line segment that passes outside the
polygon. A convex polygon has the property that any line segment
joining two points belonging to the polygon area is completely
inside the polygon.
 Cone tracing

An alternative to ray tracing in which cones are projected from
the camera center through each pixel,
where the intersection
of the cone and the scene model is used to determine the pixel's color.
 Contour

This is an image curve, often used to represent the set of
points where a given function has a given constant value. A
familiar example is a contour line on a topographic map. Here the
contour denotes where the land has a given elevation. Another type
of map contour might denote the boundary between increasing and
decreasing population density.
The equivalent concept in 3D is the level surface
or isosurface.
 Contrast

The range of colors in an image. Increasing the contrast of a
color palette makes different colors easier to distinguish, while
reducing the contrast makes them appear washed out.
 Control point

One of a set of points which control the shape of a curve by their
intuitively by position.
The curves may go through some (see Bézier curve, an interpolating spline)
or all (e.g. the CatmullRom interpolating splines) of the control points.
Positioning is often interactive and the points
are combined by blending functions to generate the shape desired. See also
Bspline and Bézier curve.
Note the distinction between knots and control points:
in an interpolating spline, knots and control points are at the same
positions in space. In a quadratic or higher order approximating spline they
are in different places: the knots lie on the curve and control points lie near the
knots, but not on the curve.
 Convex hull

The convex hull of a given set of points is the smallest convex set that
contains all the points.
 Coons patch

A Coons patch is a form of parametric bicubic spline
representation for surface patches. It allows explicit
control of patch boundary position and tangent plane continuity.
It is an example of a lofted surface.
 Coordinate system

A coordinate system is a minimal set of mutually orthogonal
vectors which span a given space. All points or vertices in the space may then be
represented using a linear combination of these
spanning vectors.
 CSG/Constructive solid geometry

A paradigm for representing 3D shapes
in terms of mathematically based compositions of geometric
primitives. Any volumetric primitives can be used provided the
primitive can satisfy an 'insideoutside' test which uniquely
partitions points in the space near it.
Typically, boolean set theoretic composition operators (e.g. intersection, union, difference)
are used.
Affine transformations may be applied to alter the shape of the primitives.
For example, the exterior of an igloo may be represented as the
union of a sphere and a cylinder, intersected with a cube.
 Cuberille

A representation of 3D space consisting of a regular array of cubes,
often referred to as voxels.
 Data visualization

The set of techniques used to turn a set of data into visual insight.
It aims to give the data a meaningful representation by exploiting
the powerful discerning capabilities of the human eye.
The data is displayed as 2D or 3D images using techniques such
as colorization, 3D imaging, animation and spatial annotation
to create an instant understanding from multivariable data.
 Delta frame

The difference between two consecutive images. Often used in video
compression algorithms that exploit the temporal coherence of image sequences.
 Depth buffer/Zbuffer

A method for solving the visible (or hidden) surface problem using two aligned
pixel buffers or images. The first buffer stores the current color
of the pixel and the second buffer stores the distance from the viewer
to the surface. When rendering a point a on a scene surface,
if the
distance from the observer to a is greater than that of a
previous point b that projects to the same image pixel, then
point a can be ignored (as it cannot be seen). If the
distance to a is less than the stored distance to b,
then distance and color of a replace
the color and distance buffer entry of b.
A zbuffer is often efficiently
implemented as a hardware buffer with entries aligned with pixels.
Unfortunately, these zbuffers suffer a lot from aliasing effects
and Abuffers are much better at dealing with
visibility problems at subpixel accuracy.
 Depth complexity

A measure of the complexity of an algorithm. It is equivalent to the
number of pieces of data written to a framebuffer divided by the total
number of pixels in the framebuffer, when a whole frame
is rendered.
 Depth cueing

Objects closer to the viewer
appear brighter and more distinct than distant objects. Thus more distant
objects or distant parts of objects are displayed with less intensity to
simulate this phenomenon and enhance perception of depth.
 Diffuse reflection

The portion of light that falls on a facet (small piece of the surface)
which is radiated diffusely in all directions.
 Direction cube

A technique used for representing spatial directions,
often used by recursive direction decomposition algorithms.
The cube is placed at the origin and aligned so that
the coordinate axes are orthogonal to the faces.
Each face of a cube is subdivided
into a number of squares. Each square represents a collection
of similar directions. Subdividing the squares on a face
increases the resolution of the directions.
 Directional lighting

A light source that radiates in such a way that rays from it are nonparallel.
 Dissolve

An animation effect that is a transition between two
sequences involving a fade from one directly to the other.
 Dithering

One of many processes for reducing the total number of colors present in an image
while retaining visual fidelity.
Dithering can be done by interleaving pixels
of selected colors to locally approximate the desired color.
Dithering can be applied to either a
color or a greyscale color space and may be necessary due to a
limited number of colors available on the display device.
 Doublebuffering

A mechanism for duplicating the framebuffer memory by using a two
buffer system in which the image in one buffer is displayed while the image in
the other buffer is computed.
The newly created image is then displayed by swapping buffer pointers rather than
having to copy memory.
Double buffering allows the CPU to have uninterrupted
access to one of the buffers while the video controller has access
to the other.
 Edge merging

The process of replacing the edge of a polygon with the adjacent edges
of neighboring polygons to prevent cracks appearing
during rendering.
 Emittance

The light emitted by a surface. This may have different intensities
and spectral characteristics in different directions.
 Explicit surface

A surface representation in which the z coordinate is expressed
as a function of the x
and y coordinates.
 Extended light source

A light source with surface
area which will cast shadows with both umbra and penumbra and thus is more
difficult to model than a point source.
 Face/facet/patch normal

A solid object can be constructed from many surface pieces which fit together. Each
piece is called a face/facet/patch. Its normal is the direction from the
surface of the object that is perpendicular to the piece's surface.
 Facet/faceting

A facet is a small piece (usually a planar polygon) of a larger surface.
Faceting is the technique used to construct a surface from multiple facets;
triangulation is an example of faceting.
 Fading

Fading is a method of switching between video sources, or images, using a
black image as an intermediate.
Fading without this intermediate is called a dissolve.
 False coloring

See pseudocolor.
 Field rendering

In interlaced video, a single image frame is sent as two fields  composed of even scanlines and
odd scanlines. Field rendering refers to a method of rendering where fields
are rendered separately in order to reduce motion artifacts.
 Fill/flood fill

These are techniques for coloring
areas bounded by line edges. The algorithms that fill interiordefined regions
(the largest connected region of pixels whose values are the same as
a given starting pixel) are called flood fill algorithms.
 Filter

1) An optical device that selectively attenuates the intensity of light
passing through it according to the light's properties. Common filters
attenuate light according to either wavelength or polarization state.
2) An algorithm that selectively modifies the intensity or color of
image data according to the image's properties.
3) An element (software or hardware) which takes in a stream of data
and produces a stream of results, on average one output for each input.
 Flat shading

Shading a polygonal patch with a single color and intensity.
The shade chosen is a function of a variety of factors, such
as light source position,
viewer position
and surface normal,
according to the shading model used.
A single shade is how the patch would appear if the surface is
genuinely planar, rather than just being approximated by polygons,
and if several viewing environment conditions
hold (distant viewer and light source).
 Fogging

The blending of a color, often light
grey, with parts of an image such that the farther objects become increasingly
obscured. (See Atmosphere effects.)
In other words, the contrast between the fog color and objects in
the image gets lower the deeper an object appears in the scene. Fogging may be
used to provide a backclipping plane where objects too distant to be seen
clearly are removed to speed up the rendering of a scene.
 FPS/Framespersecond/Feetpersecond

1) A measure of the speed
of an animation in terms of the number of complete, fully rendered images
or frames which can be displayed in one second.
2) The same, except FPS refers to the number of feet (30.48 cm) of cinema film displayed in
one second.
 Fractal

A fractal has statistical selfsimilarity at all resolutions and is
generated by an infinitely recursive process. In reality, those fractals
generated by finite processes may exhibit no visible change in detail
after some stage so are adequate approximations. So, for computer graphics
we can extend the definition to include anything that has a substantial
measure of exact or statistical selfsimilarity. This is illustrated by
three stages of the construction of the von Koch snowflake below where each straight edge is
repeatedly replaced by a copy of the entire figure.
Fractals are useful for generating natural appearing shapes or textures, such as
land and cloudscapes.
 Frame

A still twodimensional image. Often a frame is a raster image as used in the
frame buffer of a graphics display system. In computer animation
frames per second is a measurement of the number of still frames
displayed in one second to give the impression of a moving image.
 Frame rate

The frame rate of a video source is determined by the speed at which it
completes the rendering of a new image.
This is limited by both the speed at which image data can be created
and the rate at which video images can be presented on a display.
For example the NTSC system redraws at 30Hz, PAL is 25Hz and
computer displays are now usually 7275Hz.
 Frame size

A term used to refer to the dimensions of the array of
pixels forming a frame of an animation, or
alternatively the memory requirement and hence indirectly the
resolution and dimensions.
 Freeform/freeform surface

A surface that does not have a simple geometric description
(e.g. not a plane or quadric surface). It is usually
represented using a spline surface or a
triangulated surface.
 Frequency

The number of time that a periodic function or vibration repeats
itself in a specified time or space. It is often measured in
cycles per second, or cycles per centimeter, or cycles per degree
of visual arc.
 Fresnel equation

An equation used to determine the attenuation of unpolarized light
reflected from a surface, given the refractive index of the surface
material and the angle of incidence of the light relative to the
surface normal.
 Frustrum of vision

The visible region of 3D space. Projecting rays from the viewer
through all pixels in the image plane defines an infinite pyramidlike
solid shape within which all visible objects appear. The pyramid
is then truncated by a distant plane to eliminate the space which
is too far away to render, and by a closer plane which eliminates
object too close to render. The space in between is the frustrum
of vision.
 Gamut

Normally refers to the full range of
colors available in a color space. The gamut varies with resource:
photographic film, printing inks, color displays, etc. A 24 bit color
system has a gamut of 16
million different colors. Moving between systems
with different color gamuts will require quantization.
 Gaze direction

View direction is specified
as a target object rather than the more usual vector form from a camera or
eye position and direction.
 Generalized polygon

A generalized polygon is a planar shape that is
constructed using an ordered number of vertices that are connected to
form an enclosed polygonal area. It is a graphics engine's most abstract
internal representation of a shape. Specific shapes such as square and
triangle have a fixed number of vertices (3 and 4 in this example) and
can be generalized using a generalized polygon.
A generalized polygon may have holes or be concave.
Other shapes such as circle and ellipse have an infinite number of
vertices. A generalized polygon can provide the graphics engine with an
approximate representation by using a large number of vertices.
 Gloss

An object is said to have a gloss surface when specular reflection is
observed. This causes a highlight on the surface when a bright light is
directed at the object.
 Gouraud shading

Gouraud shading computes an intensity for each vertex of a
polygon using Lambertlaw shading
and then interpolates the computed intensities across
the polygon
by performing a bilinear interpolation of the
intensities down and then across scan lines.
It thus eliminates the sharp changes at polygon boundaries.
 Graphics application programmers interface (API)

A software library
enabling a programmer to produce a graphical application, typically
incorporating input handling (mouse, keyboard etc.). E.g.
OpenGL, Java3D, Allegro.
 Gray scale

A color space where colors are represented by their luminance
values only, i.e. saturation and hue are zero.
 GuptaSproull algorithm

A technique developed in the framework of Raster Graphics.
It aims to reduce aliasing when doing line drawing.
 Hedgehog

A visual representation of a surface in which the surface normals are rendered
like pins sticking out of the surface.
 Hidden line removal

A technique used in wireframe rendering
(which is when one draws the
straight line boundaries of the polygonal patches,
or polyhedral
solids that define the scene). If all boundaries are drawn, this
is as if all surfaces and objects are transparent. If all surfaces
and objects are opaque, then some boundaries would not be visible
because they are hidden by closer surfaces. Removing the obscured
or occluded portions of the boundaries is hidden line removal.
 Hidden surface problem

Sometimes called visiblesurface determination or
hiddensurface
removal. It is the problem of only displaying the parts of a surface in a
scene which are visible to the user. For a scene to make sense to a user,
any surface that is obscured by an opaque surface must not be rendered. For
raster graphics, an example of a rendering algorithm which solves the
hiddensurface problem is the zbuffer algorithm.
 Highlight

The area of a glossy object over which
specular reflection can be viewed. It is normally the color of the light
source, not of the object.
 Hollow fill

A threedimensional object whose internal volume (defined as the space
enclosed by the object's skin) is not rendered. Such threedimensional
solid objects are frequently used in virtual environments and are
constructed using infinitely thin polygons to form the skin of the
object. If the user's viewpoint is positioned within the skin of the
object, the reverse of the surface will be rendered if there is
sufficient illumination.
 Homogeneous coordinates

Normally, the transformations for scaling, rotation and
translation are treated
differently. Scaling and rotation use matrix multiplication whereas translation
uses vector addition. When the homogeneous coordinate system is used, all three
transformations can be performed using matrix multiplication.
This representation is commonly used in graphics systems because of its
simplicity of representation and use.
A homogeneous coordinate is expressed with an additional coordinate to the
point. So, a twodimensional point is represented as a
homogeneous coordinate by a triple .
Two sets of homogeneous coordinates are equivalent if one is a multiple of the other.
If the W coordinate is nonzero, we can divide each coordinate by W,
transforming
into ;
the numbers
and
are the Cartesian coordinates of the homogeneous point. If W is zero, the
homogeneous coordinate is a point at infinity.
 HSL/HueSaturationLightness

HSL, also known as HSI (HueSaturationIntensity) is a color space used to
represent images.
HSL is based on polar coordinates, while the RGB color space is
based on a threedimensional Cartesian coordinate system.
Intensity is the vertical axis of the polar system, hue is the relative
angle and saturation is the planar distance from the axis.
HSL is thought to be more intuitive to manipulate than RGB space.
For example, in the HSI space, to change red to pink requires only
changing the saturation parameter.
 HSV/HueSaturationValue

A color space that describes color
using three basis components: hue, saturation and
brightness.
(See also HSL and Munsell color system.)
 Hue

A perceptual term referring to the colorimetry quantity 'dominant
wavelength' of a color. Hue can be used together with saturation and
luminance to define the HSL color space.
 Illuminance

The amount of light falling into a patch of unit surface area.
It is measured in lux.
 Illuminant

A source of illumination.
 Imagebased rendering

An approach to rendering in which objects and environments are modeled using
image data instead of geometric primitives.
 Image file format

A representation (usually binary) used by a computer system as an agreed
format to store an image. Examples of image file formats include the
Graphics Interchange Format (GIF) and Tagged Image File Format (TIFF).
 Immersive VR/Virtual reality

A system where a
user's field of view is completely filled by the display medium and the
user can interact with the visualization in a natural way such as pointing,
grabbing, head movement to change view, etc. The user is also shielded from
external factors such that the overall perception is one of being immersed
within the visualization.
 Implicit surface

An implicit surface is defined using an implicit equation given by
for some function . The
equation restricts the interaction
of x, y and z to ensure that the point
is confined to the
surface.
 Inbetweening

Inbetweening is the generation of intermediate transition positions from a
given start and end point or keyframes. This technique is often used in
animation, where a lead artist generates the beginning and end keyframes
of a sequence (typically 1 second apart), a breakdown artist does
the breakdowns (typically 4 frames apart), and an `inbetweener' completes the rest.
 Indexed 16 and 256 color images

An indexed color image consists of a set of references to values stored
in a color table or palette.
The palette, which is often contiguous in an image file, lists all the
colors as sets of coordinates in color space.
An indexed 16color image contains a palette with 16 color entries (4
bits), whereas in an indexed 256 color image 256 colors are listed (8 bits).
 Interlaced display

A technique for displaying images at a higher
resolution than the monitor. Two images consisting of every
second row of pixels are alternately displayed during every
screen refresh (e.g. every fiftieth of a second). There is hence
a flickering artifact.
 Interpenetration

The surface of one object passing through the surface of another.
 Interreflection/Mutual Illumination

A phenomenon that occurs when a surface reflects light from other surfaces
in its environment. The effects range from more or less
sharp specular reflections that change with the
viewer's position
to diffuse reflections that are insensitive to viewer's position.
 Irradiance

A measure of the amount of light energy incident on a unit area of surface
per unit time. Measured in Watts per square meter.
 Isometric projection

This is a form of orthographic projection in which the
direction of projection and
surface normal of the image plane are parallel to one of these
eight directions { (1,1,1), (1,1,1), (1,1,1), (1,1,1),
(1,1,1), (1,1,1), (1,1,1), (1,1,1) }.
 Isosurface

A technique used in threedimensional data visualization where a surface is
drawn around points in threedimensional space that represent the same data
value. For example, the set of points { (x,y,z) : f(x,y,z) = c }
where c is a given constant.
(See implicit surface.)
 Iterated function system

A finite collection of
affine mappings in the plane which are combinations of translations, scalings
and rotations. Each mapping has a defined probability and should be
contractive, that is, scalings are less than 1.
Iterated function systems can be used for the generation of fractal objects and
image compression.
 Jittering

Jittering is performed by displacing sample locations that are initially
spaced regularly. Typically, this involves randomly shifting uniformly
positioned sample points horizontally and vertically.
Such a sample point is usually in the center of the pixel which is perturbed to
some other location within it.
Jittering adds noise to the rendered image; the advantage of jittering is
that the human eye tolerates noise more easily than it tolerates aliasing
artifacts. Consequently, humans perceive the jittered image as being of a
higher quality.
 Keyframe

An image that is stored in some way to be used as a reference point. Key
frames are often used in animation.
 Knot

A knot is usually the join point between spline curve segments.
(If the spline is a bundle of e.g. cubics model, then a knot is
the place where one cubic stops and another starts.)
 Lambert's law

A shading model in which the diffuse
component of the brightness of a point on a surface is estimated as a
scaled cosine of the angle between the surface normal and the
direction from the point to a light source.
 Levelofdetail blending

When rendering models which are defined with levelsofdetail, artifacts
can occur when one levelofdetail is replaced with another. This is known
as 'popping' and can be reduced by blending one level with the next when the
transition takes place.
 Leveling

A process applied to an image in order to have global uniform illumination.
There are many techniques for leveling. The simplest consists in subtracting
from the original image the image of the background taken under the same
conditions and then expanding the contrast of the difference.
 Light source

A source of visible electromagnetic radiation. See also
local light source.
 Line clipping

Selecting the portion of a line segment that lies inside
of a clipping window. If the line intersects the
window boundary, then the line is split into two or three
segments, one of which moves with the clipping window and
the other(s) remain unmoved.
 Line drawing

An image created only from points connected by lines. It can be described
using a series of endpoint coordinate information,
and , for each connecting line.
This can be combined with a weight
which denotes the thickness of the connecting line.
 Linear depth cuing

Linear depth cueing is a rendering technique used to give the effect of
depth to an image. This is achieved by intensities in the image plane
being reduced linearly with respect to the distance from the plane.
(See Attenuation.)
 Local light source

Light source that directly (i.e., not by reflection or transmission)
illuminates a point on a surface.
 Luminance

The absolute quantity of radiation emitted from a given light source.
 Luminosity

The relative quantity of radiation emitted by a light source.
 LUV space

A color space similar to the XYZ model,
except that the components are scaled to be perceptually linear. The
human eye perceives brightness on a logarithmic scale, and hence LUV
components are logs of the corresponding XYZ values.
 Matte shading

See Lambert's law.
 Metaball

Metaball modeling is based on the production of objects using spheres
that attract and cling to each other according to their proximity to
one another and their field of
influence (the size of their attractive field). This form of modeling may
also use cubes and other shapes, depending upon the modeler.
Metaball modeling is particularly useful for creating organic objects and
animation effects such as a group of mercury balls moving together and combining
to form an object like a soda can.
 Metameric match

Two colors that appear to be the same to a human.
They may not have identical spectral distributions, but, because
humans measure light using only three cone types, the differences
are indistinguishable.
 Microfacets

An approximation used in developing an improved specular reflection
component to surface shading. The TorranceCook
(or CookTorrance)
physical surface shading model assumes that a surface is
composed of a set of tiny planar patches, each placed according
to a distribution that depends on the surface. The
microfacet model leads to a reflection
function that gives more realistic values for the direction
and intensity of the specular component of surface reflection.
 MIP mapping

MIP mapping is a technique of precomputing antialiased
texture bitmaps at different scales, where each image in the
map is one
quarter of the size of the previous one. When the texture is viewed
from different distances, the correct scale texture is selected by the
renderer so that fewer rendering artifacts are experienced,
such as Moiré patterns.
MIP is apparently an acronym relating to the latin `multum in
parvo', meaning many things in a small place  since the texture
contains the same content at different scales. A MIP mapped texture
requires 4/3 times the storage of the original (1 + 1/4 + 1/16 +
...).
 Moiré pattern

A watered appearance usually
provided as texture on the surface of objects.
It arises from the interference between two overlapping patterns
with a similar spatial frequency.
 Monochromatic

Light (or other source of electromagnetic radiation) having only one
wavelength.
 Morphing

A continuous deformation from one keyframe or 3D model to another.
In 3D this is often achieved by approximating a surface with a
triangular mesh that can then be continuously deformed. In 2D, it is generally
performed by either distortion or deformation.
 Munsell color system

The Munsell colororder system is a way of precisely specifying colors and
showing the relationships between colors.
In this system, the color space has three parameters : hue, value and chroma (saturation).
Munsell uses scales with visually uniform steps for each of these parameters.
A Munsell Book of Color displays a collection of colored chips arranged
according to these scales.
The parameters are written in form, known as the Munsell notation.
(See HSV.)
 Nit

An equivalent name for the unit of luminance: candelas
per square meter.
 NURBS

NonUniform Rational BSplines.
A class of piecewise parametric curves or surfaces
where each curve segment or surface patch is
described by a ratio of NonUniform BSpline
polynomials. Bsplines are a class of polynomials whose coefficients
depends on a set of control points. For the Uniform
BSplines each curve segment or surface patch is defined by a
parameter domain of fixed length or area respectively, whereas
in the NonUniform BSplines the parameter domain does
not have to be uniform. The NonUniform characteristic allows
different levels of continuity between the curve segments
and the surface patches whereas it is restricted to p1 levels
for the uniform case, where p is the degree of the polynomial.
Thus, NonUniform BSplines can interpolate points more
accurately. Furthermore, rational forms can represent conic
curves and are invariant under rotation,
scaling, translation
and perspective transformations.
NURBS provide a superset of commonly used surfaces and has been
adopted as the IGES (Initial Graphics Exchange Specification)
for freeform surfaces.
 Occlusion

Visual obstruction. An occlusion occurs when an opaque surface
prevents another surface from being seen. When rendering, it is
necessary to determine which surfaces are not occluded, a problem
known as the hidden surface problem.
 Octree

A spaceoccupancy representation used for
representing 3D volumetric
objects. It is a hierarchical representation, designed to use less
memory than representing every voxel of the object explicitly.
Octrees are based on subdividing the full voxel space containing the
represented object into 8 octants by planes perpendicular to the three
coordinate axes. Octants that completely contain a single object are
denoted as being pure. Octants that contain multiple objects are
recursively split into 8 new smaller octants. This splitting continues
until all volumes are either pure, or some volume size limit is
reached. A tree data structure can be used to represent the octree.
Normally, the octree data structure will have only about as many nodes as
there are
voxels on the object surface, which can be much less than the total
number of voxels in the object. Hence, an octree representation can
save a lot of space when representing an object or scene.
 Onesided surface

A surface rendered in such a way that only one side is visible.
That side is usually facing the same direction as the surface normal.
 Opaque

Impervious to light. An opaque surface
will reflect light to some degree dependent on surface attributes. See also
Specular reflection and Diffuse reflection.
 Orthographic projection

A type of parallel projection where the direction of projection is the same
as the surface normal to the projection plane. Specialist types of
orthographic projection include frontelevation, topelevation and
sideelevation where the projection plane is perpendicular to the principle
axis. Such projections, called isometric projection,
are often used in engineering drawings as they preserve distances and angles.
 Overlay

An image compositing method where
an image is displayed over a background image.
 Painter's algorithm

An algorithm for hidden surface removal, where objects are
assigned priorities based on proximity to the camera position. When the
image is rendered to the buffer the objects with higher priority overwrite
those with lower priority. Although intuitive and simple to implement, this
algorithm has been superseded by zbuffering.
 Palette

The set of colors that may be used to compose an image.
 Pan

Camera rotation about an axis (vertically) perpendicular to the camera's view direction.
 Parametric

An approach to shape representation in which a curve or a surface is
defined by a set of equations expressed in terms of a set of
independent variables (i.e. the parameters).
This representation is convenient for curvature and bounds
computation and the control of position and tangency.
 Parametric surface

A surface defined explicitly by the range of values of a parametric function.
For a parametric function that depends upon the parameter
vector , the surface S can be defined formally as:
 Particle system

A technique for modeling irregular natural structures by a collection
of independent objects, often represented as single points.
Objects that have been represented using this technique include
fire, smoke, clouds, fog, explosions, grass, etc.
Each particle will have its own motion and property parameters,
usually drawn randomly from a distribution (perhaps constrained by or
linked to other particles, or other scene objects, such as grass being
constrained to grow from a specified surface). Because natural effects
based on particle systems need many particles for realistic appearance,
rendering of particle systems often requires specialpurpose methods
that exploit the properties of the particular particle system.
 Path tracing

Path tracing is an improvement on general raytracing techniques.
Normal raytracing uses a constant factor to estimate the contribution of
ambient light at a given surface point but pathtracing estimates the
global illumination using, for example, Monte Carlo techniques.
Images are thus generated using many paths through each pixel. Note that a
degree of oversampling is always necessary, so this technique is
computationally expensive.
 Penumbra

That part of a shadow due to a light
source which receives partial illumination from the source. By definition the
source will be an extended light source and the penumbra always surrounds the
umbra.
 Perspective projection

Perspective projection is the complete projection model of a scene onto
an image plane via a pinhole camera model.
The perspective projection of any set of parallel lines which are not
parallel to the projection plane converge to a vanishing point. In
3D, the parallel lines meet only at infinity and there is an infinity of
vanishing points, one for each of the infinity of directions in which a
line can be oriented.
 Phong shading

A shading model for surfaces based on the interpolation
of local surface normals at the vertices of a triangular patch.
The technique is used for more realistic rendering of glossy surfaces.
 Photorealistic rendering

The process of rendering images so that they
closely resemble a photograph.
Such renderings must take into account reflective properties,
light sources, illumination,
shadows, transparency and
mutual illumination.
 Photometry

Making measurements from images. One example is creating a
3D scene description using stereo image analysis, and measuring the
volume of an object in the model.
 Pixel

A single discrete sample point of an image. Image size and resolution
are defined in terms of number of pixels.
 Pixel depth

The number of bits used to generate a color at each pixel.
The number of different colors that can be displayed is
equal to . For instance a pixel depth
equal to one means that only black and white colors could be displayed;
with a pixel depth equal to four, sixteen different colors could be
displayed.
 Plenoptic function

The plenoptic function is the 5dimensional function representing the
intensity or color of the light observed from every position and
direction in 3dimensional space.
 Point light source

A mathematically defined infinitely small point l from which light
radiates. The point might be at infinity, in which all light
rays are parallel, or it might be closer to the object, in which
case light rays radiate outward in all directions. The amount of
light radiated in different directions need not be uniform.
 Point sampling

Point sampling algorithms are those which only solve for visibility at a
finite number of discrete points.
A typical example is ray tracing.
They are generally used in simple renderers.
There are four point sampling algorithms in common use today: zbuffering,
painter's algorithm, ray tracing and the
scanline algorithm.
Other point sampling algorithms are generally variations on these.
They have advantages over continuous algorithms because they are easier to
understand and implement, faster and can generate a greater range of
optical effects.
It is difficult to generate
photorealistic fully antialiased
images using a pointsampling algorithm.
 Polygon

A plane figure which is a closed
contour of straight lines. A basic primitive in the graphical representation of
objects.
 Polygon fill

A series of ordered planar vertices connected to form an enclosed area. This
area is then completely rendered using a specified color or texture.
 Polyhedron

A 3D solid that is bounded by a set of polygons whose
edges are each a member of an even number of polygons.
 Polyline

A continuous line formed from one or more connected line segments.
Polylines are specified by the endpoints of each segment.
 Portals

A method for reducing framebuffer overdraw where visible areas of a 3D
model are clipped before they are rendered. Small areas
in the model are
grouped as sectors and portals are the transition planes between them.
An initial view frustrum is defined to be as large as the image
plane.
Then all visible polygons are clipped to this volume.
A sector is rendered
only if its portal is within the clipping frame. Then a new smaller
frustrum is defined at that position and polygons visible from it are
clipped and so on, recursively.
 Procedural surface

A procedural surface uses external parameters supplied to a model that
determines how the surface will be generated. For example, a procedural
surface that generates a polygonal representation of a sphere at a specified
detail is procedural; the actual surface is generated by the specified
sphere diameter and the number of polygons that will make up the surface. An
advantage of using this approach is efficient storage and replication since
individual polygons need not be explicitly specified.
 Procedural texture

A texture generated by a model controlled by external parameters.
 Pseudocolor

Pseudocoloring recolors pixels with colored values as
a function of the grey level value in the original monochrome image.
Pseudocoloring is
used because of the limitation of the human visual system to
distinguish all the brightness range values.
 Purity

The degree to which a color is saturated.
 Quadric surface

A curved surface defined by the equation
Special cases of the surface include spheres, cones, cylinders,
ellipsoids, hyperboloids, etc.
The translation, rotation and scaling of a quadric surface is easy,
as is the calculation of its surface normal, intersection with
a ray and calculation of the zvalue (given the x and y values).
 Quadtree

A tree structure used to encode twodimensional spaces, such as images.
The image is recursively subdivided into subquadrants.
At each subdivision the subquadrants are assigned a ``full'',
a ``partially full'' or an ``empty'' label depending on how much
the quadrant intersects the region of the interest in the
image. The subdivision of partially full subquadrants continues
recursively until all the
subquadrants are homogeneous (full or empty) or
a predetermined cutoff depth has been reached.
The edges of the tree represent the different subquadrants
and the nodes to which they point represent the subquadrant labels.
 Quantization

1) The subsetting of data or a resource
to enable or speed up processing. An example of the former is where a
device has no more than an 8 bit color capability thus requiring a 24 bit image
to be requantized to 8 bit color for processing. Subsetting
large data sets can also speed up processing. An example of resource
quantization is where the processing of a screenful of data in an imagebased
algorithm can be made much more efficient by subdividing the screen, perhaps
on a binary basis, and applying the algorithm to smaller sections of the data.
2) Converting a continuous quantity into series of discrete values. For
example, continuous images can be quantized into discrete pixels,
color spaces can be quantized into a set of discrete colors, or
continuous time can be quantized into discrete steps.
 Radiance

A measure of the amount of electromagnetic radiation leaving
a point on the surface. More precisely, it is the rate at which light energy
is emitted in a particular direction per unit of projected surface area. The
projected surface area is the projection of the surface onto the plane
perpendicular to the direction of radiation. It is found by multiplying the
surface area by ,
where is the angle of the
radiated light to the surface normal.
 Radiosity

A image rendering algorithm that allows diffuse and
mutual illumination effects by evaluating the radiation of
light from light sources and reradiation amongst surfaces.
Radiosity calculations determine the steady state in the
radiative transport of light around a closed volume. Essentially, the
illumination leaving a patch is a proportion of the light reaching the
patch from all the other visible patches in the closed volume. Patch
surface normals are typically distributed everywhere
and some patches are occluded
or partly obscured from each other. The accumulation of these
radiationattenuating effects is summed up as the formfactor
between each pair of patches. The main and most timeconsuming
part of the radiosity calculation is the calculation of these form factors.
 Raster coordinates

Raster coordinates are an artifact of the method of CRT image
reconstruction where pixels are addressed and illuminated in a
toptobottom, lefttoright fashion.
Hence, raster coordinates are the 2D coordinates of the current drawing
position either in the image window or the hardware frame buffer.
 Ray tracing

A rendering paradigm that aims to produce realistic
images (rather than realtime) given a 3D model. The color of a
pixel is determined by calculating the path of a ray of light
passing through a point in the 3D model corresponding to the
pixel.
The path is traced back to a light source.
 Recursive decomposition

An algorithm where space is divided into successively smaller pieces
until a threshold is found. These algorithms can be used to draw
curves by approximating them by a chain of line segments.
This can also be used to render surfaces by subdivision algorithms,
such as the hierarchical Bspline
refinement algorithm.
 Reflectance

Reflectance is a measure of the ability of a surface to reflect
electromagnetic radiation ( e.g. light). It
is equal to the ratio of the reflected flux to the incident flux.
 Refraction

The phenomenon of a beam of light bending as the light's velocity changes.
This occurs when the refractive index of the material through which the
light is passing changes.
Let i be the normalized incident ray vector (pointing towards
the surface), which has unit surface normal n.
If t is the transmitted (refracted) vector inside a
transparent medium, then:
where
is the ratio of the
refractive indices of the inside and outside media.
(See Snell's law.)
 Render

To create an image from a description of a scene, its objects
and light sources and the viewer.
 Resolution

This indicates the number of pixels per image. It is often represented
in this format: where
N and M are the number of pixels per column and row respectively.
 Retroreflector

A type of surface with unusual reflectance characteristics, namely that
it reflects light mainly back in the direction from which it came.
This makes retroreflecting surfaces appear much brighter than matte
surfaces, if the light source is in the same direction as the
viewer, and dark otherwise.
Retroreflecting surfaces are often found on road markings and signs.
 RGB color model

The RGB (``red'', ``green'', ``blue'') color model describes a color as a
positive combination of three appropriately defined red, green and blue
primaries. If the r, g and b components are defined as scalars
constrained to a value between 0 (no intensity) and 1 (maximum
intensity) all the definable colors will be bounded by a cube and it is
typical to describe RGB combinations as coordinates on the cube (r, g,
b). For example pure red is (1, 0, 0) and the secondary color cyan is
(0, 1, 1); darker colors have values closer to (0, 0, 0) (black) and
lighter colors have values closer to (1, 1, 1) (pure white).
 RGB true color

An RGB color system with
24 bits per pixel color resolution. This gives a choice of over 16 million colors
per pixel. Such a system is generally known as a true color or
full color system.
 Rotation

A rotation is a geometric transformation that changes the
orientation of an object, extended light source
or viewpoint.
Specific rotations are often represented by a matrix R,
which then transform point p to the new position Rp.
Rotation and many other simple transformations can be done simultaneously if positions and
directions are represented in homogeneous coordinates.
 Saturation

A perceptual term referring to the colorimetry quantity 'excitation
purity' of a color. Hue can be used together with saturation and
luminance to define a HSL color space.
 Scalar

A quantity which has magnitude but no direction.
 Scaling

The process in which the size of an image or geometric representation is
modified by multiplying each component of the representation's coordinates
by constant factors.
Scaling and many other simple transformations can be done simultaneously if positions and
directions are represented in homogeneous coordinates.
 Scanline algorithm

An algorithm that renders an image one row at a time, e.g.
generates the image values for pixels lefttoright as it
scans across the image. After one row is generated, the algorithm
proceeds to the next row. One advantage of this algorithm is it
can use less memory to generate the results for only single rows
at a time. Another potential advantage is a reduction in computation
as the set of object primitives that need to be rendered at each pixel
along a scan line may not change very often, so some results
calculated at one pixel can be used at the next.
 Screen door transparency

A technique for rendering the transparency of an object.
The key idea is to only render some of the pixels associated with the
object, depending on how transparent the object is.
 Sculptured surface

A highly flexible surface
generated by the combination of surface patches which have both their boundary
curves and interior blending functions defined by polynomials of, usually,
at least order three. See also Bspline and Bézier curve.
 Selfocclusion

A surface is selfoccluding when:
a) Light cast from behind the surface does not illuminate it.
b) The light source is in front of the surface but some closer portion of the
surface blocks the incoming light.
c) The light source is in front of the surface and the surface is
illuminated, but some closer portion of the surface blocks the light
coming from the surface.
 Shading

Coloring a surface according to its
incident light. The color depends on the position, orientation and
attributes of both the surface and the sources of the illumination.
(See also Lambert's law, Phong shading
and smooth shading).
 Shadow map

A shadow map is a precomputed array used to test whether points
on object surfaces are in shadow. The array contains depth values from the
viewpoint of a point light source giving the distance to the
first object surface encountered. If a given pixel in the environment is
not contained in the array then it is in shadow. This method is useful for
quickly rerendering an image from several different viewpoints or when
several light sources are used  each would then have its own shadow map.
 Skeleton

A framework capturing the structure of an object or shape constructed from a
series of points connected by thin lines. In a similar way to a
wireframe
representation, a skeleton is used to increase the performance of the
rendering system since it is not necessary to render solid surfaces.
Objects
represented using a skeleton can be given a skin by specifying a diameter
from the skeleton used to render the surface.
In cartoon animation, a skeleton is literally a line structure
representing the position of the limbs of a figure and is not
necessarily oriented along the medial axis.
 Smooth shading

A method of polygon shading where calculations are performed for
the vertices and values for pixels inside the polygon are derived from
linear interpolation of the vertex values.
 Snell's law

A law defining how light is bent or refracted when it passes through
a boundary between two dielectric media of different indices of refraction,
such as air and glass or air and water. It is expressed by
where and
are the index of refraction of the
two media.
and
are the angles which the boundary
surface normal makes with the incident light ray and the refracted light
ray respectively.
 Spatial navigation

The process of orienting and moving through a virtual environment.
 Spatial partitioning

A technique used to divide a large task into a series of smaller ones. The
basic approach is to devise a preprocessing stage which determines
spatially coherent groups for processing. This strategy is particularly
appropriate for parallel architectures where the groups can be sent to
different processing units.
 Specular reflection

One component of light reflection at a surface point (see also diffuse reflection).
Specular reflection is observed on ``shiny'' surfaces and is characterized by
highlights on the surface.
The amount and direction of specular reflection depend on the directions
of the incident light
and the viewing direction with respect to the surface normal.
 Spline curve

A spline curve is defined using a set of control points
(). Every control point
has an associated blending
function, ,
which is described within each span (,
). The blending
function is a continuous piecewise polynomial, continuous at each knot and
weighted by the polynomials. This gives a curve ,
which is the union of the piecewise polynomials where all segments meet.
 Staircasing

Lines are scanconverted to fixed
pixel grid points. The illuminated pixels often do not lie on the true path
of the line. The result is that displayed lines are normally jagged in
appearance, an effect commonly known as the jaggies or staircasing.
The effect can be reduced or eliminated by antialiasing.
 Steradians

The unit of solid angle. The solid angle corresponding to all of space
being subtended is steradians.
Solid angle is defined as the surface area of a unit sphere which is
subtended by a given object.
 Stereo

The use of two images to generate a 3D description.
E.g. two slightly different images are displayed in a each eye
of a virtual reality head mounted display in order to
induce an impression of 3D. Stereo matching is a process by which
two images of the same scene are compared in order to deduce 3D
information.
 Stochastic sampling

A method of reducing the visual effects of aliasing by sampling in an
irregular manner, rather than on a regular grid. Recognizable aliasing
artifacts are replaced by noise, which viewers find less objectionable.
(See also jittering.)
 Superconic

Generalization of conic curve in which the trigonometric terms in
the formula of the curve are raised to an arbitrary power to control
the smoothness of the curve. It can be expressed by:
 Superquadric

A class of parametric surfaces, derived from the class of
quadric surfaces, in
which the
trigonometric terms of the quadric equation, written in
parametric form, are
raised to a power. The exponents are known as the
squareness parameters and are used to pinch or square off parts of the
original quadric shape.
A special case are the superconics.
 Surface normal

Any surface that is smooth enough for at least one derivative
calculation at a given point has a surface normal.
This is a unit vector n that is perpendicular to the plane
tangent to the surface at the given point. It is usually taken to be
pointing
outward away from the surface. Smooth surfaces have surface normals
at every point. Planar surfaces have the same surface normal at every
point that is not at the edge of the surface. At crease or fold edges,
the surface normal is undefined.
 Surface patch

This term has several usages in the graphics
community: 1) a small piece of surface with arbitrary shape and size
surrounding a surface point with a given surface normal or
2) a primitive element of a geometric surface description, such as a
spline or triangulation patch.
Graphics techniques that use the different surface patch representations
are mainly related to surface representation, visibility analysis,
illumination and reflectance.
 Sweeping

The definition of a new object in a higher
dimension produced by arbitrary movement of the originating object along a path
in the space of the higher dimension.
For example, one can create a cylindrical surface by sweeping a line about another line which is parallel.
 Tesselation

A technique to construct a surface by a small set of figures which fit
together. They are drawn repeatedly over the entire plane leaving no gaps.
 Texel

The fundamental element of a texture map.
 Texture map

A bitmap used to texture a 3D polygon model, including
adjustments for perspective correction, where vertices of the object model
are mapped onto the 2D texture bitmap.
In addition to color and brightness, textures may also be encoded with
the properties of transparency and specular reflectivity.
This kind of texture may also be procedural in nature.
A possible sideeffect of texture mapping occurs unless the renderer can
apply texture maps with correct perspective. Perspectivecorrected texture
mapping involves an algorithm that translates texels, or pixels from
the bitmap texture image, into display pixels in accordance with the
spatial orientation of the surface.
 TorranceCook (or CookTorrance) shading

A shading model that incorporates an
ambient lighting component, a diffuse component
(see Lambert's law) and a specular component.
 Translation

Point M can be moved, or translated, to a new location M' by adding
a vector T.
More concisely : M' = M + T.
Translation and many other simple transformations can be done simultaneously if positions and
directions are represented in homogeneous coordinates.
 Translucent

A characteristic of a material allowing light to pass
through partially or diffusely.
 Transparency

The ratio of the amount of light passing through a material to the amount of
light incident on the material.
 Triangulation

The transformation of a model into a mesh of triangles to facilitate
speedy rendering or other computational geometry algorithms. The initial model
might be a planar graph, freeform surface, polygonal model,
point cloud data or volumetric data.
 Trilinear filtering

A level of detail blending technique used in
MIP texture mapping.
Pixels are taken from multiple
MIP maps and blended to produce the final color. The purpose is to remove
the bands between adjacent pixels taken from different MIP maps.
 Umbra

The part of the shadow created by an extended light source that is
entirely cutoff from the source. It is surrounded by the penumbra that
receives some light from the light source.
 Vanishing point

A point in a perspective projection where parallel lines
not parallel to the projection plane
converge. A finite 2D projection of a point at infinity in 3D.
 Vector

A list of numbers, typically Cartesian coordinates or a direction
in 2D or 3D. E.g.
.
 Vector graphics

The earliest computer graphics displays were drawn on socalled vector displays,
because the electron beam which produced the image was under software control.
The beam followed a chain of vectors (i.e. a
polyline) from one point to another.
Vector graphics is sometimes referred to as linedrawing graphics
 Vertex

The points in a model at which edges terminate. E.g. the eight
corners of a cube, or the three corners of a triangle. Polyhedrons,
polygonal surfaces and triangulations are
composed of vertices, edges and faces.
 Vertex normal

The direction vector pointing directly out of a polygonal/polyhedral model
at a given vertex. This may be defined as the
average of the surface normals of the faces adjacent to
the vertex.
 Viewpoint

The location of a virtual camera in a model.
 Virtual camera

A set of parameters defining a 2D view of a 3D model. These might include:
camera location, direction, camera twist  defining the upwards direction in
the rendered image.
 Virtual environments

An artificial environment maintained by a computer which a user
may interact with or view.
 Virtual reality

A simulation of a virtual environment which
according to some must have an
'immersive' quality encouraging the feeling of being present in the
environment. Technology used with
virtual reality includes stereo image helmets, 360 degree screens
but may be as simple as a standard monitor display.
 Visible surface determination

During rendering of 3D scenes it is
necessary to determine which objects occlude others in order that the scene
looks correct, and time wastage may be prevented by not drawing shapes that
will be overdrawn. Techniques include: culling backfacing polygons,
Zbuffering and Warnock's algorithm.
 Volume rendering

The visualization of 3D volume data. E.g.
data sets such as MRI scans consisting of a volume of density
samples or voxels.
 Voxel

Volume element. A single datum in 3D volume data.
 VRML

Virtual Reality Modeling Language. A 3D
model description format suited to transfer on the WWW.
 Warnock's algorithm

A spatial partitioning technique for depth sorting a list of polygons so that they may be
rendered correctly. The algorithm subdivides the screen rectangle until it may be
painted entirely in the color of the foremost polygon or the background color.
 Warping

The manipulation of 2D images by arbitrary geometric
(i.e. position) transformations of the pixels
of some or all of an image. Some simple types of warping are
stretching, scaling, rotating,
skewing, shearing or perspective
transform (perspective projection).
This may be used to draw texture maps.
Many simple transformations can be done simultaneously if positions and
directions are represented in homogeneous coordinates.
 WeilerAtherton algorithm

A technique for clipping one generalized polygon
with the boundary of another.
 Wireframe

A minimal vector graphic rendering style in which
only the edges of shapes are drawn. This is appropriate for
polygonal objects, although many other surface representations may be
quickly converted for faster rendering, typically for editing purposes.
 XYZ color space

XYZ color space. A color model in which X specifies the red component and
Y the green component. The blue
component is 1XY (the color components are scaled so that R+G+B=1). Z
specifies the brightness.
 YIQ color space

A chrominance/luminance
color space model used in the American NTSC television standard,
Y specifies luminance, I and Q specify
chrominance. I specifies the redorange/cyan (or bluegreen) component, and Q specifies
the green/magenta (or purple) component.
 YUV color palette

A chrominance/luminance
color space model used in the British PAL television standard,
Y specifies luminance, U and V specify
chrominance. U specifies the blue/yellow component, and U specifies
the red/cyan (or bluegreen) component.
YUV is also called .
 Zbuffering

A technique for speeding up depth sorting
(See visible surface determination)
while rendering. As each primitive in
the frustrum is drawn, the distance
from the viewpoint is recorded in the Zbuffer
or depth buffer.
If a pixel has already been drawn with a closer Z value the new pixel value is not recorded.
 Zoom pyramid

A data structure which stores an image at multiple size/resolutions. The zoom pyramid for a 640x320
image would include versions with sizes 240x160, 160x80, 80x40, etc.
Zooming in to
the image quickly is therefore possible.
 Zooming

Viewing an image at different sizes. Zooming in creates an
enlarged view of a portion of the scene in the image frame. Zooming out
does the reverse.
Date of last change: Sept 12, 1999