Lecture 1: Graphics Systems and Models

Topics: 1. Applications of computer graphics 2. Graphics systems 3. Camera models Chapter 1 of Angel.

Applications of computer graphics

1. Display architectural drawings, e.g. plan of a building maps: geographical information plotting statistical graphs, e.g. share prices medical images: Computed Tomography (CT), Magnetic Resonance Imaging (MRI) scientic visualization

2. Design (interaction important) Computer Aided Design (CAD) design of very-large-scale integrated (VSLI) circuits

3. Simulation ight simulation for training pilots computer games television and computer-animated lms: Toy Story (Pixar), Ice Age virtual reality

4. User interfaces window-based operating systems: Microsoft Windows, Macintosh, X Windows internet browsers: Netscape, Explorer

A Graphics System
1. 2. 3. 4. 5. Processor Memory Frame buer Output devices Input devices

Processor

Frame buffer

Memory

Pixels and the Frame Buer Most graphics systems are raster-based. The raster is an array of picture elements pixels. The pixels are stored in the frame buer. The depth of the frame buer = num. bits used per pixel. 1-bit-deep black and white only. 8-bit-deep 28 = 256 colours. 24-bit-deep the RGB-colour system: red, green, blue, 256 shades for each. The resolution of the frame buer = number of pixels. Rasterization or scan conversion is the assignment of values to pixels in the frame buer that best represent graphical primitives: lines, circles, polygons, etc. Done by the processor. Sophisticated graphics systems use special-purpose processors for this.

Output devices The usual display type is a cathode-ray-tube (CRT). Pixels in the frame buer are displayed as points on the surface of the display. The points last only a few milliseconds, so the content of the frame buer must be redisplayed or refreshed. The rate must be high enough to avoid icker: the refresh rate. There are two ways of displaying pixels: non-interlaced: row by row, i.e. scan line by scan line, typically 50 to 80 Hz (times per second). interlaced: odd and even rows are refreshed alternately, used in television. For example, at 60 Hz, the screen is redrawn only 30 times per sec.

y deflect

Electron gun

x deflect

Phosphor

Focus

What is an image? A combination of objects, lights, and a viewer (or camera).

Ray tracing Trace rays from each light source and see whether they reach the viewer. Those rays that do form the image. We could do ray tracing in computer graphics but it is computationally extremely expensive. In practice we avoid ray-tracing by simplifying our models, for example by assuming all objects are equally bright (i.e. there are light sources everywhere). Then each object (e.g. triangle) looks like a light emitter. It is possible to add light sources and material properties and still avoid ray-tracing.
B F C

E D A

The pinhole camera:

(xp, yp, zp ) d

a box with a small hole in the centre of one side with the lm placed at the other end. It is easy to model geometrically. Orient the camera along the z axis with the pinhole at the origin. Suppose the camera has length d. Assuming the pinhole allows only one ray of light from any point (x, y, z ), the ray is clearly projected to the point (xp , yp , d), with coordinates xp = dx , z yp = dy . z

The point (xp , yp , d) is called the projection point. In an idealized model, the colour on the lm plane at this point is the colour of the point (x, y, z ). 9

x
(x, y, z)

The eld or angle of view of the pinhole camera is the angle of the largest object that is fully visible on the lm plane. If h is the height of the camera and the angle of view then = 2 tan1 h . 2d

The pinhole camera has an innite depth of eld: every point in the eld of view is in focus. The pinhole camera has two disadvantages: (1) too little light gets in, and (2) the angle of view cannot be adjusted. More sophisticated cameras replace the pinhole with a lens. The lens allows more light to enter and dierent lenses give dierent angles of view (c.f. a wide angle lens). Lenses, however, do not have an innite depth of eld; not all distances are in focus. Like the pinhole camera, computer graphics produces images in which all objects are in focus. 10

The synthetic camera model In computer graphics we use a synthetic camera model to mimic the behaviour of a real camera. The image in a pinhole camera is inverted. The lm plane is behind the lens.
y y

Camera

(y, z ) (yp, d ) Object

(y, z )

(yp,

-d ) (a) (b)

In the synthetic camera model we avoid the inversion by placing the lm plane, called the projection plane, in front of the lens.

11

The clipping rectangle or clipping window determines the size of the image.

(a)

(b)

This is similar to the eect of the angle of view in a pinhole camera.

12

Application Programmers Interface (API) We will use OpenGL. Other possibilities are: PHIGS, Direct3D, VRML, JAVA-3D. We need to specify: Objects Viewer Lights Material properties

1. 2. 3. 4.

Objects are dened by a set of vertices, e.g. line segments, rectangles, polygons, etc. A circle can be represented by its centre and one point on the circle or by three points on the circle. OpenGL also allows the denition of more complex curve and surface types (by approximating by triangles). It also provides direct access to the frame buer. The viewer is determined by: position, orientation, focal length, lm plane. Light sources have location, strength, colour, directionality, etc. Material properties are attributes of the objects.

13

The Modelling-Rendering Paradigm It can be an advantage to model and render a scene separately, using tailored software and hardware for each task. Modelling is very interactive but user does not need to see all the detail. Rendering involves mainly number crunching and no user interaction. Pixars Renderman provides an interface (i.e. le format) between the two tasks.

Modeler

Interface file

Renderer

14

The Geometric Pipeline The four major steps in the imaging process are: 1. 2. 3. 4. Transformation Clipping Projection Rasterization

These operations can be pipelined, i.e. applied in parallel, to speed up the imaging process. Several steps can be represented by 4 4 matrices. Such matrices can be multiplied together, or concatenated, a process which lends itself to pipeline architectures and parallelism.

Vertices

Transformer

Clipper

Projector

Rasterizer

Pixels

15