puter Graphics
`
`PRINCIPLES AND PRACTICE
`Foley • van Dam • Feiner • Hughes
`SECOND EDITION in C
`
`THE SYSTEMS PROGRAMMING SERIES
`
`EX-1021
`Microsoft Inc. v. LiTL LLC
`
`

`

`THE
`SYSTEMS
`PROGRAMMING
`SERIES
`
`Computer
`Graphics:
`Principles and
`Practice
`SECOND EDITION in C
`
`Foley • \lan Dam • Feiner • l-Iughes
`
`,
`
`

`

`SECOND EDITION IN C
`
`Computer Graphics
`
`PRINCIPLES AND PRACTICE
`
`James D. Foley
`Georgia Institute of Technology
`
`Andries van Dam
`Brown University
`
`Steven K. Feiner
`Columbia University
`
`John F. Hughes
`Brown University
`
`..,..,
`
`ADDISON-WESLEY PUBLISHING COMPANY
`Reading, Massachusetts • Menlo Park, California • New York
`Don MiUs, Ontario • Wokingharn, England • Amsterdam • Bonn
`Sydney • Singapore • Tok)Q • Madrid • San Juan • Milan • Paris
`
`

`

`Sponsoring Editor: Peter S. Gordon
`Production Supervisor: Bette J. Aaronson
`Production Supervisor for the C edition: Juliet Silveri
`Copy Editor: Lyn Dupre
`Text Designer: Herb Caswell
`Technical Art Consultant: Joseph K. Vetere
`Illustrators: C&C Associates
`Cover Designer: Marshall Henrichs
`Manufacturing Manager: Roy Logan
`
`This book is in the Addison-Wesley Systems Programming Series
`Consulting editors: IBM Editorial Board
`
`Library of Congress Cataloging-in-Publication Data
`
`Computer graphics: principles and practice I James D. Foley . . . [ et
`a!.]. -
`2nd ed. in C.
`p.
`em.
`Includes bibliographical references and index.
`ISBN 0-201-84840-6
`1. Computer graphics.
`T385.C5735 1996
`006.6'6-dc20
`
`I. Foley, James D. , 1942-
`
`Reprinted with corrections, July 1997.
`
`95-13631
`CIP
`
`Cover: " Dutch Interior," after Vermeer, by J. Wallace, M. Cohen, and D. Greenberg, Cornell University
`(Copyright © 1987 Cornell University , Program of Computer Graphics .)
`
`Many qf the designations used by manufacturers and sellers to distinguish their products are claimed as
`trademarks . Where those designations appear in this book, and Addison-Wesley was aware of a trademark
`claim, the designations have been printed in initial caps or all caps.
`
`The programs and applications presented in this book have been included for their instructional value. They are
`not guaranteed for any particular purpose. The publisher and the author do not offer any warranties or
`representations, nor do they accept any liabilities with respect to the programs or applications .
`
`Reprinted with corrections November 1992, November 1993, and July 1995.
`
`Copyright© 1996, 1990 by Addison-Wesley Publishing Company , Inc.
`
`All rights reserved . No part of this publication may be reproduced, stored in a retrieval system, or transmitted,
`in any form or by any means , electronic, mechanical, photocopying, recording, or otherwise, without the prior
`written permission of the publisher. Printed in the United States of America.
`
`6 7 8 9 10-DOC-99 98 97
`
`

`

`
`
`Introduction
`
`ith th
`Computer graphics started with the display of data on hardcopy plotters and cathode ray
`erlb in
`n
`tube (CRT) screens soon after the introduction of computers themselves. It has grown to
`include the creation, storage, and manipulation of models and images of objects, These
`models come from a diverse and expanding set of fields, and include physical, mathemati-
`= engineering, architectural, and even conceptual (abstract) structures, natural phenome-
`. and so on. Computer graphics today is largely interactive: The user controls the
`at structure, and appearance ofobjects and oftheir displayed images by using input
`devices, such as a keyboard, mouse, or touch-sensitive panel on the screen. Because of the
`close relationship between the input devices and the display, the handling of such devices is
`included in the study of computer graphics.
`Until the early 1980s, computer graphics was a small, specialized field, largely because
`the hardware was expensive and graphics-based application programs that were easy to use
`and cost-effective were few. Then, personal computers with built-in raster graphics
`displays—such as the Xerox Star and,later, the mass-produced, even less expensive Apple
`Macintosh and the IBM PC and its clones—popularized the use of bitmap graphics for
`user-computer interaction. A bitmap is a ones and zeros representation of the rectangular
`array of points (pixels or pels, short for ‘picture elements’’) on the screen. Once bitmap
`graphics became affordable, an explosion of easy-to-use and inexpensive graphics-based
`applications soon followed. Graphics-baseduser interfaces allowed millions of new users to
`control simple, low-cost application programs, such as spreadsheets, word processors, and
`drawing programs.
`The concept of a ‘‘desktop"’’ now became a popular metaphor for organizing screen
`space. By means of a window manager,
`the user could create, position, and resize
`
`,
`
`1
`
`

`

`2
`
`Introduction
`
`rectangular screen areas , cal led windows, that acted as virtual graphics terminals, each
`running an application. This allowed users to switch among multiple activities just by
`pointing at the desired window, typically with the mouse. Like pieces of paper on a messy
`desk, windows could overlap arbitrarily. Also part of this desktop metaphor were displays
`of icons that represented not just data files and application programs, but also common
`office objects, such as file cabinets, mailboxes, printers, and trasbcans, that performed the
`computer-operation equivalents of their real-life counterparts. Direct manipultJtion of
`objects via "pointing and clicking" replaced much of the typing of the arcane commands
`used in earlier operating systems and computer applications. Thus, users could select icons
`to activate the corresponding programs or objects, or select buttons on pull-down or pop-up
`screen menus to make choices. Today, almost all interactive application programs, even
`those for manipulating text (e.g., word processors) or numerical data (e.g., spreadsheet
`programs), use graphics extensively in the user interface and for visualizing and
`manipulating the application-specific objects. Graphical interaction via raster displays
`(displays using bitmaps) has replaced most textual interaction with alphanumeric terminals.
`Even people who do not use computers in their daily work encounter computer
`graphics in television commercials and as cinematic special effects. Computer graphics is
`no longer a rarity. It is an integral part of all computer user interfaces, and is indispensable
`for visualizing two-dimensional (20) , three-dimensional (30), and higher-dimensional
`objects: Areas as diverse as education, science, engineering, medicine, commerce, the
`military , advertising, and entertainment all rely on computer graphics. Learning how to
`program and use computers now includes learning how to use simple 20 graphics as a
`matter of routine.
`
`1.1
`
`IMAGE PROCESSING AS PICTURE ANALYSIS
`
`Computer graphics concerns the pictorial synthesis of real or imaginary objects from their
`computer-based models , whereas the related field of ifllllge processing (also called picture
`processing) treats the converse process: the analysis of scenes, or the reconstruction of
`models of 20 or 30 objects from their pictures. Picture analysis is important in many
`arenas: aerial surveillance photographs, slow-scan television images of the moon or of
`planets gathered from space probes, television images taken from an industrial robot's
`"eye," chromosome scans, X-ray images, computerized axial tomography (CAT) scans,
`and fingerprint analysis all exploit image-processing technology (see Color Plate 1.1).
`Image processing has the subareas image enhancement, pa11ern detection and recognitiofl,
`and scene analysis and compwer vision. Image enhancement deals with improving image
`quality by eliminating noise (extraneous or missing pixel data) or by enhancing contrast.
`Pattern detection and recognition deal with detecting and clarifying standard patterns and
`finding deviations (distortions) from these patterns. A particularly important example is
`optical character recognition (OCR) technology, which allows for the economical bulk
`input of pages of typeset, typewritten, or even handprinted characters. Scene analysis and
`computer vision allow scientists to recognize and reconstruct a 30 model of a scene from
`several 20 images. An example is an industri.al robot sensing the relative sizes, shapes,
`positions, and colors of parts on a conveyor belt.
`
`

`

`
`
`11.3
`1.3
`
`Representative Uses of Computer Graphics
`
`5
`
`enormous and is growing rapidly as computers with graphics capabilities become
`commodity products. Let's look at a representative sample of these areas.
`•
`«
`User interfaces. As we mentioned, most applications that run on personal computers
`and workstations, and even those that run on terminals attached to time-shared computers
`and network compute servers, have user interfaces that rely on desktop window systems to
`manage multiple simultaneous activities, and on point-and-click facilities to allowusers to
`select menu items, icons, and objects on the screen; typing is necessary only to input text to
`be stored and manipulated. Word-processing, spreadsheet, and desktop-publishing pro-
`grams are typical applications that take advantage of such user-interface techniques. The
`authors of this book used such programs to create both the text and the figures; then, the
`publisher and their contractors produced the book using similar typesetting and drawing
`software.
`=
`(Interactive) plotting in business, science, and technology. The next most common use
`of graphics today is probably to create 2D and 3D graphs of mathematical, physical, and
`economic functions; histograms, bar and pie charts; task-scheduling charts; inventory and
`production charts; and the like. All these are used to present meaningfully and concisely the
`trends and patterns gleaned from data, so as to clarify complex phenomena and to facilitate
`informed decision making.
`*
`Office automation and electronic publishing. The use of graphics for the creation and
`dissemination of information has increased enormously since the advent of desktop
`publishing on personal computers, Many organizations whose publications used to be
`printed by outside specialists can now produce printed materials inhouse. Office automa-
`tion and electronic publishing can produce both traditional printed (hardcopy) documents
`and electronic (softcopy) documents that contain text, tables, graphs, and other forms of
`drawn or scanned-in graphics. Hypermedia systems that allow browsing of networks of
`interlinked multimedia documents are proliferating (see Color Plate I.2).
`interactive
`*
`Computer-aided drafting and design. In computer-aided design (CAD),
`graphics is used to design components and systems of mechanical, electrical, electrome-
`chanical, and electronic devices, including structures such as buildings, automobile bodies,
`airplane and ship hulls, very large-scale-integrated (VLSI) chips, optical systems, and
`telephone and computer networks. Sometimes, the user merely wants to produce the precise
`drawings of components and assemblies, as for online drafting or architectural blueprints.
`Color Plate 1.8 shows an example of such a 3D design program, intended for nonprofession-
`als; a *‘customize your own patio deck"’ program used in lumber yards. More frequently,
`however, the emphasis is on interacting with a computer-based model of the component or
`system being designed in order to test, for example, its structural, electrical, or thermal
`properties. Often, the model is interpreted by a simulator that feeds back the behavior of the
`system to the user for further interactive design and test cycles. After objects have been
`designed, utility programs can postprocess the design database to make parts lists,
`to
`process ‘‘bills of materials,’* to define numerical control tapes for cutting or drilling parts,
`and so on.
`
`Simulation and animation for scientific visualization and entertainment. Computer-
`*
`produced animated movies and displays of the time-varying behavior of real and simulated
`
`

`

`1Develapment of Harr,dw re and Software fo:r Computer ·Graphics
`
`15
`
`t b iques pe il'y g d tio
`:in · ·1en ·ty of n. ighboring
`or coLr nns. Th
`1 at edges ,of prim'tives, rathe than setting pi e.ls to maxim mo zero intensity nly;
`, 14, and 19 fl r fun:h . r di cu
`of tlf important topi, .
`Chap r,
`
`Input Technology
`1.5 2
`[nput technulo :y h.a , ;al , improved ready over th year • Th, du
`, fTagi le lj h · pen of
`evel ;ped by office-
`voo1or y tern ha bee:
`repla d by th ·. ubiguJt u n
`.
`ut mation pi ne
`. · d m hm 1, and the
`· · ·
`· rnbart in the mid~ i.xd
`[ENGE68JI),
`·
`t devices that
`transparent. · · ·
`mo n.ted n the , · .. , en
`·upply n t ·
`he - ·reen, but al o 30 a
`aJ in.p t.
`·
`, boo m·ng m
`. ·· udio
`. ·m
`u~ . ( egre
`o mm nrca:tfon also ha.... ex ili pot· ntiaJ, ~nee i:t allows hands-free input nd natuiram
`, fi .. ·
`outp ·
`im · • uuc1;· o
`and o on. With ·
`tandard inpu cf vio • th
`pj
`-'-
`· .
`' ~. in:g new in.fbrmaf on or
`by typin
`user • _ · · i
`ation
`·
`b pointing ro
`·. · in:formarion
`screen. These ino
`tio, require n !knowledg
`o · progr · · ·,
`y a little keyboard use·: ·
`·
`imply by electing
`- nu butt
`·. . -
`·
`few driara te. - 1in
`by 'h
`·. 1. n !
`rm p~
`-a ing conse utive
`~ · ra\VS · .
`, . -
`12 ve • paints by
`lirl or
`, · lated b "moorh
`p in
`. bmm eel by potyg, n or p~t'nt
`moving the c
`and fill
`ar
`,ntoun with
`
`.
`
`·. r lhe
`. of _ ~
`
`o
`
`·
`
`·
`
`·
`
`· · · · c ·
`11h
`·
`
`·
`
`1.5.3 Software Portab~llcy· and Graphics Standards
`· teady adv.me
`·
`· . dware te hnol y ave thus made
`Je the evo ut · on
`from
`displ ry:
`a-kind p
`iall · u put
`•
`· .
`' , · h moo, in ·. ·
`mputer.
`. l 'W'Onder hether
`•
`· if · · aim
`.· .
`· ·
`d ex.pen
`· u tie
`extent have
`resolved?' · .·
`y rem · and ·
`aihi -
`·
`ft ·
`· ·. it
`primitive grap
`- d
`th
`has been
`·
`prooess o mat
`· n uch ·. oft.ware. ~
`.
`·.
`l!JlppJied by rnanu acwrers for their particular di
`.
`o • · ·
`.
`pend m
`d ive a
`i p
`p
`pJ
`pri.nre · .
`m re
`an h~.,
`-tj
`. ing a devic.e-i dependent packag in c-0 ~u
`· w·th a higb-l :vet
`pur
`o
`prom · , appliootion~progrom pona · · '· . • Thi · portabiWity i ·
`programm · n.g
`u
`·
`way . · a high-leve.1 1 m hine-inde1
`pl"(} · · 100 in m
`nt l nguag ( u h
`'l
`t' ng di programmer from mosll:
`FORTRAN
`' -
`· or C) provid portability: b
`·
`machine ' -
`a d ·p
`·· · "i, I n · ua ·
`readily amp 'emented n
`I ro d
`fe
`-r portabmly' ' i
`rag
`f p -
`• 'Pti gJ
`nhooc d in th:at pro ra,nm · can
`n ro • , taUabo ,, nd fir d fanuliar
`· en to yst 01 ,or even rorr ins .
`now m ve ,
`oft-ware.
`f th ne d for
`in uch d vice-independent g · phic
`tandard
`A general awarene .
`- - i cation for a D
`rrties and culminat d in a
`packages aw s
`re
`in the mid-
`
`y
`v
`
`

`

`166
`
`Graphics Hardware
`
`graphics processor with a separate pixmap is introduced, and a wide range of graphics(cid:173)
`processor functionalities is discussed in Section 4.3.3. Section 4.3.4 discusses ways in
`which the pixmap can be integrated back into the CPU's address space, given the existence
`of a graphics processor.
`
`4 .3.1 Simple Raster Display System
`
`The simplest and most common raster display system organjzation is shown in Fig. 4.18.
`The relation between memory and the CPU is exactly the same as in a nongraphics
`computer system. However, a portion of the memory also serves as the pixmap. The video
`controller displays the image defined in the frame buffer, accessing the memory through a
`separate access port as often as the raster-scan rate dictates. In many systems, a fixed
`portion of memory is permanently allocated to the frame buffer, whereas some systems
`have several interchangeable memory areas (sometimes called pages in the personal(cid:173)
`computer world). Yet other systems can designate (via a register) any part of memory for the
`frame buffer. In this case, the system may be organized as shown in Fig. 4.19, or the entire
`system memory may be dual-ported.
`The application program and graphics subroutine package share the system memory
`and are executed by the CPU. The graphics package includes scan-conversion procedures,
`so that when the application progrdm calls, say, SRGP_IineCoord (xl , yl , x2, y2), the
`graphics package can set the appropriate pixels in the frame buffer (details on scan(cid:173)
`conversion procedures were given in Chapter 3). Because the frame buffer is in the address
`space of the CPU, the graphics package can easily access it to set pixels and to implement
`the Pix Bit instructions described in Chapter 2.
`The video controller cyc.les through the frdme buffer, one scan line at a time, typically
`60 times per second. Memory reference addresses are generated in synchrony with the
`raster scan, and the contents of the memory are used to control the CRT beam's intensity or
`
`CPU
`
`Peripheral
`devices
`/\
`
`<
`
`~
`
`I DUI
`
`>
`
`System
`memory
`
`Frame
`buffer
`
`Video
`controller
`
`8
`---
`
`Fig. 4. 18 A common raster display system architecture. A dedicated portion of the
`system memory is dual-ported. so that it can be accessed directly by the video
`controller, without the system bus being tied up.
`
`

`

`4.3
`
`Raster-sca n Display Systems
`
`167
`
`CPU
`
`(
`
`Peripheral
`devices
`A
`
`......
`
`System
`memory
`
`Video
`controller
`
`Monitor
`
`Fig. 4 .19 A s imple raster display system architecture. Because the frame buffer may
`be stored anyw here in system memory, the video controller accesses the memory via
`the system bus.
`
`color. The video controller is organized as shown in Fig. 4.20. The raster-scan generator
`produces deflection signals that generate the raster scan; it also controls the X and Y address
`registers, which in turn define the memory location to be accessed next.
`Assume that the frame buffer is addressed in x from 0 to x,.. and in y from 0 to y.,.,.;
`then, ai the start of a refresh cycle, the X address register is set to zero and the Y register is
`set to Ymu (the top scan line). As the first scan line is generated, the X address is
`incremented up through x...,.. Each pixel value is fetched and is used to control the intensity
`of the CRT beam. After the first scan line, the X address is reset to zero and the Y address is
`decremented by one. The process continues until the last scan line (y = 0) is generated .
`
`....
`
`....- X llddl II
`
`~ Set or increment
`
`M
`
`u.. ~
`• ~
`llddt u ~
`
`m
`
`0
`
`r ,
`
`'1'._
`
`........ ~
`
`RM.I'«<It
`
`Horizontal
`and vertical
`deflection
`signals
`
`L-
`,....
`
`r--
`I.- Y addlcu
`
`~ Set or decrement
`
`Data
`
`-.
`
`Pial
`'illlue(l)
`
`Intensity
`or color
`
`Fig. 4 .20 logical organization of the video controller.
`
`

`

`3.50
`
`lrnpld Devices. Techn,:que -. and Interaction Tasks
`
`th
`m
`
`· .mi ht compare interacti.on techniqu - u ing differelll d . ·ces
`I.
`t k. Thus we might a ·m1 th.a experienced users Ml o .' n nter 1comrnand
`am
`or .a key
`ard 1than via m _ nu
`I • lion
`that u rs can pi
`· - · · via runcti -o
`y us'ng a. mou
`~1ec,ts m re q
`than th -y can usm a joysti.ok or -u or
`
`·
`
`l
`
`· ·
`
`·
`
`·
`
`· I
`
`.
`
`· ·
`,. · . . ·.
`
`. ui
`_
`p .rt
`· y with . no
`1n
`noecting d ioe to _ mpu,te
`; b
`_
`.
`ronou R - ·. · 2 termma1 int rface gen -rally m kin
`
`,
`
`· . also
`• ·
`
`nunon
`in , rf iDg
`
`'ntera u n ·
`l, we con jder
`d I ,gue l
`·u t indi · --
`talce f m :
`· n _
`h task . Hand m · • ,-
`an with ursor-cont ,
`-rall fast r with
`·, .are ahiead
`r th n m u e i
`- ·trol
`m
`m n ed to be on the .keyboard
`t t k in equ n after th 01Jrsor
`· .
`r ·
`pier 9. h re ·we d'eru with coo t
`_ . e
`i ail
`d · cus.~
`_
`·
`w _r i· .e.
`· nfusion ,can
`uildiog b~ock intmduood ·n
`p
`, voided
`'
`·. d
`··
`·
`eep
`·n mmd.
`l
`·
`I vel d.
`are th · elev· e
`.
`lp
`0
`. . and ·
`d
`ma·nt in bi ·
`· ,e
`Al
`
`a.·1.1 Locator Devi.ices
`h i u ful, 1to 1 - :j
`b Jute
`r
`
`-r d vi
`
`·
`
`t
`
`.
`
`. "nd _pend nt har -:teri ·jc :
`· u . u~ .
`referen e· r
`av - a [ ·
`su h mi •
`, a ongm. . . ·,w d-:Vi
`· . respec
`·. d jn
`·gm and :repo ,t nly _h;mges
`a
`tm1 jo
`ic
`· .
`tra kballs and
`b
`· _
`ao aribitradly large
`from th -ir former po i-i.on. A re ati
`change in positi n: A u er ,can move am use al g th_
`up, an pla
`· it back
`p Ii
`. it & ain. A data tab) t can be piogrammed t behave
`·
`· · '
`· m ·•tar; m
`·, · . po ·tion read -· · · · · • ·
`a
`U'a ted from au
`O 11dinates
`'
`, ·
`i
`iti n. Thi ,
`t yi Id _ .
`.
`ded to the ,
`m an · ,
`·process is con inued until
`gain
`· · •
`· ·
`.
`Relative dev·ces can -
`.
`-
`i : ·
`d a i .
`· · ·
`can b . _
`of a. rel
`th~ p ·
`on th - .c11 en .
`repOSiition the u.rsor an
`light pen . r t u h
`. ith a direct device
`reen--th _ _ · po,i nt directly a
`as
`urrogate fing r~ with an indir,e _ t d .· ·
`uch as a abt ,
`reen with a fin
`·-re n.
`oton th
`·
`urs r n .
`menu im _ ad.
`mu t be learned for th- latter;, the pr ' "feraf
`do · an en ironm nt in whth ·
`h · -
`an . d ·
`. However. d' re t pointing can
`a users.
`
`c
`
`e,,.
`
`rea ab olu -,
`pro_ ram can
`
`·
`
`'
`
`the
`
`comput
`casu ·
`arm fat
`
`, r use
`. e pe i
`
`

`

`8.2
`
`Basic Interaction Task.s
`
`365
`
`shift - i
`to move
`selection up
`using keyboard
`
`shift - J.
`to move selection
`down using
`keyboard
`
`font
`get
`Insert
`
`morg1n
`print
`put
`repeot
`
`T to scroll window
`up using keyboard
`
`J. to scroll window
`down using keyboard
`
`Fig. 8 .8 A menu within a scrolling window. The user controls scrolling by selecting the
`up and dow n arrows or by dragging the square in the scroll bar.
`
`to be paged or scrolled through. A scroll bar of the type used in many window managers
`allows all the relevant scrolling and paging commands to be presented in a concise way. A
`'fast keyboard-oriented alternative to pointing at the scrolling commands can also be
`provided; for instance, the arrow keys can be used to scroll the window, and the shift k.ey
`can be combined with the arrow keys to move the selection within the visible window, as
`shown in Fig. 8.8. In the limit, the si.ze of the window can be reduced to a single menu
`item, yielding a "slot-machine" menu of the type shown in Fig. 8.9.
`With a hierarchical menu, the user first selects from the choice set at the top of the
`hierarchy, which causes a second choice set to be available. The process is repeated until a
`leaf node (i.e., ao element of the choice set itseJO of the hierarchy tree is selected. As with
`hierarchical object selection, navigation mechanisms need to be provided so that the user
`can go back up the hierarchy if an incorrect subtree was selected. Visual feedback. to give
`the user some sense of place within the hierarchy is also needed.
`Menu hierarchies can be presented in several ways. Of course, successive levels of the
`hierarchy can replace one another on the display as further choices are made, but this does
`not give the user much sense of position within the hierarchy. The cascadi11g hierarchy, as
`depicted in Fig. 8.10, is more attractive. Enough of each menu must be revealed that the
`complete highlighted selection path is visible, and some means must be used to indicate
`whether a menu item is a leaf node or is the name of a lower-level menu (in the figure, the
`right-pointing arrow fills this role). Another arrangement is to show just the name of each
`
`Current Menu Item g
`( Accep t) (Cancer)
`
`Fig. 8 .9 A small menu-selection w indow . Only one menu item appears at a time. The
`scroll arrow s are used to change the current menu item, which is selected when the
`Accept button is chosen.
`
`

`

`366
`
`Input Devices. Techniques. and Interaction Tasks
`
`(a)
`
`(b)
`
`(C)
`
`Fig. 8 .1 0 A pop-up hierarchical menu. (a) The first menu appears where the cursor is,
`in response to a button-down action. The cursor can be moved up and down to select
`the desired typeface. (b) The cursor Is then moved to the right to bring up the second
`menu. (c) The process is repeated for the third menu.
`
`selection made thus far in traversing down the hierarchy, plus all the selections available at
`the current level.
`A panel hierarchy is another way to depict a hierarchy, as shown in Fig. 8. 11 ; it takes
`up somewhat more room than the cascading hierarchy. lf the hierarchy is not too large, an
`explicit tree showing the entire hierarchy can also be displayed.
`When we design a hierarchical menu, the issue of depth versus breadth is always
`present. Snowberry et al. [SNOW83) found experimentally that selection time and accuracy
`improve when broader menus with fewer levels of selection are used. Similar results are
`reported by Landauer and Nachbar ILAND85) and by other researehers. However, these
`
`Fig. 8 .11 A hierarchical-selection menu. The leftmost column represents the top level;
`the children of the selected item in this column are shown in the next column; and so on.
`If there is no selected Item, then the columns to the right are blank. (Courtesy of NeXT.
`Inc.@ 1989 NeXT, Inc.)
`
`

`

`368
`
`Input Devices, Techniques, and Interaction Tasks
`
`maintaining visual continuity. An attractive feature in pop-up menus is to highlight initially
`the most recently made selection from the choice set if the most recently selected item is
`more likely to be selected a second time than is another item, positioning the menu so the
`cursor is on that item. Alternatively, if the menu is ordered by frequency of use, the most
`frequently used command can be highlighted initially and should also be in the middle (not
`at the top) of the menu, to minimize cursor movements in selecting other items.
`Pop-up and other appearing menus conserve precious screen space-one of the
`user-interface designer's most valuable commodities. Their use is facilitated by a fast
`RasterOp instruction , as discussed in Chapters 2 and 19.
`Pop-up menus often can be context-sensitive. In several window-manager systems, if
`the cursor is in the window banner (the top heading of the window) , commands involving
`window manipulation appear in tbe menu; if the cursor is in the window proper, commands
`concerning the application itself appear (which commands appear can depend on the type of
`object under the cursor); otherwise, commands for creating new windows appear in the
`menu. This context-sensitivity may initially be confusing to the novice, but is powerful once
`understood.
`Unlike pop-up menus, pull-down and pull-out menus are anchored in a menu bar along
`an edge of the screen. The Apple Macintosh, Microsoft Windows , and Microsoft
`Presentation Manager all use pull-down menus. Macintosh menus , shown in Fig 8.13, also
`illustrate accelerator keys and context sensitivity. Pull-out menus , an alternative to
`pull-down menus, are shown in Fig. 8.14. Both types of menus have a two-level hierarchy:
`The menu bar is the first level, and the pull-down or pull-out menu is the second. Pull-down
`and pull-out menus can be activated explicitly or implicitly. Ln explicit activation, a button
`depression, once the cursor is in the menu bar, makes the second-level menu appear; the
`
`Cut
`Copy
`Paste
`Ctear
`
`Duplicate
`Select All
`
`..............
`
`. .. ...... .. .. ..
`eursoi
`
`··············
`
`• • • '
`
`0 . . . . . . . . . .
`
`Release ;
`b11ttn.-
`
`Fig. 8 . 13 A Macintosh pull-down menu. The last menu item is gray rather than black,
`indicating that it is currently not available for selection (the currently selected object, an
`arc, does not have corners to be rounded). The Undo command is also gray, because
`the previously executed command cannot be undone. Abbreviations are accelerator
`keys for power users. (Copyright 1988 Claris Corporation. All rights reserved.)
`
`

`

`II ut D_
`
`s. _ndl ilnte ~c
`
`1i
`
`p k ng po'nts ,on th
`
`8.3.3
`
`ID n---'m'ic Mani
`
`· . In man
`
`itu , · ,ns th u mu: t he
`
`[ci]
`~
`
`mbol ·mo n ~ - · po · ·on.
`
`

`

`s:.3
`
`Co p
`
`ffik::ta11kQl8 i
`
`a
`
`d
`moliing
`with
`wn
`
`-0
`s
`ssible
`
`-Cil
`
`tions
`
`F 1. ,8 41 Dynamjc 1rottnon.
`
`,
`,n --
`th r UD'i
`-
`
`D
`
`F· g. 1 .42 H_ ndl
`
`bj _c
`
`•
`
`

`

`9
`Dialogue
`Design
`
`We have described the fundamental building blocks from which the interface to an
`interactive graphics system is crafted-interaction devices, techniques, and tasks. Let us
`now consider how to assemble these building blocks into a usable and pleasing form.
`User-inrerface design is still at least partly an art, not a science, and thus some of what we
`offer is an attitude toward the design of interactive systems, and some specific dos and
`don'ts that, if applied creatively, can help to focus attention on the lmm(m fac/Ors, also
`called the ergonomics, of an interactive system.
`The key goals in user-interface design are increase in speed of learning, and in speed of
`use, reduction of error rate, encouragement of rapid recall of how to use the interface, and
`increase in attractiveness to potential users and buyers.
`Speed of learning concerns how long a new user takes to achieve a given proficiency
`with a system. It is especially important for systems that are to be used infrequenlly by any
`one individual: Users are generally unwilling to spend hours learning a system that they will
`usc for just minutes a week!
`Speed of use concerns how long an experienced user requires to perform some specific
`task with a system. It is critical when a person is to use a system repeatedly for a significant
`amount of time.
`The error rare measures the number of user errors per interaction. The error rate affects
`both speed of learning and speed of use; if it is easy to make mistakes with the system,
`learning takes longer and speed of use is reduced because the user must correct any
`mistakes. However, error rate must be a separate design objective for applications in which
`even one error is unacceptable-for example, air-traffic control, nuclear-power-plant
`
`391
`
`

`

`392
`
`Dialogue Design
`
`control. and strategic military command and coniJOI systems. Such systems often trade off
`some speed of use for a lower error rate.
`Rapid recall of how to use the system is another distinct design objective, since a user
`may be aw.ty from a system for weeks, and then return for casual or intensive use. The
`system should "come back" quickly to the user.
`A11ractiveness of the interface is a real marketplace concern. Of course, liking a syst.em
`or a feature is not necessarily the same as being facile with it. Ln numerous experiments
`comparing two alternative designs, subjects state a strong preference for one design but
`indeed perform faster with the other.
`It is sometimes said that systems cannot be both easy to learn and fast to use. Although
`there was certainly a time when this was often true. we have learned how to satisfy multiple
`design objectives. The simplest and most common approach to combining speed of use and
`case of learning is to provide a "starter kit" of basic commands that are designed for the
`beginning user, but are on ly a subset of the overall command set. This starter kit is made
`available from menus, to facilitate ease of learning. All the commands, both starter and
`advanced. are available through the keyboard or function keys, to facilitate speed of use.
`Some advanced commands are sometimes put in the menus also, typically at lower levels of
`hierarchy, where they can be accessed by users who do n01 yet know their keyboard
`equivalents.
`We should recognize that speed of learning is a relative term. A system with 10
`commands is faster to learn than is one with I 00 commands, in that users will be able to
`understand what each of the 10 commands does more quickly than they can what 100 do.
`But if the application for which the interface is designed requires rich functionality. the 10
`commands may have to be used in creative and imaginative wo~ys that are difficult to learn.
`whereas the 100 commands may map quite readily onto the needs of the application.
`In the final analysis, meeting even one of these objectives is no mean task. There are
`unfortunately few absolutes in user-interface design. Appropriate choices depend on many
`different factors, including the design objectives, user characteristics, the environment of
`use , available hardware and softwJre resources, and budgets. It is especially important that
`the user-interface designer's ego be submerged, so that the user's needs, not the designer's,
`are the driving factor. There is no room for a designer with quick, off-the-cuff answers.
`Good design requires careful consideration of many issues and patience in testing
`prototypes with real users.
`
`9 .1 THE FORM AND CONTENT OF USER-COMPUTER DIALOGUES
`
`The concept of a ttSu-<omputer dialogue is central to interactive system design, and there
`are helpful analogies between user-oomputer and person-person dialogues. After all,
`people have de