INTIR(W93 24-29 April1993

THE EVOLUTION OF AN INTERFACE FORCHOREOGRAPHERS

Tom W. Calvert, Armin Bruderlin, Sang Mah, Thecla Schiphorst, Chris Welman

School of Computing Science

Simon Fraser University, Burnaby, B.C. Canada V5A 1S6

604-291-4588

e-mail address: torn@ cs.sfu.ca

ABSTRACT

This paper describes the evolution of the interface to LifeForms, a compositional tool for the creation of dance

choreography, and highlights some of the important lessonswe have learned during a six year design andimplementation period. The lessons learned can be groupedinto two categories: 1) Process, and 2) Architecture of theInterface. Our goal in developing a tool for choreographyhas been to provide computer-based creative design supportfor the conception and development of dance. The evolutionwas driven by feedback from the choreographers and userswho were members of the development team, combinedwith our knowledge of current thinking on design andcomposition. Although the interface evolved in a relativelyunconstrained way, the resulting system has many of thefeatures that theoretical discussion in human interfacedesign has projected as necessary. The Life Forms interfacehas evolved incrementally with one major discontinuitywhere adoption of a new compositional primitive required acompletely new version.

The choreography and composition of a &nce is a complexsynthesis task which has much in common with design.Thus, the lessons learned here are applicable to the

development of interfaces to such applications as computeraided design.

KEY WORDS: Composition, design, user interface,dance, complexity, choreography, human animation.

1. INTRODUCTION

Like design, the composition and choreography of dance isa complex synthesis task. The design process containselements that are recognized as common to all creativeactivity. Dance is the most technically complex form ofhuman movement that exists in our culture, sodevelopment of a computer tool to assist in the creation ofdance poses several research challenges. Choreography is a

Permission to copy without fee all or part of this material ia

granted provided that the copies are not made or distributed for

direct commercial edventage, the ACM copyright notice and the

title of the publication and its date appear, and notice is given

that copying is by permission of the Association for Computing

Machinery. To copy otherwise, or to republish, requires a fee

and/or specific permission.

@I1993 ACM 0.8979J-~75.5/93/0~04/OJ 15...$1 .5(3

compositional design task that requires a set of skills telatedto creating, structuring, and forming. Building a computerinterface that interacts with a choreographer’s design skillset requires an understanding of the mental model of thechoreographer’s design process. As Herbert Simon noted,“The ability to communicate across fields--the commonground--comes from the fact that all who use computers incomplex ways are using computers to design or toparticipate in the process of design.’’[l3]

This area of research relates to the observation andunderstanding of how our creative process operates when weinteract with computer systems, and how computer interfacedesigners can help provide a more intuitive, direct, andtransparent relationship with the creative idea [7,9]. Indance, where the creative idea is a movement idea, the goalis to visualize and create body movement in an immediateand responsive way, so the computer tool must become a“visual idea generator.”. This paper summarizes the

evolution of Life Forms, a computer based system tosupport composition and choreography, and discusses thelessons learned during the development (note that during the

early stages of its development the Life Forms software wasknown as Compose). Many of the observations of the

effectiveness of the interface came from users of the system:

choreographers and dancers, as well as members of thedesign team. An example of a user who has had greatimpact on the evolution of the interface is MerceCunningham who has been using the system for three yearsin New York City. Life Forms has been described in moredetail at CHI’90 [12] and elsewhere [4].

In the years before starting this project in 1986, we hadacquired considerable experience in working with dancersand choreographers and in the development of computerbased systems for editing and interpreting dance notation[5,6]. Although systems for editing and interpretingnotation have an important role in recording dance, archivaltools are not necessarily very useful as tools forcomposition. In order to address the goal of supporting thecompositional process we began the development acompletely new system which would directly assist theworking choreographer in creating movement. Merce

Cunningham noted when he began working with LifeForms, “The thing that interested me most, from the verystart was not the memory -- it wasn’t simply notation --

115

24-29 April1993 lNTfRtH1’but the fact that I could make new things” KA Times May15 1991].

The design and development was carried out by aninterdisciplinary team made up of users such aschoreographers, as well as systems architects andimplementers. The process of designing the interface hasbeen highly iterative - beginning from a very simpleconcept many, many alternatives have been suggested andevaluated while only a few have been implemented. Attimes the introduction of a new concept or approach hasrequired a discontinuous design shift to occur in the

otherwise incremental process.

Lije Forms, which has been licensed to Kinetic Effects Inc.(Seattle, WA.), was initially developed on Silicon GraphicsIris workstations and this remains the platform for ourresearch. However, to provide a version which would bemore accessible to working dancers and choreographers wehave also developed a version for the Apple Macintosh (thisis now available from Macromexiia).

2. DANCE COMPOSITION ANDCHOREOGRAPHY

Typically, in composing a new dance, a choreographerstarts from a particular stimulus. The stimulus can be asvaried as a specitlc physical movement, a musical phrase, avisual image, or a state of mind. Even in the case ofmovement or an idea without particular contex~ what maybe called “pure dance”, the choreographer frequentlydevelops thematic material through an exploratorystructuring technique. Or the choreographer may developthe initial movement idea through some event or commentin the environment. If there is a specific context, such as aparticular exploration of space, a striking dream image, or apiece of music or dramatic plot, the choreographer maydraw out or illustrate spatial possibilities, may simplythink or muse about a striking image to make furtherconnections, or may listen to the music or observe thestory and absorb its “sense”, its dynamic qualities, itstempos or in the case of expressive dance its statement ofemotional flow [2]. From all of these stimuli thechoreographer explores or develops the generative idea,either intuitively or constructively using chance ordeterministic structuring procedures, or “interprets” music,image, feeling or narrative, creating or finding movementthat seems to successfully mirror its essence.

In the studio the choreographer works with the dancers tobuild the piece. The dancers may improvise with thechoreographer’s initial compositional material to providemovement. Some choreographers may work with notes,sketches, and floor plans, and some will record work inprogress with a video camera to act as an objective eye, anda memory aid. This is an iterative and interactive processand proceeds over a period of weeks or months until thedance is complete. Some choreographers begin first with

the broad spatial outlines and then go back to develop thedetailed movements, while others will begin with somespecific movement material, then develop phrases, and thensections. Structurally, a choreographer will always need tomove between the design of the overall dance and the design

of the more detailed levels of section, phrase, and particularmovement or gesture.

The computer based system is designed to assist, not to \replace this process. Dance itself is still a kinesthetic

experience and cannot be replaced by technology. But thecomputer based system can provide an extension, a visualidea generator that supports the iterative and interactivenature of the choreographic process. And like video, thetool also provides a record of compositional process whileit has the advantage of storing movement in threedimensional space and allowing editing and modification tooccur.

3. EVOLUTION OF THE INTERFACE

3.1. Concept

The first iteration of the system was conceived as aplanning tool which would allow the choreographer toblock out how individual dancers would move. Thus, sometoken for each dancer would be placed on a movable stageand when the configuration was acceptable, theconfiguration would be saved. The choreographer wouldthen reconfigure the dancers for the next important scene.This would in turn be saved and the process would continueuntil the piece was complete. This is similar to thestoryboard used by film directors and animators to help inplanning a scene or action sequence. By flipping through

these keyframe configurations, a crude animation could beproduced for the composer to review.

From this early concept, the system’s development hasprogressed through two main evolutionary stages. In thefirst, the tokens which served as the basic building blockfor a composition were static body shapes, or stances. Inthe second stage, the basic building block became an entiremovement sequence. A description of the system at thesestages has been given in some detail elsewhere [12, 4]. Asynopsis is be provided here to summarize the system andhighlight recent features before considering the lessonslearned during the development process.

3.2. Composition Based on Stances

In the fwst useful version of the system, key scenes areconstructed by placing figures in particular stances on astylized stage at a specific point in time, The stage itselfcan be tilted and turned to view the scene from any vantage

116

INTERCHI’93 24-29 April1993

Figure 1: a) Old Stance-based Body Editor. Figure 1: b) New Sequence Editor.

point. The stances are selected from a collection of menus

(see right side of Figure la), each of which display anumber of different body shapes to choose from. A typicalmenu, for example, might consist of ballet shapes, orperhaps different lying positions. Once several keyframeconfigurations have been entered, the scene can be playedback. During playback the system performs simpleinterpolation of position, orientation and body shape foreach figure, to give the composer a quick approximation ofhow the scene might appear in a live performance.

The Body Editor allows the choreographer to create newbody shapes and add them to the menus, or to edit existingshapes. In a separate window a simplified threedimensional representation of the human body is presentedfor editing. A limb segment can be selected and itsposition interactively adjusted with a hemispherical sliderconceptually similar to a virtual trackball [8]. When aposture has been fully defiied in this manner, the shape canbe transferred to one of the stance menus. In this way theuser can refine and extend the palette of shapes available.

1:52 1 I 21!3[41S [C.l,lol o

FIm- B

xxal

A third window, the Timeline, provides a summary of thebody stances for each figure as they develop over time,similar to the notion of a musical score (Figure 2). TheTimeline provides information about the temporaldevelopment of the composition which is lacking in theprevious windows, showing both the relative timing of thekey stances assigned to a figure, as well as the temporalrelationship between different figures. Simple editing

operations provide a mechanism by which movement overtime can be copied, cut and pasted in a manner familiar toanybody who has used a word processor.

This simple stance-based system has proved to be a viabletool for choreography and human figure animation, and isthe model implemented in the Macintosh version of theprogram.

3.3. Composition Based on Sequences

As the users gained experience with the earlier system, they

came to realize that a better building block for a dancecomposition was a movement sequence rather than a simplestance. To meet this need, a major redevelopment wasnecessary. Menus of stances were replaced by menus ofmovement sequences, which could be as short as a singlestance or as long as many hundreds of keyframes. Asbefore, composition in space involves making a selectionfrom a menu and placing it on the stage. However, ratherthan dragging and orienting a single stance, now the userworks with a complete movement sequence as a primitive.Although the stage view still displays only a single shape

to represent a sequence, a line indicating the path traveledby the figure during the sequence is also displayed.Multiple sequences can be strung together for a singlefigure, just as multiple stances were before.

Figure 2: The stance-based Macintosh timeline window.To facilitate the creation and editing of sequences, most ofthe functionality of the earlier version was combined toform a new Sequence Editor (Figtue lb), which replaced theBody Editor. In this window a single figure can be

117

24-29 April1993 lNTfRtHr9manipulated into a series of keyframe stances, which maythen be played back as an animation. A timeline stripbelow the figure offers the same editing capabilities as theoriginal Timeline window (see Figure lb centre displaystrip). In addition the timing of portions of the sequencecan be adjusted by stretching or shrinking ranges on thetimeline. A major enhancement has been the addition ofdirect manipulation of the figure with inverse kinenuuics[14], allowing the user to drag a chain of body parts aroundrather than adjusting each part individually (Figure 3).Sequences may also be imported from external sources andadded to menus. Another major addition is the provision ofprocedural movement which is generated by algorithmsrather than frame by frame specification of the

choreographer. An example is the generation of customizedwaiks[3].

Figure 3: Selecting and dragging the arm using inversekinematics - the cursor is on the hand.

With detailed timing within a sequence being adjusted in theSequence Editor, the main TimeIine becomes simpler(Figure 4). The sequences assigned to each dancer aresimply shown as boxes which can only be edited (cut, copy

or paste) as a unit. However, in this mode thechoreographer can very quickly build up complexmovement for each dancer by assembling sequences. Usersfound it helpful if the Timeli~e is displayed on top of astage view so that the multiple dancers can be animated atthe same time that the inter-relationships in time are beingreviewed.

Figure 4: The new Timeline in sequence-based system.

A recent addition to the system is the ability to add digitizedsound to the composition. The choreography can besynchronized with the soundtmck during playback.

This description reflects the current state of the systemwhich runs on the Silicon Graphics family of IRISworkstations.

4. LESSONS LEARNED

The Evolution of the Life Forms interface came aboutthrough many iterations. Each incremental developmentphase was accompanied and even initiated by newunderstanding, based on observations and experience withusing the interface, Often these lessons prompted newideas that resonated strongly enough within the design teamto affect the conception, design and implementation of theinterface. These lessons can be loosely grouped into twocategories. Process refers to the design process andchoreographic process in terms of b it was experienced

and affected by the interface, Architecture of the Interfacerefers to experiences and observations dealing more directlywith W technical interface design issues became focal

points of our work.

4.1. Process

Observation 1: An interdisciplinaryDevelopment Team is Necessary

An interdisciplinary development team involving users,systems architects and implementers is essential. In thiscase the users were choreographers, most of whom had notechnical background, and particular attention was given tomaintaining the level of their involvement in the designprocess. Users who are members of the team often becomeinvolved in the technical developments and lose their

detachment. Thus some limited turnover in teammembership can be helpful.

Observation 2: The Roie of Users

Involvement of users who are part of the development team,

although very useful, is no substitute for external userswho can bring a perhaps technologically naive yetcreatively insightful point of view to the process. At ailstages the system under development must be tested byappropriate (i.e. naive and expert) users.

There is a special role for a “Mentor User” - someone whois a true leader in the field and who can pioneer innovation.In our case New York choreographer, Merce Cunninghamrecipient of the MacArthur fellowship amongst many otherchoreographic honors and awards has served as a MentorUser for three years. Cunningham has spent almost 50years challenging assumptions and traditional notions ofwhat dance is and how it can be made, and continues to do

118

INT{R(W93 24-29 April1993

so in his use of Life Forms. Cunningham balances thecomputer’s precise representation with the realities ofhuman physicaiity, “I look at some things and say, “wellthat’s impossible for a dancer to do. But if I look longenough I could think of a way it could be done. Notexactly as its done on the screen, but it could prompt myeye to see something I’ve never thought of before.” [MerceCunningham, quoted in the LA Times, May 15, 1991]

Observation 3: Discontinuities in DevelopmentAre Sometimes Necessary

As experience was gained with the original stance-basedsystem and as hardware capability increased, it becameobvious that a higher level building block for dancecomposition was necessary. The implementation of asequence as the design building block (rather than a stance)enabled choreographers to develop design “chunks”. Thisallowed composition to occur on more than one conceptuallevel. A sequence can be created, manipulated, varied andplaced with other sequences at a rate that enables the visualresponse to provide a meaningful creative feedback loop.Major design re-implementation was required to meet thisneed, and correspondingly a major break in the developmentprocess was necessary. Merce Cunningham noted in

response to this development in Life Forms: “What was

like photographs is now like film, and what started out aswork with positions has developed into work with phrases... It’s remarkable, they keep adding things to it ... it willenlarge it [dance]. The system now has multiplepossibilities”.

Sometimes a new concept requires a major re-implementation of a system. Discontinuities of this type

should be minimized, but from time to time are necessary,especially when they provide a major conceptualbreakthrough to occur.

Observation 4: Time from Conception toimplementation

Design is an iterative process, and for this project, at least

the process of designing the way the program functions andhow that functionality is presented to the user has been noexception. The existing system is a result of manyiterations, based largely on comments and suggestions fromthe users within the development team. Unfortunately therealities of software development mean that the turnaroundtime from idea to implementation is often measured inweeks (or more). In our experience, this not only inhibitsthe whole design process, but also may tend to makedevelopers resistant to good suggestions that make priorwork obsolete. A strict adherence to standard modularcoding practices has been beneficial, limiting the scope ofsource code changes required to support new features. Aisothe adoption of a GUI toolkit and interactive interfacedesigner has been a great help for rapid prototyping andmodifications to the interface. The emergence of morecomprehensive toolkits for interactive 3D applications,

such as SGI’S Inventor [11], also promises to decrease thetime horn idea to realization.

Observation 5: Impact ofChoreographers Work

Often, choreographers whoreported that it had changed

Life Forms on the Way

learned to use the systemthe way they thowzht about

choreography and how they went about composing. This isinteresting, but indicates the difficulty in objectively

evaluating a tool, which changes the user’s conception ofthe task. At the same time it illustrates that the ideas andwork of ail members of the development team have a directeffect on one another, and that artists are willing to risk, toexperiment and to ask questions about new ways ofworking. Cunningham, always willing to expand hisexperience of looking at and creating movement, has said,“The common thread in all these dances [made with LifeForms] is that they are ail different. That’s what interestsme. I am not interested in the idea of repeating something.What’s exciting is when I come across an idea that I’munfamiliar with, or when I have a question for myself, ....if its something that interests you deeply, you will find away.” [Anchorage Daily News Feb. 23 1992]

Among comments from other choreographers who used LfeForms to create movement were the following: “I was ableto work more directly with internal imagery. It freed mefrom my own personal movement bias. I was able to havedancers on the screen take on movements that I wouldn’thave dreamed of attempting on my own”; or, “I was muchmore likely to use movement outside my comfort range and

vocabulary and venture into the realm of the unknown.”;and, “Working with the system inspired me in new wayssimply because of the visual nature of working with it.”,and “I found my choreographic process with Life Forms tobe more intuitive, more ‘out of body’, not hampered by myown limitations or concerns for dancers well-being, noreven concerned with the outcome necessarily.” [YorkUniversity Choreographic Workshop, Summer 19921

Much of the creative process reflected in the previousresponses relate to the visual nature of the interface, and thecreative process that is evoked by working withchoreography in a visual way. Choreography is also veryoften a kinesthetic experience and it is well known thatdifferent choreographers have preferred methods of working.One choreographer in particular found that, “I discoveredhow little I rely on visual awareness in the studio, that Itend to work with a kinesthetic awareness of the body, andof movement coming from within the body. I initiallyfound it quite difficult to make the transition fromrecognizing the computer image on the screen and seeing iton the body in the studio”

These comments reflect both the multiplicity of methods inwhich choreographers conceive and create dance, and also anunexpected willingness to work with what exists, evenwhen it maybe different from what has been leaned before.

119

24-29 April1993 INFRM’93Again, perhaps Cunningham says it best when he stated,“My point in working with Life Forms is not to complainabout what it can’t do, but to look and see what it can do.L~e Forms can enlarge possibilities and help the way thatpeople look at dance.” &A Daily News, May 9 1991]

4.2. Architecture of the Interface:

Observation 6: Reduce Task Complexity withAlternate Views

Early in the development process it was realized that the

original concept of letting the choreographer do alldevelopment on the Stage (as in a real studio) was limiting.The only way to review how the piece developed in timewas to play it through as an animation and although theanimation is useful, just as in reviewing a musicalcomposition, it is often possible to obtain insights fromthe score which are not obvious in a performance. Thus, itwas decided to add a display which set out a representationof the piece as it develops in time.

Another simple addition to the original system was awindow in which the body stance of individual dancerscould be built up, edited and customized. After considerableexperimentation we developed a Body Editor similar to thatshown in Figure la) which was later extended to theSequence Editor shown in Figure lb).

These observations indicate that task complexity can bereduced by providing alternative ways to view the

development of the piece. This is consistent with otherexperience in complex problem solving [13].

Observation 7: Separating High LevelSpecification from Detailed Choreography

detail and higher level actions, so these changes must bemade as easy as possible.

Observation 8: Access to Explicit Knowledge

It is obvious that those involved in creative synthesis tasksdo not start from scratch. They bring to each newcomposition the history of their own previous work, aswell as that of others. They also have a background in thestandard concepts of their field. The menus in our systemare visual archives for these types of knowledge. Browsingis supported by allowing the user to click on a menu itemto reveal a flip-book animation of the stored sequence.

In this class of tasks, the most powerful access toknowledge is through the memory of the artist. But thisshould be supplemented where possible with readilyaccessible files of previous work and with proceduralgenerators.

Observation 9: Procedural Generation ofMovement Sequences

While the sequence-based system empowers the user, themovement sequences still have to be generated somehow;building up sequences frame by frame in the SequenceEditor is extremely tedious. A more powerful approachinvolves finding a procedural method to generate a wholeclass of movements such as locomotion. Bruderlin [3] hasdeveloped a procedural approach to walking - this cangenerate a wide variety of walks which can be used inbuilding sequences. The interface shown in Figure 5 allowsthe user to customize the walk by changing parameters(e.g.. velocity, bounciness, torso sway) to characterize theperson being animated - an older person, a child, a tiredperson, etc.

When the originrd stance-based system was replaced by asequence-based system, the power of the user to createdances was greatly increased. This change allowed the userto work with a higher level primitive. Of course, it wasstill necessary to create these sequences on a frame by framebasis, unless a procedural method was used.

As noted, when a body editor was added to the originalsystem, it made it possible to separate the detailedchoreography of individual figures from the more generalchoreography of multiple interacting figures. In the sameway, when the Sequence Editor was added to the sequencebased version of the system, it made it possible to edit thetiming of the moves of an individual figure separately fromthe more general timing in the Timeline of the sequencesassigned to all dancers.

These observations confirm that the process is hierarchicaland that the composer is empowered by separating detailfrom higher level actions. However, we have observed thatusers frequently switch back and forth between low level

Figure 5: Interface for real-time, procedural walkingmodule.

The knowledge of movement available through a proceduralgenerator greatly increases the power of the user providingthe procedures can be customized. This tool can replace apart of the choreographic process by automaticallygenerating a feasible movement pattern. But it need not take

120

lNTkRcHr93anything away from the creative choreographer, since thesequence generated can be modified for specific situations.

Observation 10: Simple Abstract Models may beas Useful as more Realistic Models

As shown in Figure 6, a variety of body models have been

used - they include a simple stick figure, an outline figure,a figure defined by contours and a shaded solid figure. Inearly implementations the simplest figures were usedbecause they could be drawn most quickly. With fasterprocessors it has become possible to use more complex andrealistic figures. However, users often find that the simplestfigures are effective for their needs - research in psychologyhas shown that we can perceive human movement quiteaccurately from only a few points on the body [10]. Theshaded figures certainly add realism - but it is not clearwhether users prefer this to an abstraction. Also, users aremore likely to be critical of imperfections in more realisticfigures while they accept abstractions for what they are.

Body models which are simple abstractions are often quiteacceptable and may be preferred over more realistic shadedmodels.

24-29 April1993

kinematic chain. This is intuitively attractive and users likethe facility, but our experiences have shown that it maybe

difficult to accurately duplicate spxific body stances.

Users like direct manipulation, but it does not always givethe best results.

5. DISCUSSION AND CONCLUSIONS

The composition and choreography of dance is a complexsynthesis and design task. Many computer based tools havebeen developed to assist designers but very few of thesesystems give users sufficient freedom to experiment and becreative. Our goal in developing a tool for choreographywas to use the computer to provide creative design support

during the process of conceiving of and developing a dance.Design evolution was often initiated by feedback from thechoreographers and users who were members of the designteam, combined with our knowledge of current thinking ofdesign and composition [1, 13]. We started with a verysimple concept (blocking figures on a stage) and over a six

year period have evolved a very sophisticated system.

Figure 6: Four different body models: stick, outline,contour and “fleshed out” figures.

Observation 11: Users Prafer Direct Interaction

In the original Body Editor, the user had the choice ofadjusting limb orientation either by directly manipulating iton the screen with the mouse or by using linear or sphericalpotentiometers on the screen. Direct manipulation wasgenerally preferred, but sometimes it was difficult to seejust how the 3-d body was being moved on the 2-d display.Inverse kinematics allows the user to move the endpoints oflimbs without first adjusting other segments in the

The principal conclusion that can be drawn is that thismodel for the development process has been successful andthat a very useful system has evolved. The system evolvedincrementally with one major discontinuity where adoptionof a new compositional primitive required a completely newversion. Another conclusion is that although the interfaceevolution was driven in a relatively unconstrained way bythe choreographers in the design team, the resulting systemhas many of the features that theoretical discussion hasprojected as necessary [12,13]. Specitlcally

121

24-29 April1993 lNTfRtH1’

● The process is hierarchical and tools should suppcmt highlevel (more conceptual) and low level (more detailed)spec~lcation.

. The complexity of the task can be minimized by makingalternate views available (e.g. Stage, Sequence Editor,Timeline).

“ The composer is empowered by access to knowledge. In

general, the internal knowledge of the user can besupplemented with (a) explicit knowledge in the form ofstances, sequences, etc stored in libraries and displayed inmenus; (b) procedural knowledge where algorithms havebeen developed to capture the essence of commonmovements (procedural walking is discussed above} and (c)declarative lmowledge, where rules and constraints are storedin a knowledge base and are available to an expert systemfor resolution of complex interactions, The currentinterface provides access to explicit and proceduralknowledge and users have found this to be valuable.Methods to incorporate declarative knowledge still have tobe developed.

Other observations are that users prefer direct interaction inspecifying body stances although they may sacrificespecificity of exact joint location in space. Users preferinteractivity to accuracy. Users also prefer abstractions inobjects such as body models over realism when realismaffects real-time display.

One of the more startling observations we have made inrelation to working with choreographers is that not onlyhas the interface evolved in response to user feedback, butthat choreographers method of creating has evolved orenlarged as a result of working with a computer-basedcreative tool. Our conclusion is that the design process issymbiotic, and that both the design of the interface and thedesign of the task for which the interface is created effectone another deeply.

6. REFERENCES

1. Akin, O. How Do Architects Design?, In Al and

Pattern Recognition in Computer Aided Design, ed.

Latombe, IFIP, North-Holland, 1978, pp. 65-103.

2, Brightman P. Making dances with algorithms:Towards a theory of choreography, based on the useof computer programs and Laban concepts. M.A.Thesis, Columbia University, 1984.

3.

4.

5.

6.

7.

8,

9.

10.

11.

12.

13.

14.

Bruderlin, A. and Calvert, T.W. Goal-Directed,Dynamic Animation of Human Walking. Computer

Graphics (SIGGRAPH 89), vol. 23, (1989), pp.233-242,

Calvert, T.W., Welman, C., Gaudet, S., Schiphorst,T. and Lee,C. Composition of Multiple FigureSequences for Dance and Animation. The VisualComputer, vol. 7, 1991, pp. 114-121.

Calvert, T.W. Towards a Language for HumanMovement. Computers and the Humanities, 20:2,(1986), pp. 35-43.

Calvert, T.W., Chapman, J,, and Patla, A. Aspectsof the Kinematic Simulation of Human Movement.IEEE Computer Graphics and Applications, vol. 2,(November 1982), pp. 41-50.

Card, S., Moran, T., and Newell, A. ThePsychology of Human-Computer Interaction.Lawrence Erlbaum Asscc., Hillsdale, N.J., 1983.

Chen, M., Mountford, J. and Sellen, A. “A Studyin Interactive 3D Rotation Using 2D ControlDevices”. Computer Graphics 22,4, 1988, pp.121-129.

Hartson, H. R., and Hix, D. Human-ComputerInterface Development: Concepts and Systems.ACM Computing Surveys 21, 1, (1989), pp. 5-92.

Johansson, G. Perception and Psychology, 14, pp.201-211, 1973.

Strauss, P., and Carey, R. An Object-Oriented 3DGraphics Toolkit. Computer Graphics (SIGGRAPH92), 26,2, (July 1992), pp. 341-349.

Schiphorst, T., Calvert, T., Lee, C., Welman, C.and Gaudet, S, Tools for Interaction with theCreative Process of Composition, Proc. CHI’90,(Seattle, April 1990), pp. 167-174.

Simon, H.A. The Sciences of the Artificial. MITPress, Cambridge, MA., 1969.

Welman, C. “Inverse Kinematics ThroughDifferential Manipulation”. In Proceedings of theWestern Computer Graphics Symposium , (1992),pp. 123-128.

122