Holds in the computer animation of New Vogue Dances
 Don Herbison-Evans

 3rd International Conference Computer Graphics, Imaging and Visualization
 University of Technology Sydney, Australia, 26 - 28 July 2006
 New Vogue dances are the most popular dance style in Australian Dancesport competitions. They are dances with the same prescribed choreography for each couple: usually a man and a woman. A feature of these dances is the variety of holds used between partners. Many of the dances have been published as scores in the internationally recognised Labanotation, and the files for these scores published in machine-readable form. This article describes an interactive computer program for a PC which interprets such a file producing animation of two figures dancing together.
 Previous work has described the production of such animation from descriptions in an animation language (NUDES), and of the translation of Labanotation scores into movements of a single figure (LINTER).
 The computer program described in this article (LINTEL) combines these to produce animation of two figures dancing together from the Labanotation files. The main barrier to producing such a program has been the generation and maintenance of the holds between the two partners, given the difference in height (and leg length) of the partners, and the high heels worn by the woman, which make for different step lengths of the two partners. In particular: the techniques used to infer the holds between the partners from the notation, the simulation and maintenance the holds as the dancers move, and the display of changes in the holds in a natural fashion are described.
 The interactive computer program LINTEL also allows zoom in and out, rotation about the 3 axes, and shifting in 3D, while the figures are dancing, as well as variable speed and single frame forward and back animation.
 New Vogue dances originated in the 1930s and '40s, when some Australian dancers rebelled against the formal balletic foot work of the English Old Time dances (Boyd, 1984b), and started to choreograph sequence dances based on the Modern Ballroom technique. Len Hourigan of Brisbane coined the term "New Vogue" for these dances. They have many open positions, which makes them attractive to watch, like the English Old Time, and unlike Standard Ballroom dancing in which observers see only the backs of the couples. The dances also have only the footwork, alignments and basic holds prescribed, leaving scope for the dancers to add their own shaping and styling, which makes them very expressive to dance and to watch.

 Canadian Three Step, bar 5, beat 1                                                Canadian Three Step, bar 5, beat 1
                (author with partner Anna Piper)                                         (Fred Fortran with partner Ginger Gigabyte, ) 

 New Vogue Dancing is now very popular in Australia, being danced at social dances in clubs and public halls around the country. Dancesport competitions and championships held around the country, there are usually more entries in the New Vogue events than in either the Standard Modern Ballroom or the Latin & American events, and this makes Australian competitions somewhat different from those overseas, such as those of North America or Europe.
 The Australian New Vogue dances are sequence dances for couples, each couple consisting usually of a man and a woman. In sequence dances, every couple on the dance floor performs the same steps at the same time, and at the end of the sequence, the steps are started again. This makes New Vogue dances relatively easy to learn, as a beginner can easily copy the movements of adjacent dancers on the floor. They typically have 8, 16 or 32 bar sequences, and so need music with a similar musical phrasing. New Vogue dances have been choreographed to all the dance rhythms, including Viennese Waltz, Modern Waltz, Slow Foxtrot, Quickstep, Tango, March, Bossa Nova, Samba, Rumba, Cha Cha, and Jive (Swing). Over the years, many hundreds of New Vogue dances have been choreographed Many are presented at regular competitions which are held to provide showcases for such new choreography. However, only a limited number have actually gained wide popularity. Over the years, a number geographical variants of many of the dances evolved, and so in 1967 the Australian Dancing Board of Control started standardising a subset of the dances for competitions and dancing championships (Boyd, 1984a).
 Many methods have been proposed for prescribing dances on paper, but one attaining international usage amongst amateur and professional dancers is Labanotation, which was invented by Rudolf Laban in the 1920's (Hutchinson, 1954). This is based a vertical staves, one set for each dancer, each consisting of three lines. Symbols alongside the central line represent ground support, which in New Vogue dancing indicate movements and postions of the feet and legs. Movements and positions of the arms and other parts of the body are represented by symbols on either side of the outer staff lines. Basic symbols represent positions and movements in 8 horizontal directions Three levels of height are indicated by the shading of the symbol. Other symbols are used as modifiers to show to what part of the body each basic symbol applies. Bow lines are used to indicate physical contact between parts of the body and between dancers. The notation can be used to describe any human movement to an arbitrary degree of precision. The lexicon of the full notation has over 1400 symbols.
       LED in operation, showing the start of the  Canadian Three Step

 A simple editor, LED, has been written (Hunt et al, 1979) to write scores involving a basic subset of just over 100 symbols of Labanotation. This has been used to notate the New Vogue dances commonly used Australian Dancesport competions. The data files produced by this editor are simple: they are ASCII files with one line per Labanotation symbol. Each line records the type of symbol, its direction, its x and y coordinates on the score, its height and width, and its shading.
 This is a program that generates animation of one or two animated humanoid figures doing the movements prescribed in a Labanotation score generated using LED. The notation is first translated into a script file for the NUDES animation system, (Herbison-Evans, 1979) and this is then reread and animated using OpenGl. The program allows the user to zoom in or out, pan in three directions, rotate in three dimensions, freeze, advance single frames forward and back, and speed or slow up the movement. This latter is done by simply changing the number of polygonal facets used to approximate each ellipsoid.
 Locomotive steps in each of the eight directions of the basic Laban symbols are generated by separate routines in LINTEL. These generate the stylised movements used in ballroom dancing, which keep the torso over the moving foot, which in turn inhibit partners from treading on each others' feet. Also the step is defined from an open position of the feet to the next open position, in-keeping with the ballroom notion of 'foot pressure' during a step.
 Turns also need attention. In ballroom dancing, a turn is usually only made on the standing foot. Turning on the moving foot is physically difficult.
 A variety of holds are used in New Vogue dancing. The ones so far implemented in LINTEL are the Closed Ballroom Promenade, Counter Promenade, Open, Counter Open, Shadow and Semi Shadow Holds.
 The problem of animating the holds between the partners can be broken down into four sections:
 1. Working out and storing the angles required for the arms and body for the approximate maintenance of each hold.
 2. Detecting from the notation when a hold changes.
 3. Implementing a change of hold smoothly when a change is detected.
 4. Maintaining a current hold as the figures dance.
 In this hold, the partners face each other directly. The hold requires the maintenance of five points of contact between the partners while they are dancing. These consist of three hand contacts:
       1. the man's left hand holding the lady's right hand,
       2. the lady's left hand resting on the top of the man's right upper arm (behind the arm in the Tango),
       3. the man's right hand placed on the left shoulder blade on the back of the lady.
 In addition to these 3 hand contacts, there are two more areas of contact:
       4. the lady's left elbow rests on the man's right elbow,
       5. the right area of the chest of each partner touches that of the other.
 However with the approximate nature of the notation used for these examples, these latter two contacts were not notated, being subsumed under a prior knowledge of the dance style.
 Two further aspects of this hold are that the partners hold their heads turned to the left, and they tilt their torsos to the left. Again these were not notated, but were assumed from the dance style knowledge.
 This hold has the same five contacts as the Closed Hold, but the partners face each other obliquely, both prepared to travel to man's left. The heads are both turned to the man's left, but there is no tilt of the torso.
 This hold also has the same five contacts as the Closed Hold, but the partners face each other obliquely, both prepared to travel to man's right. Both heads are both turned to the man's right, but there is no tilt of the torso.
 This is the simplest hold, with the man's right hand holding the lady's left hand. There is no other contact. The heads look straight ahead, and there is no body tilt.
 This is also simple. The man's left hand holds the lady's right hand. There is no other contact. The heads look straight ahead, and there is no body tilt.
 In this hold, both partners face the same way, with the lady in front and somewhat to the man's right. The man's left hand holds the lady's left hand, and the man's right hand holds the lady's right hand. The heads look straight ahead, and there is no body tilt.
 This is a more complex hold. Again both partners face the same way, with the lady in front and somewhat to the man's right. The man's left hand holds lady's left hand, the man's right hand is placed on the front of the lady's right hip. The heads look straight ahead, and there is no body tilt.
 This is notated by a release sign, and this sign can be used to indicate the end of one hold and preparation for the adoption of another.
 NUDES subroutines were written to bend the approriate joints by suitable angles using the FLEX, ABDUCT, and ROTATE commands of the NUDES language. The resulting quaternion angles are then stored in NUDES variables for used by subsequent BENDTO commands when a hold has to be adopted. The FLEX, ABDUCT, and ROTATE commands are much easier to modify when a position needs editing, than direct use of the quaternion angles used by BENDTO. The routines are called between frames 0 and 1, and the figures then returned to 'No Hold' positions before the animation starts.
 All the holds involve contact between the hands of the partners, so a contact bow with one end on a hand signifies that a hold needs to be identified.
 If a left to left hand contact is found, there is ambiguity between Shadow and Semi Shadow Holds. Sympathetic notation can show this contact in the notation slightly after the the contact of the man's right hand is notated.
 Similarly, if the hold is between the man's left hand and the lady's right hand, there is an ambiguity between Closed, Promenade, Counter Promenade, and Open Holds that needs resolution. This can be resolved more easily by having notation that is sympathetic to this problem: by notating this hand to hand contact slightly after the notation of the other parts of the hold. The ambiguity between these holds can then be resolved using the relative facings of the figures when one of the other contacts is discovered.
 To identify a hold in LINTEL, each hold is assigned a counter. When a release sign is encountered, all counters are zeroed. Then as indicators of each hold are discovered, the appropriate counters are incremented until one of them reaches a threshold value for that hold.
 The time taken for a new hold to be adopted in the animation in LINTEL is somewhat arbitrarily chosen to be about two musical beats. A more sophisticated choice might be to use the time from the start of the first symbol indicating the hold to the end of the last one used.
 The movements are performed using a natural quadratic acceleration-deceleration cycle, using quaternions which give a direct move from the current orientation to the new one.
 This sometimes results in parts of one figure passing through the other. These occasions are also inclined to occur in reality, and the dancers have to make particular movements to avoid hitting each other. This is particularly important at the changes from Closed Hold to Shadow or Semi Shadow Hold, when the lady usually throws her left arm straight up to avoid hitting the man's head in her turn. This avoidance movement needs to be included in the notation, to avoid embarrassment in the animation.
 With the difference in leg length of the partners, the convention is that the lady has to accommodate the stride length of the man. She does this in reality by using articulation angles for her legs that are sympathetic to the man's movement.
 In LINTEL, the spacing between partners is again maintained by modifying the lady's movement, but in a simpler but less physically possible way, by sliding her along the floor to maintain an inter-pelvis spacing preset for each of the holds.
 The hand-hand contact is then maintained by raising or lowering the man's arms to create contact, using Buckdale's algorithm (Buckdale, 1983).
 This repositioning of the lady and the movement of the man's arms are done after performing all the other prescribed movements which occur between each frame of the animation.

 The LINTEL program and the LED labanotation files have been made freely available on the web, so allowing anyone in any part of the world to see, appreciate, and learn the significant dances that are so much a part of the Australian heritage.

 Many thanks are due to the late Phillipa Cullen who inspired this this work, to generations of students who each solved many pieces of the puzzle involved in this project, to the tolerant and kind staff at the many Universities who have hosted the development of this work on their computers, most recently to Kevin Suffern and Stephen Gowing at the University of Technology, Sydney, for their support and encouragement, and to Ron Chambers of Conroe Texas who designed and implemented the interactive environment and animation output, and who led the way to finally putting music to the animation to make it an artistic experience.
 Boyd, N. (1984a)
 New Vogue Sequence Dancing, 165 Bobbin Head Road, Turramurra, NSW, Australia, Revised Edition.
 Boyd, N. (1984b)
 English Old Time Sequence Dancing Guide, 165 Bobbin Head Road, Turramurra, NSW, Australia.
 Buckdale, R.S. (1983)
 "Normal Feet for Dancing Ellipsoids", Proceedings Graphics83 (ed.: H. Hvistendahl), Sydney, pp. 71-76.
 Herbison-Evans, D. (1979)
 "A Human Movement Language for Computer Animation", Language Design and Programming Methodology (ed.: J. Tobias), Springer-Verlag, pp. 117-128.
 Hesketh, R. (1989)
 Revised Technique of the Thirteen New Vogue Championship Dances, Clayton Dance Centre, 296 Spring Road, Dingley, Victoria 3172, Australia.
 Hunt, F.E.S., Politis, G., and Herbison-Evans, D. (1989)
 An Interactive Graphical Editor for Labanotation, Basser Department of Computer Science, Technical Report 343, University of Sydney.
 Hutchinson, A. (1954)
 Labanotation: The System for Recording Movement, Theatre Arts Books, New York.
written 20 May 2006, updated 23 July 2011