Generative Dance – A Taxonomy

Summary

This research tries to contribute to an understanding of the specific role that Generative Art currently plays in dance. It does so by proposing a taxonomy of topics that cover both common and dance specific aspects of Generative Art. This taxonomy is used for comparing a wide diversity of generative works that have been created in the context of dance. m Wegner, teacher and researcher at Offenburg University, Germany. A detailed description of the research has been published here.

Taxonomy

Taxonomy of Generative Dance

The chosen taxonomy identifies six main topics, each of which
is divided into several subtopics.

Topic – Domain

The topic Domain refers to the domain of application of a generative system. Subtopics are:

  • Creation: systems support dance creation
  • Performance: systems employed during performance
  • Exhibition: systems created for exhibitions
  • Teaching: systems for dance education
  • Dissemination: systems created for explanatory purposes
  • Research: systems created for research

Topic – Contribution

The topic Contribution deals with the form of involvement of dancers in the realization of a generative system. Subtopics are:

  • Performer: dancers participate as performers on stage
  • Data: dancers contribute to data acquisition
  • Survey: dancers evaluate a system
  • Consultation: dancers are consulted as experts
  • Participation: dancers directly contribute to development
  • Collaboration: dancers are members of the project team

Topic – Manifestation

The topic Manifestation specifies how a generative system is made perceivable. Subtopics are:

  • Human: humans play the role of a generative system
  • Score: systems create scores for humans
  • Artificial Dancer: systems appear as artificial dancers
  • Media: systems generate synthetic media

Topic – Autonomy

The topic Autonomy addresses the level of agency of a generative system. Subtopics are:

  • Control: systems respond to discrete inputs with discrete outputs
  • Instrument: systems respond to continuous input with continuous output
  • Support: systems support human creativity
  • Collaboration: systems contribute to decision making
  • Autonomous: systems make decisions of their own

Topic – Representation

The topic Representation deals with the abstractions a generative system operates on. Subtopics are:

  • Pose: systems manipulate dance poses
  • Motion: systems manipulate dance motions
  • Behavior: systems specify behaviors
  • Cognition: systems model mental processes,
  • Group: systems specify group behaviors
  • Structure: systems deal with choreographic structure

Topic – Process

The topic Process refers to the operational principle of a generative system. Subtopics are:

  • Random: systems mainly use randomness
  • Rules: systems employ computational rules
  • Simulation: systems employ computer simulations
  • Evolution: systems adapt through artificial evolution
  • Machine-Learning: systems adapt through machine-learning

Examples

Domain / Creation : Cochoreo

© Carlson et al. 2016

Cochoreo is a choreographic support tool that generates dance poses. Poses are created by a Genetic Algorithm that employs a fitness function which favours unusual poses, such as poses with asymmetry, uneven reach space, and instable balance. These poses are meant to serve as seed material in a choreographic process and encourage experimentation with new movement material. More information is available here.

Domain / Dissemination : Counterpoint Tool

© synchronousobjects.osu.edu

The CounterPoint Tool is one of several interactive tools and visualisations that have been created to explain the counterpoint technique employed by choreographer William Forsythe. The CounterPoint Tool focuses on the alignment principle which it imitates through an algorithm that directs the motions of graphical widgets. Depending on the level of alignment, the widgets rotate and move across the screen in unison or not. More information is available here.

Domain / Research : Dancing to Music

© Lee et al. 2019

“Dancing to Music” employs machine-learning to generate dance movements from music. The system subdivides pose sequences and audio features into temporally normalised units. Variational autoencoders are then used to translate from audio units to movement units. More information is available here.

Contribution / Survey : Web3D Dance Composer

© Soga 2002

This “Web3D Dance Composer” is used to generate utilitarian choreographies for teaching the petit allegro to beginning level female dance students. The system employs rules to generate step selections, repetitions, and transitions that fulfil the requirements for physical exertion and memorability. The suitability of this system for dance education was assessed by dance teachers in a survey. More information is available here.

Contribution / Participation : Tonight We Improvise!

© Jochum and Derks 2019

This setup for “Tonight We Improvise!” employs a repurposed motorised wheel-chair base as a robotic dance partner with which human dancers can improvise. The robot tracks a dancer’s position and responds to it via one of several response behaviours. These response behaviours have been developed through several iterative sessions in which dancers participated. More information is available here.

Contribution / Collaboration : Performative Body Mapping

© Saunders and Gemeinboeck 2018

The methodology “Performative Body Mapping” translates the embodied skills from human movement experts to non-anthropomorphic robots. At the beginning of the transfer process, dancers embody the robot by wearing a costume in the shape of the robot. The costume’s movements are recorded through motion capture and then later used to train the real robot through imitation learning. More information is available here.

Manifestation / Human : Adaptive/Responsive Movement Approach

© Pitcher 2015

The Adaptive/Responsive Movement Approach (or ARMA for short) provides a shared language for new media and dance that is inspired by systems theory, programming languages, and directed improvisation techniques. The language establishes interaction rules for dancers according to which they have to respond to various trigger conditions such as the proximity of audience members or the loudness of music. More information is available here.

Manifestation / Score : Lansdown Choreographic System I

© Lansdown 1978

A historical example is the work by Lansdown on generative choreography. For his first choreography, he employed a state machine to handle transition probabilities between elementary dance movements. Using this state machine, a choreographic score in Benesh Notation was created and presented to professional dancers to perform. More information is available here.

Manifestation / Artificial Dancer : AI am here

© Berman and James 2018

The dance piece “AI am here” stages a duet between a human and artificial dancer, the latter of which learns in real-time to imitate the movements of the human dancer. As the learning progresses, the visualisation of the artificial dancer changes from abstract to human-like. More information is available here.

Autonomy / Instrument : Encoded

© Johnston 2013

The dance piece “Encoded” employs a real-time fluid dynamics simulation to generate synthetic audio and visuals. Dancers interact with the simulated fluid by stirring it with their movements. This approach shows that physical models can offer intuitive and rich forms of interaction. More information is available here.

Autonomy / Collaboration : Sound Choreography <> Body Code

© Sicchio and McLean 2017

At the core of the dance and music piece “Sound Choreography <> Body Code” are two live coding languages, one that specifies a dance score and one that controls a sound synthesis system, and two human performers, one dancer and one musician. The codes and the performers engage in a complex collaborative setting, in that the code controlling sound synthesis changes in response to the dancer’s position on stage, and the code controlling the score changes in response to the frequency content of the music. More information is available here.

Autonomy / Autonomous : Becoming

© Wayne McGregor 2017

In the installation “Becoming”, a simulated abstract body plays the role of a fully autonomous artificial dancer to which human dancers can relate to through kinaesthetic empathy. The artificial dancer has been employed in rehearsal settings of the Wayne McGregor dance company. In these settings, both the artificial and human dancers are each tasked with solving individual choreographic problems through body movements. More information is available here.

Representation / Behaviour : Dance Verbs

© Hsieh and Luciani 2005

Hsieh and Luciani developed a choreographic system that models dance movements by means of a physics simulation. For this purpose, a dancing body is deconstructed into a minimal set of essential interactions between masses. The dance movements are reconstructed by modelling the initiation, progagation, and dissipation of energy among the masses. More information is available here.

Representation / Cognition : Robodanza

© Infantino et al. 2016

In the dance piece “Robodanza, a single humanoid robot learns to improvise to music. The robot possesses a cognitive architecture that represents four motivational states: effectiveness, confidence, social acceptance, and basic needs. These states affect the robot’s capability to learn. More information is available here.

Representation / Structure : DANCING

© Nakazawa and Paezold-Ruehl 2009

“DANCING” is a proof of concept choreographic system for generating ballroom dance for a single couple. The system employs a genetic algorithm that operates on step and position-based representations of dance. During evolution, the system rates the quality of each sequence based on the use of the stage, the facing of the dancers, and the step distribution. More information is available here.

Process / Simulation : Vishnu’s dance of life and death

© Antunes and Leymarie 2012

The work “Vishnu’s dance of life and death” employs a computational ecosystem in which the agents’ needs for metabolism and procreation determine their choice of actions. This ecosystem is translated into a dance performance, with agents born and dying entering and leaving the stage, and actions played back as dance movements. More information is available here.

Process / Evolution : Performative Ecologies

© Glynn 2008

The installation “Performative Ecologies” hosts several robots that hang from the ceiling and perform rotational movements. The robots can learn new movements from visitors and share learned movements among themselves. The fitness value of a movement is based on the amount of visitor attention the movement receives. More information is available here.

Process / Machine-Learning : Chor-RNN

© Crnkovic and Friis 2016

The choreographic system “Chor-RNN” trains a recurrent neural network on motion capture data from a single dancer. Once trained, the system is able to generate movement sequences that mirror the style and syntax of the dancer. More information is available here.

This website was created using Coventry.Domains, a service of the Coventry University Group. The information, views, opinions and discussion contained on this website are those of the author(s) and do not necessarily reflect the views and opinions of the Coventry University Group. For more information on privacy, cookies, takedown requests and more, visit our policies page.