Emotion by motion: expression simulation in virtual ballet

Learning and teaching choreographies can be an arduous task. In ballet, most dancers learn by emulation i.e. "watch, copy and learn". The teaching process not only instructs the order of steps but also requires explaining the quality required for the performance. Notational systems are used in almost all fields of study, and dance notation with inferred domain rules are used to aid in the teaching of choreographies and dance. Few professional performers can read written choreography let alone visualise the movements involved, and this represents a considerable barrier to the utility of choreography in its written form. Real-time computer graphics are ideally suited to bridge the gap between written choreographic notation and performance, via the creation of a virtual dncer. It would be useful for professionals to better understand choreography and notated ballet scores as well as assisting to teach dance at all levels. To understand the needs and derive methods for a virtual ballet dancer system, there are three distinct parts and these provide the structure for this thesis. The first part researches into the fidelity that is required for a virtual ballet dancer. From this analysis, expressive motions are parameterised using Laban's effort parameters and results presented how participants distinguished between the different emotions performed at various levels of fidelity. The results provide understanding on how Laban's parameters define variations performed during the different expressive movement and the level of interpolation required for a user to distinguish the expressive performances. The second part presents methods for setting and evaluation of specified ballet positions (key poses) which form the foundation for ballet steps. Mathematical rules are developed and explained within the context of the ballet domain rules being represented. The resulting poses defined by the dance notation and the mathematical descriptions are evaluated by professional teachers and notators. The results are presented showing how basic ballet positions are accurately posed for a perfect dancer and variations from the perfect pose to the real-world. These poses are used as the foundation for layering the expressive algorithm on. The final part presents how Laban's Effort factors can be used for expressive interpolation between key poses, i.e. the quality of movement. Methods are analysed and algorithms implemented to develop variations in the movement between the set dance positions. These variations are matched to the expressive performances of real dancers analysed in the first part of this research in order to evaluate the algorithms derived with actual expressive performances. The results presented are the first major steps to produce an animated "virtual ballet dancer".

Name dc:type.qualificationname

Ph.D

Level dc:type.qualificationlevel

doctoral

Grantor dc:publisher.institution

University of Leeds

Year dc:date.issued

2005

Author dc:creator

• Neagle, Royce James

Advisors dc:contributor.advisor

• Ng, K.C.
• Ruddle, R.A.

Identifier

uk.bl.ethos.529705

OAI identifier oai:identifier

oai:etheses.whiterose.ac.uk:1331