RESEARCH PROCESS

A student from the USC Kaufman School of Dance, Valerie Chen, learned and performed the three movements series using the USC School of Cinematic Arts’ Performance Capture facilities. The motion capture data was supplemented by filming the performance which allows a visual validation of the movements. The resulting performance data was then gathered, conformed and subjected to computer vision libraries in order to perform a meta-level analysis of the three distinct movement traditions. The results are also being compared with in utero movements that have been posited as universal embryonic gestures. 

Motion capture is a technology with huge potential for the arts and humanities and there is some nascent work in this area that is exciting. Indeed, motion capture can recreate complex movements and realistic physical interactions and intricate movement patterns and transfers of weight and forces that can be easily recreated in a physically accurate manner, opening them up for study using increasingly sophisticated computer vision algorithms. However, motion capture is extremely cost prohibitive requiring spacious facilities, expensive hardware, specialized software and operator expertise, all of which has limited its uses in the humanities and in arts-based research. Fortunately, the USC school of Cinematic Arts has a sophisticated performance capture apparatus—in fact it is the most sophisticated of any university in the county and this affords us relatively unprecedented access to these tools.

Arts based research is iterative in nature. It spans knowledge fields offering a way of testing theories and keeping them grounded in the material world. Motion capture facilities are scarce, the hardware is expensive, and the process requires specialized training and expertise and this prevents the sort of iterative research that we are poised to complete for this project. Our particular approach to arts-based research is one of mixed methods, which involves choices of qualitative and quantitative approaches as are appropriate to the project, with a special emphasis on those that will render results from multiple viewpoints. For this endeavor, we will employ the techniques of computational analysis, choreographic and physiological analysis, media-artistic analysis, and data visualization. Findings will be interpreted technically, semiotically and culturally in conversation with our expert advisors from the fields of Dance, Computer Science and Bio-Kinesiology.

Working with human motion capture in this way is a technically iterative process which involves setting up software, and calibrating the motion capture suits with cameras. Motion capture of the body includes a three-part process of sensing the motion, processing the sensor data, and storing the processed data. Additionally, with our advisors we will be assessing open kinetic movement datasets and refining what physiological and/or choreographic elements we will be analyzing.