ACM - Computers in Entertainment

Kabuki Mono: The art of Kabuki transformed

By Naoko Tosa, Ryotaro Konoike, Iroshan Horathalge, Yohan Fernandopulle, Kasthuri Jayarajah, Owen Noel Newton Fernando, Ryohei Nakatsu

The limitations created by geographical and technological boundaries profoundly influenced the development and sustainability of traditional artistic cultures, which have long been popular among the locals to whom the art form is native. Today, however, with the “connectedness” computing brings along, it is possible to transform such art forms into globally accepted traditions. Japan is a nation that flourishes with tradition. From traditional Kabuki to pop culture oriented Manga and Cosplay, the fundamental motivation behind these traditions and modern cultures remains as the ardent desire of the human to transform himself as another who is more appealing in character and aesthetics. Kabuki Mono is a system that binds tradition with modernization leveraging the power of computing.

The existing Kabuki Mono is a series of computer vision and image processing techniques including facial feature recognition using active shape models, image labeling and image warping. The system intakes an image of a human, detects and recognizes the face and selected features therein to virtually apply the Kumadori makeup to the person’s image and adorn with the Kabuki wig. The system also brings alive Manga and Cosplay.

BeforeAfter

In addition to providing an original method for increasing the accuracy for facial feature recognition, devise a “Feeling Map” (see below) based on the unique classifiers of the Kumadori makeup variants. The details of related work are accessible from http://www.naokotosa.com/.

Currently, related work is being pursued as improvements to the existing Kabuki Mono. A segment of the work is concentrated on increasing the accuracy of some of the intermediate processing steps, namely, the image labeling step which segments the human face form the rest of the image based on skin color, and the image blending stage of the human face with the Kumadori makeup.  For more accurate facial skin recognition and blending, experiments are under way by utilizing Markov random fields and Poisson seamless cloning respectively, both of which are expected to result in more realistic visuals. Currently, we are also focusing on extending the same technology in the 3-dimensional space and also, making it accessible via mobile devices based on a client-server model. Extending to the third dimension, gives rise to a series of new challenges including pose estimation, feature tracking, illumination dependencies, etc.

The Future

The vision of such a system is to transform the way in which people experience traditions and break the regional barriers making way for global reach and acknowledgement. The future of this system promises an interactive media wall that reflects on the general public as Kabuki actors.

Further Reading

Tosa, N., Konoike, R., and Nakatsu, R., KABUKI-MONO: The Art of Kumadori Facial expression for Manga and Cosplay. Second International Conference on Culture and Computing, Kyoto Japan, pp 98 – 103, 2011.

Copyright © 2019. All Rights Reserved