When players in online games engage in conversation with each other, often through a chat window, their graphical avatars typically do not exhibit interesting behavior. This paper proposes using a model of face-to-face communication to automatically generate realistic nonverbal behavior in the avatars based on what is going on in the conversation. It describes Spark, a flexible XML based architecture that makes this possible and reports on a user study that shows how such avatars could significantly improve online conversations.
|Title of host publication||Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|
|Number of pages||12|
|Publication status||Published - 2004|
|Name||Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|