Mapping beyond the uncanny valley: A Delphi study on aiding adoption of realistic digital faces

Abstract

Developers and HCI researchers have long strived to create digital agents that are more realistic. Voice-only versions are now common, but there has been a lack of visually realistic agents. A key barrier is the “Uncanny Valley”, referring to aversion being triggered if agents are not quite realistic. To gain understanding of the challenges of the Uncanny Valley in creating realistic agents, we conducted a Delphi study. For the Delphi panel, we recruited 13 leading international experts in the area of digital humans. They participated in three rounds of qualitative interviews. We aimed to transfer their knowledge from the entertainment industry to HCI researchers. Our findings include the unexpected conclusion that the panel considered the challenges of final rendering was not a key problem. Instead, modeling and rigging were highlighted, and a new dimension of interactivity was revealed as important. Our results provide a set of research directions for those engaged in HCI-oriented information systems using realistic digital humans.