AMIS: An Audiovisual Dataset With Multiple Representation of Featured Actors' Performances

What is it about?

In this audiovisual dataset, we have captured performances by three actors from different ethnic backgrounds (monologues and two-person conversations) simultaneously in four different visual formats. The formats are talking-head videos, which look similar to the videoconferencing-style frame; full-body videos, which introduce the added aspect of body language; and animated avatars and volumetric avatars, which offer a realistic representation of the actors in a shared virtual space.

Featured Image

Photo by visuals on Unsplash

Photo by visuals on Unsplash

Why is it important?

In the context of social XR and the futuristic teleconferencing solutions, many visual representations of the conversation partners are conceivable. It is possible to talk to a person through a 2D display like a TV monitor, or it is also possible to virtually be present in the same space with the remote user with the help of new software and networking solutions that make it possible to have a teleconference as realistic as a face-to-face interaction. This dataset provides all of these possibilities as passive interaction materials, which in turn can be used to test user preference and impact on interpersonal bonds.