Realistic Digital Avatars and the Uncanny Valley
And, unlike current desktop and mobile technology, it does so through deeply immersive scenarios. These virtual experiences enable presence, embodiment and immersion.
The VR avatar is one of the most important tools for creating presence, embodiment and immersion. In computing, an avatar is a graphical representation of a human user, or of the user’s character or persona. Customisable digital avatars were first used in a computer game by the 1979 role-playing game Avatar. Avatar was one of the forerunners to the modern Massively Multiplayer Online game. Avatar ran on the University of Illinois's PLATO network and found fans globally. Today, avatars are used in a variety of online settings including social media, virtual assistants, instant messaging platforms, games and digital worlds such as World of Warcraft and Second Life.
Why are avatars so important? Humans evolved to communicate face-to-face with one another. A key aspect of this is non-verbal communication through body language and facial expressions. We react better and are more engaged with human interaction-based digital experiences. Tools such as emoji, GIFs and media snippets simulate this expressiveness, but they only compensate for a lack of face-to-face interaction. Hence, avatars are crucial.
Key to creating richly immersive VR is the capacity for digital avatars to communicate using non-verbal cues. The success of digital avatars depends on their appearance and their behaviour. They need to be realistic and bear a resemblance to the person whose avatar it is, while also being expressive but not exaggerated. In short, they need to seem human. If they do, then people can use VR to establish a level of intimacy that they cannot using desktop or mobile devices.
But not all VR avatars are equal, and their present implementation tends to leave something to be desired. Today, realistic VR avatars tend to fall into the Uncanny Valley. The Uncanny Valley is a phenomenon where users have a negative reaction to digital representations of people that just aren’t quite human enough. This manifests as a cold, eerie sense that leaves users feeling less, rather than more, immersed.
There are a few theories as to why the Uncanny Valley effect exists, but new research suggests that it is due to a cognitive bias against objects that stray from typical patterns. Human faces are a very well-known pattern; hence they are very prone to falling into the Uncanny Valley. This has repercussions for nonverbal communication: while cartoonish avatars can be less eerie, as they don’t try to look human, it is harder for them to communicate via facial expressions and body language. But what is the solution for this problem?
Enter ALIVE: Avatar Learning Impact assessment for Virtual Environments. This three-month project based at the University of Glasgow’s Advanced Research Centre and undertaken in collaboration with Edify experimented with ultra-realistic digital avatars. The project used the Unreal Metahumans engine to create realistic digital avatars. It compared how easily participants could identify facial expressions and how people felt about the avatars when compared to more traditional cartoonish avatars.
Although the participants generally found it easier to identify facial expressions in Metahuman avatars, they tended to find them uncannier than traditional avatars. This might have been because the Metahuman avatars, while more realistic than cartoon ones, still need to exaggerate their facial expressions and non-verbal cues. Metahumans may still be somewhat uncanny, but they are easier for VR users to empathise with, and it is easier for users to identify emotions.
A possible avenue for future research is motion capture. ALIVE focused on facial expressions, but body language is another vital part of non-verbal communication. Avatars that can move fluidly and expressively might be perceived as less uncanny and easier to empathise with. Understanding the role of embodiment and perception of social acts in virtual settings has the potential to be used for social skill training and treatments outside the classroom. As a result, the project's findings can be applied to further research and practical applications.
One issue that ALIVE faced was processing power. The study used Oculus Quest headsets. Although these headsets are capable of impressive things, they still acted as somewhat of a bottleneck, as did the rendering capabilities of the Metahumans engine. Creating a virtual environment in which every participant is represented by a Metahuman remains some way off.
This isn’t to say that Metahuman avatars can’t be used, period. Rather, they might be used in educational contexts such as demonstrations, for example. In cases like this, the demonstrator would be represented by a Metahuman so the audience could benefit from the demonstrator’s avatar’s increased fidelity. In this sense, ALIVE is ahead of the curve: as the technology that powers VR headsets matures, their ability to render complex graphical assets such as Metahumans will increase. This might not be far off. Just think about how quickly smartphone technology has advanced in the years since the iPhone first launched. As the technology that powers VR headsets becomes more mature, the Metahumans engine will inevitably pose less of a challenge.
ALIVE won a prestigious award for research into digital pedagogy and immersive learning. This was presented at the 8th Annual Conference of the Immersive Learning Research Network (ILRN). The team behind ALIVE presented the results of the research project at this conference to a global audience of VR researchers, practitioners and experts.
Want to learn more about ALIVE and the technology that powers VR avatars? Subscribe to updates from Edify for the latest in VR research.
Curious about how you could leverage virtual reality to enhance learning outcomes?
Find out more about how we partner with universities on our dedicated higher education page.
