Will digital avatars replace runway models?

As we scoot ever closer to the strange and wonderful future that Hollywood has always promised, we continue to fulfill what were once far-fetched expectations. As such, researchers at Manchester Metropolitan University have created unbelievably...

by i-D Team and Adam Fletcher
22 September 2014, 9:30pm

Image courtesy Manchester Metropolitan University

These avatars, by the way, are not the wimpy holograms of yore. Initially developed for ballet analysis, the avatars begin as data collected from a highly sensitive 3D infrared body scanner. Using standard motion capture technology, joint movements are plotted to create a virtual skeleton. The skeleton is then synthesized with the 3D avatar with the help of software like MeshLab, Maya, ZBrush, and MotionBuilder. Designers now have the option of sending these avatars down the runway in digital dress, piloted by real models in motion capture suits, or going the low-key route by rendering an entirely virtual fashion show to video file, to be dispersed as easily as sending an email. We spoke to Andrew Brownridge, one of the researchers on the project and co-author of its findings, about uncanny valleys, alien-like humanoids, and what it's like to meet your virtual self.

How did you first become interested in developing these avatars?
The interest in creating the avatars came from ballet analysis, which uses motion capture data displayed in a rather basic format. After seeing the analysis of the motion presented post-performance, we considered that it would be more aesthetically pleasing to have more than just a 'stick man.' My personal interest stemmed from research on the 'uncanny valley.' [A hypothesis in human aesthetics which states that when human features look almost, but not exactly, like real humans, observers respond with revulsion. The uncanny valley is a dip on the graph of the comfort level of people as subjects they view progress towards the semblance of a natural, healthy human.] 

You were interested in the 'revulsion' aspect? 
I was interested in the revulsion but also the opposite. It was stated by Mori, an uncanny valley theorist, that movement had a greater impact than appearance. I wanted to see whether combining this with a physically accurate representation of the same subject would be emotionally acceptable to the viewer. 

Have they been? Do these avatars look almost exactly like a living human?
Yes, the avatars look remarkably lifelike and have been quite readily accepted by viewers. I was, in fact, the first test subject! I personally found it fascinating seeing an accurate representation of myself move within the virtual environment. 

Very cool. Are the avatars about as lifelike as wax figures, like the ones at Madame Tussauds?
I find wax figures rather creepy! The addition of dynamic motion makes the appearance much less creepy. 

Do you need a model or could you just create someone from scratch?
Virtual characters are dependent on the skill of the 3D artist, whereas our avatars do not require the time or expertise of a digital modeler. We have animated self-modeled virtual characters for art exhibitions in Manchester and Bangkok.

What was their role in the exhibitions?
One was demonstrating augmented performance, where an alien-like humanoid was dancing in a fantasy woodland environment. The other was based upon a knight character acting out the final omitted scene from A Clockwork Orange while placed in a virtual world used as a virtual studio. 

How do you imagine these avatars could be applied in other uses and disciplines?
The ability to animate the characters should enable use of the avatars and interaction across the internet. This could be used, for example, for a dance teacher in Tokyo coaching a student in Alaska in real time and with accurate analysis, and similarly within the fashion world for virtual fitting rooms or fashion shows. 

I read that one option for fashion shows is to have a model wearing motion capture outfits?
The reason the model would wear the motion capture suit would be to drive the animation of the virtual character. So, the model would perform the motion and the avatar would represent that motion to the audience. Therefore, the model would not need to change outfits between walks, only the character or outfit would need to be changed within the virtual environment.

In the future, do you see these avatars replacing people entirely in some areas (such as ballet or modelling)?
I don't think people will ever be entirely replaced. There is and always be a desire for real human interaction. I do think, however, that it will offer many more options and facilitate what in the past has only been contemplated. I think it will increase accessibility to the viewer or user, enhance the creative power of the designer, and benefit the teaching of motor skills and performance.

When do you estimate these avatars will become commonly used?
These avatars could be used to some degree now, and I foresee that we will begin to see them used more frequently within the next 12 months.


Text Hannah Ghorashi
Image courtesy Manchester Metropolitan University

Think Pieces
fashion interviews
hannah ghorashi
andrew brownridge