(Part One of this series is available here.)

To be clear: 3D customer avatars are in their infancy at the moment, so this is not a pressing challenge that the industry has to solve today.  But it is one that we need to be acutely aware of as we move further and further towards an all-digital shopping experience.  Because to get us there, the way we capture and render consumers’ digital counterparts is going to have to become a lot more sophisticated.  Which, as I’ll explain now, is not necessarily the same thing as becoming more accurate.

Even if only a small percentage of the potentially huge customer cohort for digital trying-on were to be deterred from using the technology because its results don’t look enough like “them”, that would prove a significant problem.  But there’s also another side to consider: a lot of people also recoil at seeing obviously doctored images of the kind that smartphone camera software used to serve up when “beauty mode” or an equivalent was engaged.  We’ve all seen these: skin so smooth and rouged it looks like paint, and fake soft focus everywhere. Scroll through any professional Instagram influencer’s feed and you’ll notice that trend for over-processed photos was very short-lived; today there’s a much more careful balance being struck between aesthetics and honesty.  And that’s where I believe the fashion industry will need to go if digital avatars are going to start appearing on e-commerce storefronts: not photoreal, not too obviously flattering, but somewhere in between.

Like many other aspects of the fashion industry, the luxury sector may need to be in the vanguard for finding the right balance between accuracy and aspiration when it comes to creating pleasing aesthetic representations of their customers.  This is, after all, what luxury does best: making customers feel special by tastefully stepping beyond the everyday.

Discerning luxury customers, then, are likely to be the best possible test cases for a new era of digital avatars. And luxury brands and houses, who have also pushed the envelope with new experiences and digital innovations in physical stores, are in a unique position to create a new, more personalised digital ‘boutique’ – and they have the will, the need, and the extra margin to put the customer (or a better digital representation of her) at the centre of it. 

Speaking scientifically for a moment, accurately recreating a person’s face in photoreal form is a hard problem to solve – whether you then go on to tweak it or not.  Even the entertainment industry’s best attempts at fully CGI humans still falls short of being completely believable.  Ultimately, it’s probable that a kind of ‘mathematical formula’ that works with faces will be developed to make this process easier.  Contrary as it sounds, although perfect beauty is fairly rare, beauty itself is often the expression of the average.  It is the divergence from the norm that usually detracts from the attractiveness of the individual – prominent features such as an extra-long nose, ears that protrude more than usual, or a chin that is less defined than is normally seen.  These are typically the aspects of ourselves that we do not like to be shown or reminded of – because our minds tend to exaggerate their impact on the way we look far more than the features themselves warrant.

For this reason, it might be advantageous for a personalised digital avatar platform to merge each individual’s scan or photograph with the closest idealised ‘beautiful’ image to them, to create a hybrid ‘flattering’ facial representation of the person.  This would be distinct from applying an overblown filter over a photograph, because it would involve making far more subtle changes to the underlying geometry of the face, helping to ‘smooth out’ divergent facial characteristics the same way we do when we look for our best side in a mirror, or angle our heads a certain way to pose for a photograph. 

Skin textures (such as scarring or wrinkles) could also be ironed out, provided this was done equally subtly, and skin colour evened up – with special attention paid to preserving racial variations.  The final result would then be recognisably ‘the customer, but on a good day’: tending towards a deft homogenisation of all individuals using the system.  In a sense, we’d be coming back to what Apple tried to do with Memojis, but starting from the individual and working backwards towards a less accurate reflection, instead of other way around.

We, as an industry, should remember, though, that personalisation should always be optional – lest it end up feeling intrusive.  People may choose to use an automatically-generated generic avatar for simple privacy reasons (and after the slate of personal data breaches that hit the headlines last year, I can’t say I’d blame them) or they may be part of a demographic such as plus-size, where being shown an unflattering render of themselves would run completely counter to the message of empowerment and self-respect the brand in question was trying to communicate.  Speed and efficiency will always be of the essence to the shopper who simply wants to buy, rather than experience, so we must be careful to make sure that technologies like personalised avatars are seen as ‘added value’ extras, rather than coming across as an unnecessary hurdle for disinterested shoppers to overcome.

Ultimately, though, unlike Memojis (I don’t expect an avalanche of these to start after this article is published) there will be a lot of fun to be had with e-commerce avatars, albeit with a deadly serious business purpose underneath.  As hard as the hurdles might seem to overcome right now, I believe there will be a time when e-commerce avatars are fully customisable – probably sooner rather than later.  Starting with high-end fashion and eventually trickling down, consumers will be able to change up details such glasses and hair colour, add tans for trying on beachwear, vary their makeup, and even add false eyelashes and nails for a complete look with occasional apparel.  Added value could be gained from watching animated avatars participating in the activities (such as sport, for example) that the purchase of the apparel is anticipating.

And, with the consent of the customer, there will no doubt be avatars used by pricey, celebrity-endorsed designer brands, which, although they demonstrate how well the apparel fits (the main purpose, after all, of their deployment), will also be air-brushed and Photoshopped further (or the digital equivalent) to create the fantasy of perfection that works so well for selling fashion today.

This is avatar, not just as fit tool, but as truly personalised mannequin and hugely valuable sales assistant.  The thought of a throng of a customer’s own (convincing enough, but never too cruelly accurate) avatars treating their originator to an exclusive ‘fashion show’, as they strut the catwalk towards the checkout, is enough to make any retailer’s heart glow.  And avatars powered by the same underlying technology could soon be doing the influencers’ work for them as they stroll across social media platforms in real-time, displaying an endless aisle of stock.

If we get the first impressions right, these technologies are going to last.