This is the third in a series of articles where – inspired by a conversation with a leading 3D vendor – I commit to helping fashion businesses who are interested in 3D, of which there are more than ever right now, to better understand the potential benefits and plan their ways forward.
First I looked at the practicalities of material scanning; second I explained the importance of best-practices that can ensure that virtual materials look, feel, drape, perform, and share other tactile attributes with their real-world counterparts.
Next, I want to talk about fit. More specifically, I want to look back at the history of sizing surveys – how, why, and where they were deployed – and how their impacts are still being felt in pattern development, garment fitting, and product return rates today. I’m also going to make some suggestions on how we might re-examine the process of gathering sizing data, and the potential that new methods offer for providing brands and retailers with continuous, accurate sizing data from their target demographics. After all, consumer loyalty is now one of fashion’s most valuable currencies, and better-fitting clothes translate into happier customers and more brand advocates.
Where did the idea of sizing surveys for fashion and retail come from? Not surprisingly, they have their roots in other disciplines where there was a similar need to understand the proportions of human bodies in a particular region and at a particular time. These anthropometric studies were a big part of evolutionary science – where researchers set out to understand physiological differences between modern humans and our ancestors – as well as in medicine, health, and fitness.
Sizing surveys with garment fitting in mind started by adopting the same principles as those academic surveys, but with much larger cohorts. The first national sizing survey took place here in the UK, around 70 years ago, with a view to producing a detailed data set of body measurements that retailers and home shopping businesses could use to test their block patterns against the real measurements of the general population.
Interestingly, since this survey came just a decade after the end of the Second World War, with rationing still a fresh memory for most people, the UK population came across as being leaner than it perhaps would have done at a time of more stable peace.
Beyond retailers, this data was also used indirectly by at-home dressmakers (an industry with a strong association to wartime) who, through the 1950s and 1960s, would base their creations on generic cut-out patterns and grades provided by mass suppliers like Simplicity and Vogue.
The military also provided the basis for a long-running series of sizing surveys carried out in the United States, including the well-known Army Anthropometric Survey (ANSUR) which assessed 240 points of measure across 75,000 enlisted men and women. Indeed, the US military ran more than 40 of these surveys between 1945 and 1988, culminating in the ANSUR.
There was then a gap of around half a century before the UK returned to nationwide surveying. A massive undertaking, the next national survey was conducted by SizeUK between 2001 and 2004.
I remember this time well, because I saw the results being used first-hand. Then, I was working with GUS (Great Universal Stores) and Burberry – both of whom were using the SizeUK data to help improve fit with the goal of lowing returns. This was an especially important metric in home shopping; at the time catalogue sales were accruing return rates of between 40% and 60%.
SizeUK measured 11,000 participants, which in today’s standards is a relatively small sample size. 140 key landmark points (body measures) from each participant were recorded, which in total provided some 1.54 million data points. This was, according to the researchers involed, the minimum number of subjects required to give an accuracy of ±1cm with a 95% confidence level. For SizeUK the primary recruitment criteria were women and men, with seven age groups spanning a numbers of years such as, 16-25, with participants coming from across three geographical regions in the U.K. containing eight data collection locations.
Commercially speaking, SizeUK was a collaboration between the UK Government, 17 major UK retailers, leading academic institutes and technology companies – including those associated with 3D scanners, 2D CAD/CAM and PDM (Product Data Management) software.
Indeed, SizeUK was also the first survey to take advantage of 3D body scanning to automatically extract measurements, rather than relying on painstaking manual methods. The survey also included a qualitative element, with researchers trying to gauge participants’ satisfaction with the clothing they usually bought. The results showed that over 60% of UK shoppers had difficulty finding clothes that fit, and they also revealed that the average female waist size had increased by 16.5 cm since the 1950’s post-war era.
After SizeUK finished, around 2004, other European projects emerged, and the idea began to coalesce for a framework of sizing standards. This would help to enable the seamless integration of 3D scanners, 2D CAD and 3D virtual technologies into the fashion industry, bringing some degree of consistency and standardisation to international sizing, and opening the door to some of the virtual fitting and customisation technologies we see today.
So, as remarkable an effort as SizeUK and other large-scale surveys were, I believe the industry needs to ask itself whether their results are still relevant today. And, by extension, should we, as an industry, be thinking about undertaking new surveys on that sort of scale – and if we did, what new avenues would we have for collecting that kind of data?
Since SizeUK began, in 2001, 3D body scanning has marched ahead in terms of accuracy. Today, scans are accurate up to a tolerance of 0.1mm (a 50% improvement since 2001) and can obtain anywhere up to 250 body measurements from a single person in less than 90 seconds. The output is a set of three-dimensional XYZ axis coordinates made up of between 600 and 800 different data points.
How does this all work? There are 3 main technologies that the 3D scanners on the market are utilising today:
- Photogrammetry: The process of photogrammetry actually involves no scanning of the subject. An array of pictures are taken from multiple angles of the subject, they are then combined into a single 3D entity using a predefined algorithm. This is a quick method of ‘scanning’ which also captures colour and texture but lacks intricacy and resolution.
- Structured light: In this method multiple bands of light are projected onto the subject. When they hit the subject they become distorted, much like running a length of string along an object, which will move based on the changes in the object’s height. As the projector applies light to the subject, a camera records the distortion of the light in order to calculate the structure of the subject through triangulation.
- Hybrid: A hybrid of these two technologies can be used if it is necessary providing colour to a blank structured light scan.
These approaches cover essentially all the bases for dedicated, physical body scanners. But the new wave of size scanning – pioneered by businesses like 3DLOOK and Sizestream – have other tools for collecting body data using either standalone hardware or consumers’ smartphones, together with A.I. & M.L. algorithms. At the time of writing, though, there is still a lack of adoption from brands and retailers of either at-home or in-store. This is surprising because both methods offer a clear benefit – accurate insight into real sizing across a retailer’s customer base – with a minimal or non-existent physical footprint.
This, in my opinion, is where the future of sizing surveys lies. Rather than national surveys of large population groups, they can be linked to retailers or a brands customers or even on an individual basis, with the resulting data being used to make instant fit recommendations for the customer who scans themselves – as well as continuously tweak the block patterns to improve overall fit across the retailer’s entire range.
What’s more, if we could seriously consider moving to new business models including Made-To-Order, informed by those individualised scans, then we would be able to leave the model of high waste and high returns in the past. It’s not exactly tailoring – which remains its own hyper-specialised discipline – but these measures could become part of a very large mass-customisation dataset. This could enable retailers and brands to serve not only that individual but also people of very similar sizes and shapes, meaning that they could manufacture small runs based upon the mean average size. This would result in holding less redundant inventory, and it could help turn today’s traditional business model from a supply-chain to a demand-chain.
There is little reason that the 3D models generated during the scanning process could not also be used for virtual-try-on and other augmented reality applications.
The technologies themselves, though, are not going to have much impact on consumer buying patterns. Unlike past size surveys, which were used by brands and retailers for purposes that the customer did not directly see, the new era of 3D scanning and body data will need to be actively promoted by brands and retailers, since it requires customer participation. So, a message to retailers and brands: if you really want to reduce your return rates than it’s time to take action, there’s no other way than a scientific approach!
Retailers should also consider incentives for customers who share their measurement data, ensuring that as well as improving fit for commercial reasons – reducing returns dramatically, and cutting down on waste – new technologies will also improve the reputation of the businesses that use them.