The world of fashion in recent years has begun to digitalize rapidly. Between virtual showrooms and online shopping that was bolstered by the pandemic, fashion retailers have had to adjust to the needs of customers.
One of the industry’s biggest challenges has been to get a shopper to purchase the items in their cart, which can be difficult for customers who can’t imagine how the garments or accessories may appear on themselves instead of a model. That is what StyleScan Inc. is trying to change by offering artificial intelligence-powered visualization technology to improve the online shopping experience.
The company, based in Brentwood but with a primary studio in Van Nuys, was founded by Larissa Posner in 2018. Posner holds a certificate in business application of machine learning from the Massachusetts Institute of Technology and more than 10 years of experience advising companies on investor relations and strategies. Posner also spent time working as a catalog model.
While working in quantitative finance in 2018, Posner learned she had no time to shop at physical stores and realized several were closing around her. She considered that the “retail apocalypse.”
“Barneys closed on the Upper West Side next to my house. Every major retailer was closing chains of their stores because at the time they were calling it the ‘Amazon effect,’” Posner recalled. “Everybody started shopping online, not going into stores. At the same time, I looked at Shopify and some of these other platforms, and their stocks are rallying and they’re doing amazing. Naturally, you have to assume convenience overtakes and e-commerce will be the future. So, what can we do now to build a product that the market will later need?”
Posner said she wanted to be her own customer. She had the idea of uploading her photo and being able to see herself in a retailer’s inventory. She assembled a group of computer scientists, venture investors, and fashion executives to develop virtual dressing technology.
Posner went to the team in financial services at Corporate Profile, an investor relations firm representing publicly traded biotech, tech and financial services companies and asked if they could predict stock prices, was it possible to predict women’s sizes and potentially show it on women?
“The answer was ‘yes.’ We now have plenty of breakthroughs in computer vision and AI. We can take one garment from one photo superimposed on another photo,” Posner said. “We’ll need to train and build neural nets and train algorithms. But yes, it is possible. And so, I raised the first million dollars. That was 2019.”
That’s when things got serious.
How it works
The StyleScan image-processing technology takes a 2D photo of a clothing item or accessory and converts it into a 3D volumetric mesh that can be superimposed onto humans of different shapes and sizes.
Posner said this technology could be used to make several areas of the fashion industry more sustainable and cost-effective, one of them being traditional photography.
“A single photo of a garment, let’s say, on a mannequin. Now, that garment can be shown in each size instead of in just one sample size. Now, it can be shown on women with different skin tones, body shapes, sizes,” Posner said. “And it’s all digital. (We) can generate zero waste. We can turn things around in a matter of 48 hours, whereas if you were to do a traditional photo shoot it would be two weeks at a minimum, probably three weeks. And then cost, you can’t possibly compare cost.”
The StyleScan team surveyed 12 apparel companies of various sizes, from small to large, to see how much e-commerce photoshoots cost them. For smaller retailers, one final image can run about $50, while larger luxury brands pay up to $1,000 an image.
“With StyleScan, it’s the cost of a bottle of water,” Posner said.
In addition to being cost-effective and sustainable, Posner said retailers who are using their technology have seen an increase in their sales and conversion rates.
The company offers two types of services: Model Switch and Style Switch. In the first, a customer can choose one item on the retailer’s site and choose from different models to visualize the piece.
“Our next product is Style Switch, where you can also dress the model in the items from across the retailer’s inventory selection and be your own stylist and build full looks and create a kind of preview before you buy,” she said.
Possible clients looking to use the company’s technology will choose from a list of subscription tiers based on volume and how many models they want for their site.
“Do they just want one item, one model, or do they want one item across multiple sizes, multiple skin tones? … It’s based on volume, revenue and how relatable they want to get. Also, it requires quite a bit of integration into our partner’s e-comm. Integration is just us managing their Shopify account.”
The company is an official Shopify partner. Clients give them access to Shopify, where they upload any images are included in the subscription package.
StyleScan also plans to offer different tiers of models.
“Maybe somebody will be an influencer with a certain amount of following and that will be something you can access with your higher premium package,” Posner said. “A package where you can also not just switch a model but also switch styles. That is really our premium creme de la creme package.”
After 30 days with StyleScan, jewelry seller Millianna saw its conversion rate increase by 93% and total dollar sales go up by 367%.
“Ordinarily, they photographed earrings on the white background, nothing else. And what they found is that a consumer comes to the site and then can’t visualize the item. How big is this earring? Where does it hit you? And so, the customer goes now to Google and starts researching the item and you have a potential of losing a customer,” Posner said.
“With StyleScan, what we did is we added our magic button ‘virtual model try-on’. Now, you can see it on different women with different skin tones and you can see the item and visualize how big it is.”
Posner said it is key that they do not operate as a black box company where customers submit photos of the garments and then download output images.
“We have professional designers, professional stylists (and) professional creative directors on staff, so it’s almost like a cyborg where a human operates the software and a human has quality control of anything that comes out and that is presented to the client,” Posner said. “We make sure that there is an aesthetic oversight over the software result.”
With the crossover between tech and fashion come a lot of unknown territories, but Posner said technology companies and decision-makers will always be going through a learning curve when trying to understand how to navigate these changes in the industry.
Posner also said she is proud to see her clients have seen a direct effect of StyleScan on their businesses and who are excited to work with them even more.
“Since we started…we already are seeing customers coming back and doubling their subscription and upgrading and wanting more SKUs on the per month basis…powered by styles. Now that they’re seeing not only the aesthetic upgrade, but they’re also seeing the efficacy data,” Posner said.