Venturing into Fashion Tech

Build It Series: Harnessing Truthful Data & AI in Fashion Fit Tech with Katy Schildmeyer

• Beyond Form • Episode 63

Share episode

Fashion's Fit Dilemma:
Today's episode features fashion garment technologist and fit expert Katy Schildmeyer. As online shopping increases, the challenge of ensuring accurate garment fit remains ever-present. Katy and Peter Jeun Ho discuss the pressing need for enhanced fashion education, particularly in the communication of fit and the creation of garments that cater to all body types, including the aging population. They reflect on the historical transformation from shopping experiences of the 90s, to the rising return rates in online retail today which can be more than 30% for some fashion brands.

Ensuring Data Integrity to Find the Fit Truth:
The interview examines the integration of AI into fashion tech. Accurate data is crucial, and Katy emphasises the importance of truthful data in training AI models to avoid misguided innovations. We spotlight industry leaders like COS, Tapestry, and Japanese denim companies, commending their practices towards better fitting products. But how does one find the truth in fit? Find out in this episode.

Connect with Katy: linkedin.com/in/katherineschildmeyer
Katy features Beyond Form startup AIMIRR in her research: aimirr.com

Send us a text

Support the show

--------
The show is recorded from Beyond Form, a fashion tech innovation platform. We build, invest, and educate fashion tech entrepreneurs and startups. We’d love to hear your feedback, so let us know if you’d like to hear a certain topic. Email us at podcast@beyondform.io. If you’re an entrepreneur or fashion tech startup looking for studio support, check out our website: beyondform.io

Peter Jeun Ho Tsang:

Hello, I'm Peter Jeun Ho Tsang, founder and CEO of Beyond Form, and welcome to the special series Build It, where I speak to founders and their clients as to how they're building fashion technologies. This series gives you a glimpse into topics such as personal struggles, technical challenges, working in fashion and more. On today's episode, I'm speaking to Katy Schildmeyer, who is a fashion fit technologist and an academic researcher in fashion fit tech. Her most recent paper included the Beyond Form AIMIRR, where she discusses digital avatars, body fit, body scanning and why and how some brands are getting it so wrong.

Katy Schildmeyer:

Anything that you're dealing with mathematically, it has to start from a place of truth in AI, because if you're going to train any engine off of that AI, it has to be honest, and so if you have a dishonest representation of the data and information, yes, you are shooting yourself in the foot, and you have to then go back and untrain that data, and to do that is a very sticky situation.

Peter Jeun Ho Tsang:

Let's get this conversation going with Katy on today's episode of Venturing into Fashion Tech. How are you today, Katy?

Katy Schildmeyer:

I'm good. How are you Peter?

Peter Jeun Ho Tsang:

I'm good, thank you. I'm looking forward to today's conversation. It's the first time we've actually had somebody talking about academic research in the space of fashion technology and, for this episode, specifically fit technology and 3D avatars and simulation, which obviously is a very hot topic right now within the industry. But before we get stuck into your story and the research that you have been conducting, I just want to set the scene for our listeners. So in 2024, research in fashion fit technology is advancing rapidly. In October this year, for example, online retailer zalando, based in europe, launched the feature for their customers to be able to create their own 3d avatars based on their own body measurements, which is quite novel, because not every online retailer does this, just does this. Every online retailer even has that feature of creating virtual try on outfits to go on the avatar. So it is advancing.

Peter Jeun Ho Tsang:

In a recent study that they actually conducted in partnership with you gov in the uk, it actually found that around 36 percent of adults said a negative fitting or shopping experience would stop them from purchasing, with 23.6 percent indicating that fit was a major concern for them. So I'm sure you have pretty strong opinions on that. Why it's such a high number? So I'm going to dig into that a little bit deeper on. In terms of academia and research, many academic institutions are looking into the issue of fit. For example, mit in the USA released results this year on their 4D knitted garment research looking into custom fit and custom design, and researchers from Cornell University released their findings on fit models and 3D prototyping in September. So quite clearly it's a topic that many people are looking into.

Peter Jeun Ho Tsang:

However, fit is still such a major, major issue for the industry at large and I'm sure you have lots of opinions as to why some brands get it right and why some brands get it so wrong, shall we say. But before we get into your opinions on that, let's lead into you as an individual. You have worked with AI Mayor, which is part of the Beyond Form portfolio. However, how did that get you into before you got to that stage? Actually, how did you get into fashion technology, to the fashion industry, retail and fit technology? Tell us about your journey into the industry. Was it always about fashion for you as a child or did you just stumble into it?

Katy Schildmeyer:

no, I always about fashion. I am the middle of eight children, which meant everything I got from my sisters was a hand-me-down. Now I have a sister who's massively much taller than I am, and then another sister who has a completely different body shape, even as children, from what I was, and so anything I got from them it was always ill fitting. And you know you, you make do with what you have when you're in that size of a family. And so by the time I picked up like a sewing literally a sewing needle and a thread and not even a sewing machine I was teaching myself how to take those things in as a child.

Katy Schildmeyer:

And I think my earliest memory of like the first garment I made was when I was seven or eight and it was a little beret hat. I made it with a plate where I cut out the pattern from a plate and then an internal pattern, and then I stitched it together, literally needle and thread, and it was polka dot, various colors, and I made a little bow for it. So that was as early as I can like recall making something for scratch. And then obviously it just extended and I was lucky enough to go to an art school where I could do those things in high school and then you know, obviously, college and so on. But, yes, I've worked pretty much any kind of job you can think of and and so I have a kind of 360 view of the ecosystem that is um in existence today, but also what it was in the last, you know, 30 years of my life, um, and where things have gone wrong and where things have gone right.

Peter Jeun Ho Tsang:

You said that you've touched several parts of the industry, and I read on your bio that you have done design, you have done fit, you have done retail. Is there one part that you've always gravitated towards, or has it been just the way it's happened Fit, fit, fit, fit Hands down from it, basically. So why is?

Katy Schildmeyer:

that why, yeah, pattern and the fit process. But even how? How do we communicate fit to people? Because that that was a big part of the retail integration that I did, where I would basically go into failing divisions of of different companies as a store manager or team leader in those areas and I would fix them. So it was either do we retrain the employees or do we fire them completely and bring in people that can be molded to do these things and communicate well with consumers, because that's a huge part of any company in a fashion is that we're making stuff for other people and that product for other people must have communication with it.

Peter Jeun Ho Tsang:

Obviously, you're obsessed with fit. You recently co-wrote an academic research paper entitled Building Novel Dynamic Pattern Theory via 3D Body Scan, which is which is the title.

Katy Schildmeyer:

Tell us about why you chose this topic and what it's about, then so really that comes down to um, an organization I'm part of called the ocac it's the open circle apparel coalition and what we do is we literally try to solve the problem of fit I why things fail and also where technology may fail or how we can improve technology right to become better at doing the job of selling goods to consumers accurately. And that's really kind of where it started, where I was basically helping a company called Belladena that does made-to-measure product and she started this other forum, brought me in and that's where I met Pritesh and from there we ended up saying, okay, this is definitely a problem.

Peter Jeun Ho Tsang:

Let's figure out what's going on with body scanners as a whole in the mobile arena and how do we really solve this problem of getting accuracy out of a 3d mobile body scan and how that can then give the fashion companies real, true good data what is frustrating you the most, like through your research, because obviously the fact that you've written the paper and you've done all of this research and you've worked with companies as well means that you must have some sort of a personal book bear as well about what the industry is doing, most likely wrong so that, and that's a big question, but I'll break it down more simply into three parts.

Katy Schildmeyer:

So the first part is how we communicate with the customer. So consumers don't understand how a product is made or pattern making. And you know, I would say as a teenager in the 90s right, I understood that because we had brick and mortar stores and there was a greater integration of somebody teaching you about your body and how also garments were made, and they had whole like booklets that they would send out to their sales team to teach people about how things were made. And then, when the dot-com came through, right, you had a limitation of data in the internet at that point in time, and so they had to decide what do we scrap? And it was the communication of fit. And so a good example is a brick-and-mortar retail store would see anywhere from 8% to 14% return rate in goods sold, you know, interactively from the store with a good sales team. I think 2008 was when I started to really see this uptick in interchange of oh my gosh, we're seeing like 30% return rate at store level. So what is that? And then we looked at it and that difference was online sales. So we could already see that that margin of error was there. It was already baked into the process and then it's just continued on.

Katy Schildmeyer:

And in the last I think 2018 to now, right. In the last I think, 2018 to now right there's no information about what this garment was actually made for. Like I know, yes, your brand caters to these size ranges, but what is the body type, what is the shape, what is the age right of these consumers, and brands do a really bad job of communicating that. So that's one. And then two is again a deficit in the education system. So I have seen students coming out of two-year schools, which I would advise don't go to a two-year school If you can go to a four-year school where you're getting some form of fit education into your regime. So, understanding how to go through a technical fit process, make adjustments to the pattern, run it back through the system and ensure that that garment is made well and understand why and that requires a lot of understanding of textile as also body movement and motion to really make core judgments that are really good for fit. So that's another area that I think is wanting and we do better.

Katy Schildmeyer:

And then the other component is lack of enough body information for the core markets. Majority, most dress forms are made for people under the age of 30. And I deal with tailors all the time that end up with elderly people coming in to get their last suit made for them that they will most likely be buried in and they cannot find any dress form to make it on that that figure. So they literally tape up the person, cut it off and create like the shell mechanism to create something for that consumer. And it's unfortunate because our work can be very sacred in that sense where we're making the last garment of a person's life and they're going to be buried in it or, you know, shown that way in it. And so that you know, with respect to that, we should have more diversity in our dress forms and we just simply don't diversity in our dress forms and we just simply don't.

Peter Jeun Ho Tsang:

So it sounds like that's a really grassroots level problem. Then, starting with the fashion education, just to clarify for our listeners you mean fashion school, like those are four-year programs instead of a two-year programs. It sounds like there's a problem there and I completely agree with you. I went to fashion school and I don't think we spent nearly enough time looking at fit and I did a program in fashion design which I think says everything. I think that's quite shocking that all of the body forms are not catering to anyone above 30 and obviously our bodies change drastically, decade yep and how does that make you feel?

Katy Schildmeyer:

well, that's a problem that I have been working on, and so I myself have kind of developed a new rig system for things like that and that had to be worked on alongside chiropractic doctors and osteosurgeons to really ensure like we had correct formation data right. And in my family, you know, I've had the opportunity which I do believe it's a sacred opportunity to take care of the aging population within a family and when you know your father or your grandmother can't wear a garment because it drops off of them, because they've, like, lost their backside and nothing has been made for them, right, I can turn around and make something for them or I can adjust the garments accordingly in a way that gives them freedom, and that, I mean, that's part of, I think, the privilege of being a designer and understanding fit is that you can give freedom to a whole subsection of people, and we overlook that consistently.

Peter Jeun Ho Tsang:

I completely agree with that. And then your paper. Let's go back to the paper then. A lot of our listeners don't come from product development or design backgrounds, so even if they wanted to read your paper, they might not necessarily understand everything that's in the paper. So what are some of the, for example, the three main insights that you want somebody to take away from it?

Katy Schildmeyer:

You really have to validate your data that you're getting. When you're taking a body scan of a person, right, you really need to understand, like, what is the true shape of that person? And how does this company do a body scan through a mobile device? So are they using a figure that they've morphed out into different shapes right to get the derivative, or are they actually doing much more of what Pritesh has worked on, which is trying to get the truth of that figure into a 3D output for a brand? So that's the first portion is what is the truth of that data?

Katy Schildmeyer:

And then two is understanding that making patterns is not simply a matter of extending a shape. So quite often people think that pattern making is taking a shape and expanding it out, and that's not the case. There's so much underlying information that goes into how a pattern gets made that not all AI systems in pattern making are viable and, quite frankly, I would caution that because of the amount of information that I've even had to do in programming AI pattern making, the level of information that goes into those documents and the time it takes is massive. So anybody who's saying that they get the AI pattern off the top, I caution that and double check your work.

Peter Jeun Ho Tsang:

Do you think they're lying or do you think they're a bit misguided, misguided, okay.

Katy Schildmeyer:

Misguided Because a lot of the technicians that work on that are not necessarily pattern makers and they don't have the background that I have, where I understand how to make those pattern adjustments.

Katy Schildmeyer:

I understand how to do draping versus flat pattern making and why you would do one or the other and how those things kind of come out right so I can look at a pattern and know who it was made for because I've worked with them long enough. So a person who might be misguided doesn't understand those nuances of what a pattern maker really does. And if you're really going into a space of innovation because I see this with companies all the time they want to have innovation right. But the person that might be heading up innovation is maybe a little misguided as well, because they don't have the pattern making background or they don't have the consumer intelligence background of retail to understand what it is they actually need to be innovating, and so I always want people to ask the question is what is it we're truly trying to innovate and what is the impact of that innovation that we want to drive home to not just, obviously, our shareholders, but to our consumers, to create a better, stronger brand story?

Peter Jeun Ho Tsang:

better, stronger brand story. I mentioned in the introduction initiatives like Zalando's, for example, where they're now letting their customers create their own 3D avatars. What do you think about such products? You mentioned that obviously, a lot of the tech teams in those tech retailers don't come from a pattern cutting, pattern making or even a fit background. So do you, do you think such innovations are a little bit, shall we say, shooting their own foot, basically right from the start, because it's not set up properly? Or do you think it's a lot of lip service? It's a lot of just nice and glossy, but not necessarily adding that much value?

Katy Schildmeyer:

everything kind of holds a certain amount of lip service regardless, yeah.

Peter Jeun Ho Tsang:

Most lip service in retail.

Katy Schildmeyer:

So that I mean that's a given, I think, on anything in our culture, but especially with the foundation of AIs oddly, been privileged to have friends for a very long time, since I was in junior high school making clothing, that are now, you know, working as heads of AI for different companies.

Katy Schildmeyer:

So it's a very interesting inflection point for who I know and how they work, and so I've been able to pick their brains a little bit on this subject and they said yes, anything that you're dealing with mathematically, it has to start from a place of truth in AI, because if you're going to train any engine off of that AI, it has to be honest, and so if you have a dishonest representation of the data and information, yes, you are shooting yourself in the foot and you have to then go back and untrain that data, and to do that is a very sticky situation.

Katy Schildmeyer:

I would say that market is almost as big as the training markets for large language models is the untraining of large language models, because they have to go through the whole system and try to figure out where the links are and pull it out. It's something that I would caution anybody who's working on parametric technology to really understand is like how are you vetting the efficacy of those things? And not to mention that there's a whole lot of laws coming out that are related to waste reduction systems. Things like that in apparel, and anything that you're making in apparel that doesn't work is going to waste, and so you know there's got to be a tracking mechanism for that.

Peter Jeun Ho Tsang:

Another conversation. I like what you say there about starting from a place of truth. I like what you say there about starting from a place of truth. And whenever I meet a new founder, an early stage fashion tech startup that says they're trying to deal with fit, I always have a big, massive question mark. Because, as a startup, how do you get that data in the first place to train, as you say, your models to make it accurate and make sure that whatever you're spewing out at the other end is actually truthful? And I think, as the fashion tech I guess category gets even bigger, I think a lot of brands are probably overlooking or not even thinking about the facts that maybe the data just isn't accurate enough yet out there for it to be truthful enough that's both actually so not enough data.

Katy Schildmeyer:

and also the again, if, if you have a head of innovation that really doesn't understand, like, what a pattern drafter is doing right, yeah, and what really a technical designer is doing, which, quite frankly, in the 3d world of things, those are going to be your strongholds. You want to have more pattern makers, you want to have more 3D designers and you want to own your patterns internally as a company, not the factory. So you want to be able to own all of that data and if you don't understand those nuances of why you might want to do that, it becomes a matter of you're innovating but in the wrong direction, because you're following a trend and the lip service of those mechanisms rather than seeking the truth of the mechanism that you want really to add value to the brand with.

Peter Jeun Ho Tsang:

You must see so many infuriating things that people get so wrong all the time when it comes to fitting products. Can you share some of those areas that you do see, or some of those things that you just go? What are you doing?

Katy Schildmeyer:

So a big one is posture, which we do have another paper from the IEEE that's posture and position paper. So it reviews like why you would want to know that information. Part of fitting a garment really relates to the posture, and then also the other is understanding that fit is not a fixed point. So a good example that I always give to people is I have two sphere objects, one's an egg, one's an orange. Right, they're going to wear different clothes and the textiles are even different. Like the building materials that they wear are different, and the way the orange is, you know, covering it works and behaves is vastly different than the eggs, and so we can imagine clothing very much like the shell of the egg and the shell of the orange. It's going to be different, they're made for completely different spheres.

Katy Schildmeyer:

So humans wear different sets of spheres.

Katy Schildmeyer:

Right, my preference of a shell is different than your preference of a shell, so the fit over ease that I want in a garment might be different than the next person.

Katy Schildmeyer:

Good example is if I have a gentleman who loves to wear suits, right, like you can see him at a pity show in italy, he's got, he's just dressed to the nines beautiful tailoring, good, sporty, cut fit right, and then I on the other end, I've got snoop dog, yeah, and he's dressed, you know, obviously maybe more in a louis vuon like very different fit dynamics and preferences of that person. So fit is not just a well-made garment for the fit over ease in relation to the spherical object, right, it's also a matter of understanding that style dynamic of your brand. So and that's where things can kind of go wrong is you might have people who have different definitions of fit and you really have to get down to the. What is your root definition of that? And then let's tear that apart and build that into our brand story and how we share that on the. The website is going to really impact how we ingratiate customers into buying more or saying this this doesn't actually fit my shell.

Peter Jeun Ho Tsang:

I like it, but it doesn't fit my shell I just have snoop dogg now in my head and I just think that nothing ever fits.

Peter Jeun Ho Tsang:

Snoop dogg, whenever you see everything's like oversized or they're just like hanging off his body, just doesn't fit yeah but, um, we talked a lot about, okay, what people aren't doing right necessarily, so let's flip the coin around a little bit. So what do you think brands are getting right when it comes to fit? Of course, we talked about zolando there, but who else do you think is doing great things in the space?

Katy Schildmeyer:

it's a brand called cos cost and they're out of sweden. What they've done with pants, right, is they, they have the adjusters inside the pants, but they're well-made pants, like they're just. They're not like fast, fast fashion right there. They get closer to bridge apparel, um, and they've understood that if they build in more material to the garment you can adjust it and have it for a longer period over time, right, they also understand that, and I know that that shorter people might kill me for saying this, but when you have length built into a garment, it's far easier to shorten that garment than it is to lengthen the garment. So if you're making items too short, you're kind of adding to the waste, because you can't sell short things to a taller market and even short people may not be able to wear it if it's too short. So they get it right in the sense that they've added variables to the hemlines and things like that, so that you can adjust accordingly, which is great. So that's one brand that I really love.

Katy Schildmeyer:

And then, of course, I think Tapestry, as an overarching corporation, is really doing a lot of their own research and data on trying to utilize technology to impact the fit of garments for various brands that they have and do it in a way that's also ecology-minded. They always run with the aspect of how is waste going to be affected or impacted by anything that we use, and I think that type of gravity in innovation is really important. Honestly, I think a lot of Japanese premium denim companies do a great job and there are a lot of emerging Asian companies that are coming out that are also, I think, doing a really, really fantastic job of trying to educate their own consumers a match for North American or European consumers but they do a good job of saying we make goods for this type of body shape or this type of body shape. So they've got it down a little bit better, I think, than a lot of North American companies.

Peter Jeun Ho Tsang:

I think there's aspects of culture in there as well, because it seems like japanese designers and product um garment tech just seem to be so obsessed with fit. They just seem to be on another level when it comes to pattern cutting and those technical details that do come out at the end. And whenever I go to japan, I just buy everything, because everything looks amazing, to be honest which we don't get here in europe and, from what I see in the us, you don't get that there either. Yeah, so I completely agree with that, that there. So then let's go back to the paper, let's go back to amir as well. Why did you choose to feature such a technology within the research?

Katy Schildmeyer:

it was a mix of different types of providers, so they all do work through a mobile device right? That was one of the key criterias. Is mobile device, not the scanners, because the variance of information is different. And then it was a matter of how are they assembling the information that they capture from the mobile device to the consumer or to the brand right? And AI Mirror was different from the other three companies that we utilized and from my perspective. So when I was working on assembling the patterns that I was getting from University of Manchester because they participated in this as well was I was taking that information and applying it in the 3D program. And when I was looking at the figures, AI Mirror had various distinct, different body shapes across the board that were closer to the reality of the user than some of the other body shapes, and so from that I was able to disseminate. Not all 3D mobile app scanners are created equally, and if I am a person that has maybe a difficult body shape, I would want to know that.

Peter Jeun Ho Tsang:

Which I think goes back to your earlier comment that no shells are the same and that everyone's got a different body shape, essentially. So then, what do you think is the future of FitTech? Do you think we'll ever get to an industry consensus or industry standards to how we should be approaching the notion of fit?

Katy Schildmeyer:

Well, I think that, yes, eventually that will come to an area of happiness with fit. I don't think that we're ever going to get the 100% just because our bodies are ever changing. Also, we struggle with, also body dysmorphia of the consumer, so there's some variables in that that we're just never going to be able to again psychologically. So it really comes down to the truth of the math and how you work with the math and, again, the technology's ability to deliver the truth of the human, and I think that, yes, we will be there. It's just a matter of dealing with the technology, the changing of the technology. Ai does a huge amount of that work that typically processors had to do, and so that will make a big difference because, as you know, I'm sure, with other technologies, they may struggle with the size of some of the data related to mobile devices, and that's always a variable between mobile devices as well. So interoperability isn't always friendly, but I do think we'll get to a unified place where we'll have to have that in order to make many systems work.

Peter Jeun Ho Tsang:

That sentence. There, the truth of the math. I think my husband would be very happy to say that he's a mathematician, so he would completely agree with you there, the truth of the math. So I just want to finish off this conversation with a quick fire round of questions. So the first answer that comes to your head Are you ready, Katie Fashion, Should it fit like a glove or fit to comfort?

Katy Schildmeyer:

Fit preference.

Peter Jeun Ho Tsang:

Humanite or actual humans. Actual humans, engineering or fashioneering.

Katy Schildmeyer:

I would say fashioneering.

Peter Jeun Ho Tsang:

Top vital skill to make it as a fashion technologist Learn, always learn, especially when it comes to fit, and humans always learn. I'll finally and learn, yeah and finally.

Katy Schildmeyer:

best tip for wannabe fashion tech entrepreneurs don't let your ego get in the way of the truth thank you so much for your time, katie yeah, you're welcome.

People on this episode

Head Shot

Peter Jeun Ho Tsang

Host