Fashion Technology Live Demo | Black Swan Textiles | PI Apparel

Today we're going to talk about digitalization and mediocrity, probably not two words you would normally hear at a conference like this. I've worked in my career for various brands, starting with Ralph Lauren, Target, Fruit of the Loom— I was the grapes— JCPenney, and Under Armour. In 2017, I started Black Swan. So first, we're going to talk about Fabric. This was intended to be a live demonstration of two products, but due to some technical issues with systems and stuff, this is actually going to be close to a live demo, but more with screenshots. But it represents the technology that I want to showcase. So first, PLM systems. There's some PLM providers here. They love their product. I think it stands for Pedelec Mad because it's probably one of the most hated systems I've come across. However, the intention is great because it is intended to collect all the data throughout the process from concept to production. But fabric tables typically aren't very useful in these systems. They look something like a glorified spreadsheet with some data that's useful, other data that's— not really useful. It doesn't provide a standard definition, so there's no simple way to determine if fabric from supplier A is the same as fabric from supplier B because they're identified by article numbers. There may be a master part number in the system, but most times, even that's wrong. When we look at fabric at Black Swan, we see 6 attributes that need to be defined digitally or objectively. First, the way it looks, how it's made, the construction, the hand feel. Since everything is going digital, we're seeing fabrics on screen, but hand is probably the second most important aesthetic property after color of any fabric, whether it's a designer or a customer looking at this stuff. The provenance, where that fabric came from, increasingly important, especially with digital passports and all that kind of stuff. Performance, how does that fabric stand up to standard test methods that are associated with end use? So is this fabric suitable for yoga pants? Is it suitable for football jerseys, whatever? And that's a function of those test results. Feasibility, how difficult or simple is that fabric to make? That in itself can be a problem. If you've selected a fabric that is slow to knit or produces a lot of defects per square or per 100 meters, something like that. So we're looking at how do we get digital data, objective data around all these things. This was our start, the BFD, Black Fox, Black Ops Fabric Decoder. My name's Keith Hoover. I'm president of Black Swan Textiles. Black Swan Textiles focuses on digitizing fabric. In the apparel industry, we're a little behind. Where everyone else, they've got robots, they've got sensors, IoT, not so in the apparel industry.

The current way to analyze fabric, a potential customer will send a swatch of fabric to the knitting mill. The technician will physically pull this and check each individual stitch and then translate what the knit, tuck, and miss stitches are over to a CAM layout for the mechanical settings on the machine. You'll also need to punch the piece of fabric to figure out the weight, and then you'll also have to analyze the fabric for the yarn content. Once we have all of that, then it comes time to put it on the machine and start your sampling.

The whole process is very costly, really inefficient in a mill, and has a moderate success rate.

It's a long time.

Weeks and months, a year of fabric development. Speed to market is really key.

What this program potentially can do is eliminate eliminate a lot of the time and expense out of that by digitizing the parameters of these knit styles.

It's going to benefit anybody who wants to quickly identify materials.

Black Swan Textiles created for us one of the very first images of what digital fabric actually looks like down to the yarn level and gave us the first digital transformation in the form of digital fabric attributes.

There are various aspects of fabric that important from both a design and a manufacturing point of view. What the fabric looks like, how it's made, the way it feels, the draping properties or the physics of the fabric, how it performs in a given situation, and finally how, how difficult or simple it is to make. So these digital properties help us define the end use at the beginning and then help us drive manufacturing at the end.

We have an active library of over 3,000, 4,000 styles that we've produced over the years. Customers are looking for very specific fabrics, and they're usually really subjective components that they're looking for, right? With the Black Swan program, basically we would be able to plug in objective components to those requests and measure them against our existing styles and not have to even create new product.

What the consumer wants is individual personalization. That personalization that we can create by having a digital platform affecting radical change at the commercialization level of textile sewn products.

The early stage of any innovation is the original idea is going to morph into multiple aspects. I could easily see this be used on the quality side to scan fabrics to see if they're coming off the way they're supposed to. It's about getting it to deployed and seeing where the end user takes it. I do believe Keith is on to something, but that something is going to evolve, I think, and that's exciting.

That's cool.

So the important part about the digitization of the fabric is once you create this process once for a particular fabric, that digital component can be programmed in for the next set of fabrics. So you don't have to redo this project. Once you get the digital settings on a fabric, you can apply to any fabric you want. So that preemptive work on digitizing the fabric is absolutely enormous.

The big advantage here is speed. If we can shorten the time it takes to evaluate and digitally offer a solution to someone versus this whole other really expensive and tedious process, that's the big one. And then obviously for, for mills like us to be able to somehow digitalize our library and monetize something that we've put hundreds of thousands or maybe millions of dollars into over the years that's kind of sitting there untapped, it's tremendous.

When I first saw it, I was like, wow, that speeds things up. I wish we had that. Uh, how much is it, right? How much is it going to cost me? And do my competitors have it? Because I want it for myself. Those are the 3 things I said when I saw it.

This project establishes the fact that the textile industry can be driven as a digital platform.

There's a whole opportunity to reinvigorate this industry. It enables us to make in the USA, but more importantly, rather than where we make, it's how we make it that we're addressing. And I think that's the fascinating part of this whole thing. So all of this stuff is basically You take a piece of fabric, you scan it, you run it through some algorithms, it tells you how to make it. That's nice. But the real value, like Keith Dartley said, is how can I access my fabric library at the mill of 5,000 to 10,000 fabrics? Because that's the big thing. And that's what I'm going to show you here. This is the visual browser from the Arsenal. You can see in this one, there's 9,816 fabrics loaded in this library. In the center part of the screen, these are all the fabrics. Over on the right, there are filters that you can use to navigate through this huge library of fabrics. So there's one thing you can do. I can apply— I can do a color picker, and I can apply color to everything. Kind of a bell and whistle. I can also use tags that mills typically assign to fabrics, like there you can choose mesh, and when you click it, it shows you all the fabrics that have been assigned mesh as a tag. But this mesh is kind of an ambiguous term. It means different things at different mills. So we've applied, derived something called visual cues from AI so that we can assign aesthetic or graphical components to a fabric that reflect any graphical images of the fabric, structures, lines, that kind of stuff. So we're going to take the mesh and we're going to apply— which of these look like a checkerboard, you know, a 90, 180-degree layout? And these two do. And then we're going to go back and say, "Okay, skew that 45 degrees." And these are a grid, but they're offset by 45 degrees. So already, we can start to We can assign these tags to fabrics using technology instead of someone sitting through adding tags. We can also— we can look at the different types of colored yarn. The solid color, in this case we're going to show only fabrics with heather yarns. So we can quickly, if we're interested in sourcing heathers, We can only look at heathers, and then we can apply some of these other filters as well. We can also look at slub yarns. So in this case, we could choose slubs, and these fabrics have the irregular yarn construction in it. This one, we're going to look at— in the front of these fabrics, typically what you'd call, like, 1x1 rib, we're just calling columns, because everybody knows what a spreadsheet columns and rows is. So we're gonna show all fabrics that have 3 whales that make up a column. And then we're going to apply this other slider to show only the ribs that have us— that are prominent, that are really strong patterns, so we can further filter this thing down. We can then look at the— of that fabric, what's the backside of that fabric look like of these fabrics that have a front side with 3 ribs? And you can see it starts to get interesting when you start looking at the backsides of fabrics. So now we're gonna look at fabric, the backsides of fabrics. And typically in single knits, if you look at the backside of a jersey, it looks kinda crappy. You'd never put that in the front of a jersey. But based on the complexity of stitches, we've come up with a slider called complexity. If it's set at 50 cent or below, then we can look at fabrics with interesting patterns. And we can look at fabrics that have rows and column structure. And we can look at fabrics that have— we can start— these are fabrics that have a diamond shape in them, but not a hexagon shape. And then we can reset it in both diamond and hexagon. So the user can explore the library and start seeing some cool fabrics. And now we're seeing the back view with— these fabrics all have prominent columns. And these have prominent row structures. So it allows you to see stuff that you couldn't do before. Now we can look at this particular fabric has a zigzag pattern. Let's take a deeper dive. So this fabric, you can see the front and back. You can see all the colors that have been applied, run in production on it. You can see this thing called buckets. It's the predictive features about this fabric, whether it's durable, what end uses it's appropriate for. We can look at the details of the construction of the yarn. We're measuring the haptic properties of the hand feel. So we can know exactly where this plots among these three attributes of hand feel. Then we can look at production lots and all the testing that's been done on every production lot by specific test methods. So we're getting a much better view of how this fabric is going to be in the product. And then the structure that we showed in BFD, this is actually— this is the key to this fabric. This is how to set up a machine with tuck and knit stitches to make that fabric. So everything's here. It's truly a digital twin of the fabric because it's not just the appearance, but it's all the behavior as well. Now, we can also go into fabrics, and we can click on this Show Similar Fabrics with the Same Visual Cues. When you click that, then it shows here's 6 fabrics with identical visual cues. And then I can go back and show me all the fabrics that feel the same or have similar haptic properties, similar hand feel. And across all constructions, it pulls these fabrics out. So this would enable a brand to have a signature hand feel, like some brands have signature colors. When a customer interacts with that fabric, they always get that same sensation. We can filter by end use properties, by— we can do a grid search, we can punch those stitch counts in and find the fabrics. In this case, We found the one fabric that meets that definition. And then we can do typical issues that a sourcing person would do to weight or fiber content, that kind of stuff to help them navigate a library. Now we get into design. We're done with fabric. And true design versus 3D, which is kind of drinking your own spit when you think about it. You're taking existing patterns and you're just repurposing them. Not in a nice way, but I do have— I mean, I started my career as an illustrator back when digital art were these digits instead of 0s and 1s. So I drew stuff for brands. But 3D design is not a design tool. It wasn't designed that way. In fact, if you go back and look at the Eurotex flagship product, this was like 2005. This is a blueprint for how the digital industry evolved. And they were trying to figure out how to do everything digitally from designing to production. And Chapter 4 was the blueprint for how every 3D product today is built. This is a platform, these are the assumptions. And look up Yaron Goldstein. He worked for a company you may be familiar with, but this is basically— This is the thinking about how this product was developed. They said that this is for pattern makers and technical designers after a design is approved. And they wanted to make a virtual sample instead of a prototype. So this was meant to prototype something that was design approved. It's an engineering product. It's not lightweight product. It requires specific knowledge. And they said at that time that the more intuitive approach of sketching to create 3D design, it seems appealing, well, because that's how people work. But they didn't think that's the way the industry worked, at least not at factories, so they didn't pursue that, nor was that technology available in 2005. But they acknowledged at the time that there was a better solution for designing product. So, NuArc is a— It's a startup out of UK. I came across them. I'm not working for them, but I thought it was a cool product. So they're basically— they take— a designer takes a sketch that the designer draws. And then they can basically color it up. So they upload the sketch. They write a simple description. So if you took English in college, it probably comes in handy here. It generates 4 to 8 samples in 20 seconds. And then you can edit each one to do more variations. And then you get the high-res PNG file. So what I'm going to do is show you some examples that I created. Took this sketch, wrote this prompt, 20 seconds, I get that. Oh, this sketch of a white gauze dress, boom. This, you can change the description of the fabric if you wanted to make it look flannel. But in this case, it's a nice silk dress. A puffer. This— all this stuff is available in seconds. So you, as a designer, can sit through and just do iteration after iteration after iteration using this tool and your own drawing before you commit to a design and then go to a product like Clothing tech where you actually build it in 3D. So it's kind of fascinating. It even understood what black watch plaid was. I was surprised that it did. There was one thing I found. The artist here tried to make drapey lines. And the AI thought it was stitch patterns in the woven, in the knit. So it basically, you have to be careful what you draw because it's going to render what you draw. This jacket, not bad. I mean, if you're a designer and you're presenting these ideas to a buyer to get your design in the line, I think you're going to be much better with the image on the right than the image on the left. And it literally takes slightly longer. This sketch was drawn by a friend of mine at Ralph Lauren way back when. And she's a little anorexic, but not bad. I drew this in 1992, and he looked pretty cool. This guy looks kind of goofy here, but looks great. Shoes, just by putting alligator in there, boom. You can take a sketch, you can make it a dress shoe, or you can make it a casual shoe. You can even, make variations like it's stepping in dog shit. But not that— it had trouble finding out that one, but it did come up with brown splotches, so. Footwear, you know, I was actually surprised it figured out some of these structures. And even flat goofy socks. So design is a missing link in everything that's been going on over the past 20 years. We've been talking about product development, we've been talking about manufacturing, But we've kind of ignored design. And design is what makes or breaks a product, in my view. So if you look at this, you've got the digital stuff on the left, you've got production on the right. Typically, we started with 3D, and then we've basically connected either with DXF files that goes to a marker, color, the spectral data that goes to dyeing, And now BFD will enable knitting. But if we start with a visualization tool like Newark or Straight Lines out there, they're working on a similar product. There's probably many others. But this is the— this is the next thing that's going to drive design forward, much to the consternation of a lot of the 3D companies that have been pumping this stuff out for 20 years to a low adoption rate. So I think it's just a matter of technology catching up with what we really need to do in apparel. And that way, we connect the first concept design with the final product. And that's it.

Wow.

So we've got some time for questions or snarky remarks or whatever you want to do. I like how you add humor to your presentation.

Well, otherwise—

Well, sometimes I just— you got to say, I mean, I know what people think. I mean, I've worked in these jobs and, you know, there are certain companies, if they ask me what I hated most about my job, I'd probably say PLM. But you'll never say that out loud because the company's spent a lot of money on PLM systems. So I think it's just a matter of Supporting what we do, which is— there's a question. I'm wearing a white shirt. Why is my shirt white? And a dyer would tell you because we didn't add dye to it, we added optical brighteners. But that's how it's white. Somebody else might say, well, when you look at it, these signals go to your eyes, go to your brain, and your brain says white. Well, that's what white is. My shirt's white because a designer wanted a white shirt. And that's what everything is about here. Technology is our friend, but it shouldn't rule us. It should serve us.

What was the name of the design service from London?

Oh, NewArc, N-E-W-A-R-C.ai. They have a free promo. You can do like 10, upload 10 things. And it's like Well, it's $29 a month. So it's— I'm sure the price will go up. But definitely go online. And if you see other products, try them out too. There's another product called Midjourney that basically you just write a prompt. Philosophically, I'm against Midjourney because it takes the human creation out of it. Basically, it's something else. But more practically, I'm against it because— If a designer can write a prompt in Midjourney, then a merchant can write a prompt in Midjourney, and a vendor can write a prompt in Midjourney, and therefore there are no more designers. So you got to be careful what shortcuts you go for. Any other?

Yep.

When you digitalize those fabrics, [Speaker:AUDIENCE] What different file formats are you able to do? [Speaker:JAMES WHITTAKER] We're calling it a KDH file, Knit Design Hierarchy, or Keith David Hoover, actually. But all it does is it captures the stitch pattern in a unique format, the courses and wales in a field by themselves, and then yarn diameter. So it's a very simple format. We're working with 2 mills right now, one in Los Angeles, as a matter of fact, and another in Faisalabad, Pakistan to digitize their libraries. The other aspect of that program is we saw there was a presentation earlier today on traceability. Well, traceability is very difficult in our industry because there's so many factories and producers that we're in. Ralph Lauren has over 600 mills just for apparel. So if you want to know where every fabric came from, One approach is to have the vendor tell you and write it and fill in a form. The other one is just to link into their ERP systems, Enterprise Resource, whatever P stands for. But it's the systems that if I go into a mill and say, I want to make this fabric, they go and pull certain information out of their fabric so they know how to set up the machine, they know what order, how to order for yarn. So it's that system that they have that their business runs on, You just link to that and that tells you the traceability all the way back to yarn and fiber. A much simpler solution, but we'll see what happens. Anything else? Well, cool, I finished early. Thanks, everybody.