Is 3D Dead? The Truth About Digital Product Creation | PI Apparel

Thanks everyone for coming. I think first I want to say we would like this to be as interactive as possible. So if you have any questions, either put your hands up and talk about it or send it through the app that I think then Alina might be able to tell us questions about. So in general, this panel is coming at a time, I think, of great uncertainty and great change in the industry. And in a way, it's a bit worrisome that starting last year, people are saying, is 3D dead? And there's even a panel, I think, tomorrow about whether 3D is really dead or not, and there's a lot of concern. So I think our panel is, if I might say, based on the premise that it is not dead, but you need to take the whole journey into account. That is, it's not just 3D for DPC or for virtual showrooms, that the big value is if you get all of this together. And that is, in fact, a playbook called Industry 4.0 that has been very successful in other industries. And the key point is that you need to get the return on investment on these assets throughout the journey. And we can't just expect that there's gonna be a shortcut, some magic pixie dust. You have to do a lot of things right. So there's no magic, but there is a science and there is an art of how to do it. And we're going to explore that in this panel. So maybe we'll just start with just a round of introductions. And we kind of organized it as a— so there is people from different parts of the journey with more experience. Everyone is quite experienced here. So I'm going to just ask the panelists to say just a couple of words about yourself and your experience with this. And maybe just in terms of the pipeline, I would start with Raj, and basically would tell us a bit more about the early DPC part of the journey.

All right. So my name is Raj—

sorry.

My name is Raj Rieu. I am the senior director of technical design at Athleta. I work under one of the Gap brands. I've been in DPC around 10 years. And so I think introduction, just, I started off definitely as a non-believer. I've come up through the technical design world, which is obviously very realistic. We work in precision and numbers. And through my experience in developing digital product, I've, you know, found the way to work within the industry and find some real advantage to it. So I was under Gap. I was situated in the Hong Kong office, and we kicked off a lengthy tool selection process for our brands. And I set up a fabric digitization team. sorry, a fabric digitization team in the office over there. So we built a very accurate way of digitizing our fabrics and also realized since our team was very small that we would accredit our vendors and mills and get them into a consistent fabric digitization process to help us. We then also built 3D standards. So we rolled those out, onboarded about 80 different—

80 vendors.

And of course, our vendor community were long time, very experienced in 3D. I then moved to San Francisco and started working in Athleta. And what we did there is we had all been sort of dabbling and, you know, in sort of many POCs, and we scaled the process to 100% of our product with 3D. With the exception of seamless and accessories. So everything else we were building in DPC. And so, you know, with that, we were getting our 3D protos coming in. And I think the one thing— so design and tech design were thrilled. Suddenly we're seeing the digital twin. We're able to fit. We were able to iterate and develop the product. But there was a missing element to this. And this was really the merchandising team was not understanding the 3D product. And so this is where the wheels started to fall off the bus. And what was missing was really that alignment in the process, right? So, you know, if you're putting this much effort, you're getting— no matter who's doing the work, whether it's your in-house teams or it's your vendors, you've got exertion of many hours of work, right? You still have to build patterns. You still have to build the digital assets. And so there's a point in time where there's a level of accuracy that's needed and something something that is more creative and more accessible for design to really sell those concepts to the merchandisers.

No, thank you very much. And so maybe we'll just do another— just quick round of other introductions to get overall just to— okay, you know, and we'll come back to doing deep dives on some of the issues. Maybe the next in the pipeline would be Richard. Who has experience with all aspects of the whole product lifecycle management systems.

Yeah. Sure. Thanks, Dinesh. Great to be here. Always a pleasure to be at PI because I also find that the audience is incredibly well-informed. So always great to talk to people who are well-versed in the subject. So my name is Richa Agarwal. I have been in the industry for about 25 years. I have seen— I've been involved in the DPC journey ever since the inception of DPC Technologies. And obviously the DPC ecosystem has really built up over the years. I have advised a whole bunch of companies, some of them are in this room right now. I've also managed technology portfolios for very large brands and really also helped companies scale DPC. Some of the largest brands in the world scale their DPC programs. So very, very familiar with the entire DPC ecosystem. And I will, after we go through the introductions, I will pick off where Roz left off around the very key thing that she mentioned around how DPC adoption has often stalled with the merchandising folks. And I'm not obviously blaming anybody. There's probably some merchandisers in the room here, but that often becomes a very, very key point where DPC journeys sort of gets stalled in merchandising adoption. So I'll talk about that in a second.

Great. Thank you. And Ermina?

Hi, everybody. I'm Ermina Avdisha. First time here in PIA New York. I usually go to the European one. And I've started with DPC 10 years ago, actually. But in fashion, I started working as a designer 15, 16 years ago. Yeah, I work in large retail brands and sportswear brands, both American and European. And I started as a 3D designer. Then my journey was until 3D platform owner, the last one in Mango, and now as a digital fashion transformation consultant. What I learned and what I did through my past jobs very quickly, it was starting to utilizing the tools as a designer and introducing those tools within all the offices in the company and actually building real programs where you can unlock the value of, of technology or DPC tools throughout the whole value chain. This is my experience.

Great, thank you very much. I think as you can see, there's the the whole spectrum here. Just a little bit about myself. I'm Dinesh Pai. I'm actually— I've been working in 3D and AI for— maybe I shouldn't say exactly how many years— many decades as a university professor. And about 10 years ago, I founded a company called Vital Mechanics to bring some of that technology from computer science. I'm a computer science professor to the fashion industry. Specifically, we were the first to introduce soft avatars in this industry based on measurements of the human body, which I'm very grateful that 3DRC recognized us with a couple of Grand Challenge Awards in the past. But I'm really interested in bringing the digital technology to the fashion industry. And as I mentioned earlier, it isn't just about some 3D or AI. In fact, I think it's— I would like to— you to think about that it's not 3D, it is HD. And in fact, there are 3 HDs that you should think about when you go back. One is the human dimension. And that's a remarkable aspect that is ignored by quite a bit of the industry because, you know, you're not making tablecloths or curtains on the— you're making things that go on the human body and you ignore that at your peril. The second HD is high definition, that is, it's not enough to have a pretty picture that— or something that will go in the virtual showroom at one end or just for early design approval or something, what you want is to capture the details. You have to— it requires a bit more investment in time. But once you get this high-definition garment, then it becomes a true digital twin. And that means that you can use it throughout this process, this whole retail journey. And the last HD to take away is the huge data opportunity that I think Richard and the others will talk about as well. Because there is, I think Richard calls it exhaust data, something like, there's so much of data byproduct that is coming from your returned data, from all the way from your product initial creation, as well as these new technologies allow you to generate huge amounts of data. So right now, for example, we can run 10,000 fit in parallel in the cloud so you can generate realistic variations and you can capture the data and use that. So basically, I think it's what we want to explore without minimizing the challenges is, you know, what are the hard things that we need to do? What is difficult? What's the challenge to make things work and see how that goes? So maybe I'll start again with Richa because I think Roz made a beautiful segue into one of the big blockers that we are hearing, which is the merchandising end of things. Do you want to do it?

Sure, sure. I'll talk about the merchandising in a second, but I like, I do like how you framed it because it's that triple HD opportunity. And from some of the presentations I've seen this morning, especially, I don't know if a lot of you saw this morning, there was a panel around some of the startups that work here who are working on the different aspects of the value chain as far as DPC value chain is concerned. So, you know, often the digital asset is treated as a digital asset for a specific use case, but that digital asset doesn't travel through the entire value chain. We need to reframe this conversation because we need to start creating digital assets that are high definition and high fidelity from the get-go. And we mature these digital assets through the value chain. So you, you start with the digital asset that is created for product development purposes or concepting purposes, and then you take that digital asset all the way to the marketing purposes. Now along the way, there is a whole bunch of what I call, as Dinesh pointed out, Thank you for mentioning my blog, Dinesh. But I recently wrote an article about this. So there is a lot of data that is created along the way of the digital asset journey, right? Right now, most companies and most systems are not leveraging this data. There's a huge bunch of metadata that's being generated. For example, when you start to digitize fabrics, there's a lot of metadata that's created. And that metadata is kind of just sitting with the digital asset, maybe in a DAM if your company has a DAM, maybe not. You need to start connecting that metadata into your PLM systems. Because if you don't, you are only leveraging the digital asset as a pretty picture, which is either used for your merchandising assortment or it's used for your B2B or your B2C marketing purposes, but using this metadata or using this exhaust data, now that we have proliferation of AI tools in a lot of companies, you can really leverage this rich data to do all kinds of things. So getting back to merchandising adoption, right? So often, a lot of times where I've seen DPC journeys stall is with merchandisers. The reason the merchandisers, in my perspective, aren't really adopting DPC the way that, you know, tech design and design teams have adopted DPC is because they just see that as a pretty picture. And, you know, they are missing the touch and feel factor. So they want to touch and feel all of the samples. Now, if you could leverage all of this metadata that is being generated through AI and offer them as intelligence to your merch teams, it will make your DPC adoption a lot easier. The other thing we also need to think about is how do you leverage the historical sales data and provide that at the point of assortment, right? So a lot of times you've got a lot of return data. That return data never makes it back all the way to your PLM systems and your designers and your tech designers and your merchandisers are generating, are assorting product, creating product without connecting the dots. You need to connect the dots and I think when you start offering this rich data to your merchandise teams, it will make the adoption of your DPC journey a lot easier. So I'll stop right there. So I can pass it on.

No, that's great. Thank you very much. It— maybe just to complete, like, the merchandising cycle, it would be very helpful to know, you know, in the, in the virtual showrooms, which are almost the end of the journey, what are the challenges and how can we get the feedback? I think one of the actually the maybe underappreciated things is that once you digitize the whole process, it becomes more of a two-way street. It's not a waterfall, not a pipeline going from, you know, DPC to the showroom, but it can be very fast feedback channel. So maybe you can say a little bit about that.

Yeah, okay. Yeah, for me, like, for the virtual showrooms I had, like, I ran off a proof of concept last year in Mango. Where we were trying to use 3D assets from, of course, of 3M, well, digital product creation from the collection to use it for the virtual showroom. But the idea was basically to shorten the go-to-market by 2 weeks. So we were trying to get those assets digitally to sell earlier in the process and make the early decisions from the sales team. Actually, the idea was every 2 months it was to create the— to have the physical showrooms. But this time, the idea was to have portion of this virtually and the rest physical in order to figure out if it actually is something that works, that helps everybody. So how we did it was— as everybody, I think, imagined can be done. We created an avatar. But the first avatar, it was a little bit— like, it was real-looking avatar. But it was a little bit taller and slimmer than actually our product development avatars. So when we started then to actually create the garments for this avatar, once everything was decided from design and buying team, so every— the trims, materials, and the patterns, we actually need to adjust them on this avatar. Because the idea was actually to showcase the collection as it was a real outfit and a real model. So this was the first challenge. And it took a lot of time, actually. It was really a slow process, very manually. Even though we had IT department working with us in order to automate certain processes with coded solutions. Still, it was like a very slow process because we needed to as well go to approval stages with styling department, with visual merchandising department, with everybody. And the goal of it was in the end creating still images and video, 360 view videos in order to be shown in the B2B platform. So it wasn't actually a virtual showroom, but kind of preparing to that direction. We decided in this way because it was more important for the clients to have way more high fidelity looking of the garments versus actually being like an immersive space where to look to the collection and buying in. So yeah. We went all in with this. We actually achieved the parts of the collection digitally. And when it came for, well, the wholesale clients in the showroom, they actually wasn't expecting it like digitally. They weren't prepared. So they weren't actually prepared to take decisions without actually seeing the real product because usually they are used to touch and feel the product, especially if they're like new styles. So even though we achieve digitally to represent the best-looking avatars and with the garments and really had— we had the feel of the garment, how like truly almost 100% real. We couldn't achieve the fact that people wanted still to touch and feel the products. So that was a problem. We also did actually another— we actually also ran another test with AI try-ons. So instead of having 360 videos, but at least AI virtual try-ons with the AI avatars in order to to mix and match 3D garments and physical samples, you know, to achieve like a full range of outfits. But the styling wasn't replace— reliable on that time yet. So in the end, we couldn't style the garments in AI and 3D the way how it works in the reality.

Right. No, that's great. And I think there seem to be some recurring themes and where maybe one of the threads underlying this is trust.

Trust.

That is that, you know, again, you are— most of you at least are building physical products that go on humans, and how do you trust that what this is showing you is real? And it is— it's, it's good to be skeptical. And also, I was just thinking that even though you said Atleta doesn't do, you know, virtual showrooms, but you do a lot of e-commerce, and I think the same issue is there for e-commerce returns too, because people don't trust trust what they see on the website or the recommendation that's being given. So maybe the question I could throw out is, you know, how do we get trust? And maybe a second part to that is if you ask someone, do you want to touch and feel the real thing and put it on? Of course, the answer is yes. They'll be, you know, I might as well try it. So what they don't appreciate is there is a cost. There is a price to be paid for that. And so there's a trade-off. How do we articulate that trade-off? And what is the value you get by missing touching it? How do we show people— I mean, if we can't convince people there's any value, then why bother? So how do we do that? I'm just throwing it out open.

So you're talking about two different use cases here. One is touch and feel in the B2B space, and then there's one use case in the B2C space. And they're very different answers based on— so which one are you trying to address?

Sort of both in the sense that I think that there are different, but there is a common thing that is you want to assess the actual product by touching and feeling it.

Yeah.

So—

Yeah, I mean, obviously, you know, there is no substitute for touching and feeling. But what is the value add that you're offering to, for example, an e-com customer, right? So a customer goes to your website, looks at your product, and obviously can't touch and feel. And obviously e-com has its own huge benefits. It has opened— I mean, this is obviously very self-explanatory, but e-com has opened this massive sales opportunity for customers that didn't exist prior to e-com. So customers can order samples from the comfort of their home. Now, how do you increase trust in that asset? Again, let's go back to how the digital asset journey starts. A lot of what I have found, a lot of companies will digitize assets just for the e-com purposes. And I am seeing a proliferation of DPC tools which will create this virtual try-on, which really isn't a virtual try-on. It's literally just the representation of the garment that you can superimpose on a customer. That is it. That's not virtual try-on, right? And we all know that in the e-com space, there is a massive amount of returns. Like every single company, you know, in this conference right now is grappling with e-com returns because it is such a huge problem. The way that you will increase trust in digital assets is by creating high-fidelity assets that start with the use case of you are creating a true digital asset that is wearable, that can be scaled up and down, and you're using demographic data to scale those assets. And when you're grading your 3D assets, it is actually using your real-size scale, and it's using your digital— your demographic data. And that's how you will increase. Now, that comes— I don't want to get into GDPR situation, but— But that comes with, you know, a lot of privacy and sort of things which we don't have time for, but that's one of the answers.

I think so. I don't know if that is our warning time or we should wrap up relatively quickly, but maybe is there something in the e-com and merchandising space that I think from—

Yeah, so I'm going to answer your question a little differently because I think Trust means something different to different parties, right, depending on what they're really after. So, um, so for example, if 3D— we were trusting 3D but merchandising wasn't— then you need a different tool to get them to trust. And I think that's where the integration of AI comes in, right? Because they— the merchandisers really needed a storytelling piece. They want to do They wanted to see the full outfit. They wanted to see it in a realistic way that they could imagine a customer in it, right? So it was much more customer-centric.

[Speaker:GUY_SPIER] Right.

[Speaker:CHLOE_ZUCKERBERG] And so I think there are different tools. And we've sort of arrived at a point where we need a toolbox to get from A to B, you know, and then the final fit approval. But I think the goal is really setting your sights on What is the goal at the end of the line? And ours was really to improve the fit approval, like the pencils-down moment when we got to the physical sample at line adoption. And so building that process and putting the right tools in place at the right milestones is really where we've seen success.

So, I don't know, I think we are kind of out of time, but I think, sorry to— But if there are time for questions from the audience, something like that. So are there questions on the app or? In Slido? Okay, great. So, okay, good. I don't have to look back. So we have several questions. I don't know how much time we have. So what— so most of the talk in the industry revolves around digitizing established materials. For developing new materials from scratch. I think that's a whole new thing. I think several companies are working on it, but you do know— do you have something to add to that?

Sure. I don't want to dominate the stage, but I'll take this very, very quickly. So there are already companies that are working on this use case. Have industry solutions on this use case. You can— what metrics are needed to develop materials from scratch digitally? All of the same metrics that you need when you are actually digitizing the product, which is, you know, the weft, the warp, the weight, the stretch, all of the variables that go into creating an actual textile. And because now we have so much data, a lot of the fabric digitization companies have so much data in place, you can now start to develop materials virtually without ever having to actually physically create this material. You can even test these materials digitally. So I'll leave it there.

So maybe we can jump to this last question because it's an interesting one. Okay, maybe you'd last absolutely last question, which is, you know, it's a version of how do you get someone to understand something when their job depends on their not understanding it? So how, you know, how do you deal with this threat? But maybe we are totally out of time. We're gonna stop. But, you know, hey, we are here all night, so come and see us and talk to us. Thank you very much.

Thank you very much.