Advancing the Science of Fit with Dinesh Pai | Seamless

Hi Dinesh, thank you very, very much for joining us today. It's been a while, so lovely to see you. Let's start with a bit about you. So please, can you introduce yourself to our audience and let them know a little bit as to how you got into this industry?

Okay, yeah, thank you very much, Michael, for the invitation. It's always a pleasure to talk with PDRC. So I'm actually wearing two hats. I'm a professor at the University of British Columbia where for many years we've been doing research on simulation of soft bodies and other things like that. And about 10 years ago we spun off a company called Vital Mechanics, and Vital Mechanics has developed the first soft avatars in the industry. And actually that is something that got the 3DRC 2021 Grand Challenge as an industry for that. And then, but we continue to do some basic research at the university and applied research trying to get this technology out in Vital Mechanics.

Amazing. So yeah, you've just, you've just mentioned this, but you, under your two hats, you've actually won the Grand Challenge twice. So I believe I'm right in saying 2021 with Vital Vital Mechanics 2023 at the University of British Columbia. So can you give us— let's start with the Vital Mechanics win in 2021. Can you give us a brief overview as to what the solution and technology you submitted for that year looked like?

Yeah, absolutely. So for the technology for that was our soft avatar technology. You know, for many years people have always been saying that soft avatars are really necessary for predicting fit accurately, but it was considered too hard and too complex. And I should say it is a very hard problem. I can talk more about this later, but I think it's underestimated how hard it is. So, so we actually first approached it at the university where we decided, first of all, to see, you know, how do we actually predict how the human body deforms with clothing. And we looked around at the literature and actually there was nothing in the literature which said, you know, what are the soft tissue properties of the body and how do they vary over the body? So we actually did a basic research project where we developed a tissue measurement probe. We measured hundreds of participants in the lab and we developed a soft tissue model and also finite element methods of how do you simulate how clothing fits on soft avatars with friction and all those complications. And so we came out with this first soft avatar simulation technology that we call VitalFit. And that was what we submitted to 3DRC in 2021, as I think at that time it was— until that time it was primarily like an industry challenge. And so we won the industry grand challenge in 2021 for that, which we are really grateful because it showed that yes, you can do soft avatars for fit testing.

Yeah, amazing. Well, look, let's stay on the 2021 win for now, because I want to get, get on to how it's evolved, and then we can of course talk about the 2023. But back in 2021, what was the specific, uh, problem or challenge within the industry that you were trying to solve? You've obviously said fit, but if you can elaborate on that a little bit more.

Yeah, I think that's a, it's a, it's a big question, and I think the fundamental, I think, motivation that I, I think underlying this is we realized how wasteful the apparel industry is. And to be honest, I didn't know about this until we started working with some industry partners. And then I saw that the traditional supply chain and processes are incredibly wasteful. So, so we wanted to, to see how can we reduce waste. And there's a lot of the waste was actually due to challenges with fit. And so we wanted to bring the prediction of fit to the industry. You know, if I may, the analogy, I come from an engineering and computer science background, you know, and I was aware, for example, then in the automotive industry or in aerospace, people don't just make, you know, make planes or cars and then say, you know, does it, is it going to work? You do a lot of simulation, you know, with virtual wind tunnels and things like that, so that ahead of time you have a pretty good idea of how this product is going to work, and then only you make the physical product. So in apparel, people have for the longest time said, okay, hey, clothes are cheap, you know, let's just ship them to the store, and if people don't buy them, let's ship them to the to, to the landfill, maybe.

Yeah.

So, and that has been the mindset. And, and to me, that was a shock. And I really wanted to do— make a difference and to see, can we bring the same type of a playbook, uh, from the technology industry to apparel? And, you know, the price of computing is coming down constantly. So, and, uh, and the actually the impact of waste in the apparel industry is going up due to impact on the environment. So we felt this is the time to try to help with that.

Amazing. So if I'm right in saying this, the majority of brands, and I imagine it's still the case today, although I hope it is changing with solutions like yours, um, they use kind of like hard, non-dynamic, um, avatars, right? Whether that be, uh, for a physical creation or within digital tools as well, there is, um, there is ways in which to simulate the actual cloth itself, but a lot of the times it's being done on a rigid avatar. Is that correct?

Yes, I would say so. And I think in a way there's a whole spectrum of things, you know, as I think it was William Gibson who said, you know, the future is already here, but it's not evenly distributed. So there are some companies that are well ahead And we are working with a lot of them. And there are some companies who are even now just bringing, you know, their paper patterns into digital, right? So there was a discussion of should we go digital? And then there's a discussion, should we do it 3D? You know, and then, and then to all the way to what we like to think of it as accurate representation or digital twins. So in other words, it isn't just that the avatar is dynamic or, or, or, uh, 3D. It is that, you know, how much are you actually trying to reproduce the actual physical properties of either the garment or the avatar and how it fits? So, so yeah. So our focus has really been on that. And, um, I have to say, The industry has been slow, relatively slow in moving in that direction and fits and starts. I think a lot of it is due to the fact that it's a very big and traditional industry to some ways.

Absolutely.

So changing such a large juggernaut, you know, making a change of course takes some time.

Yeah, absolutely. Okay. Well, When you first developed the solution that you put forward in 2021, I guess, what were the biggest technical or creative challenges that you came up against?

Yeah, so I think that the challenge the whole time really is still the, you know, solving the fit prediction problem. I think that's really— we in fact have a sort of a motto that fit is fundamental. You know, that's— you should put that on a t-shirt. So fit is fundamental. And it's really, you know, people have been aware of fit all the time. Everyone knows about fit. You know, it's like, I think it was a friend of Mark Twain's who said, you know, everyone talks about the weather, but no one does anything about it. So I think there's something similar. Everyone talks about fit, but no one does anything about it. And I think it's really because fit is much harder than people appreciate. And as I said, even we appreciate— we didn't appreciate it during this time. In fact, in the early days, I just realized as a startup, you know, people are always trying to be in a kind of a sprint mode and realize this is a hard problem. You have to be more of in a marathon mode. Yeah. There's a lot of things that need to be challenged, to be taken care of. So people have wanted to solve fit, but what really is needed is a pretty deep, what I call a technology stack. So, and so you need to start at the foundation with roughly speaking what you'd call soft avatars, which is a realistic digital twin of human bodies that change shape and behave the way real humans do. And also not to forget, even though people have been doing garment simulation for a long time, you also need very good representation of the garment. And it isn't just the fabric properties. We pay a lot of attention to the construction details of the garment. The seams make a lot of difference. For example, I like to think of it roughly as like the skeleton that's going to hold up the body. So So you have to pay attention to all of that and the interaction between the two. How does friction work? How does, you know, how does sweat and lubrication affect those things? You have to think about all of those things. So the first, I think, foundation is the challenge is to get the basic soft avatar accurate digital twin, which I've I'm pretty confident now that we have a good solution that, that can predict fit realistically. And it isn't just that it is soft, that it is soft in the right way. Right. For example, just to another small detail about our 2021 contribution, it isn't just the whole— the body isn't soft uniformly everywhere. It isn't made of rubber. Right. So it's in some parts Even in people with a lot of body fat, some parts like around the hips, they're actually— the bones are close to the skin. So those parts don't deform as much as other parts. So you really need to pay attention to what is one of the most complex objects in the world, which is the human body, right?

Yeah, absolutely.

So then the next level on top of it you know, on the way to doing something like our 2023.

Uh, yeah, let's talk about that now.

Yeah, yeah, yeah. So yeah, I can, uh, maybe, uh, uh, I can talk about that maybe, and then I'll fill in the what's, what else is needed in between the two. Uh, so the, so the 3DS 2023 was, um, it's interesting, it was an invitation to both universities and, uh, I think industry about AI and how do— how can AI be used in apparel. And I know this is something that the whole industry is struggling with because, you know, absolutely everywhere, but how do you use it in a meaningful way? So, so our solution at the time was to say, okay, we can actually use AI to make pattern edits so that you can change the garment pattern so it fits, but to do that in a goal-directed way. So the idea is goal-directed pattern edits, not just editing the pattern in 2D and making that change. So I have a couple of slides, maybe I don't know how this—

Absolutely, let's give it a go.

Just 2 slides. I don't want to impose, you know, death by PowerPoint or something like that.

It's always good to have an aid when you're talking about technical things like this.

So yeah, I think, you know, but to use another cliché, a picture is worth 1024 words. So, so, so the idea is is to learn using AI to make goal-directed pattern edits. And, and by goal-directed, I mean something like this. So suppose you have a simple— this is a sports bra pattern, and it is, uh, you know, you, you run a fit with VitalFit and you find that the fit is not right. And how do we know that? Because VitalFit can predict fit attributes quantitatively. So these are things like what is the pressure, what is the ease, what's the stretch of the garment, all those kind of physical things. So these are physical attributes of fit, but how exactly that relates to fit intent is really up to the brand or the user, right? Because how do they want that to fit? But the idea is that it is much more intuitive to provide feedback to the system the way a fit model or a consumer would. That is, the consumer wouldn't say, you know, take out a quarter inch here or something like that. Yeah, much easier to say, you know, my left shoulder, it feels too tight and I have too much pressure there, or it's, you know, too loose here, I want it to be reduced. So this is what we mean by goal-directed. We want to say what's the goal of the pattern edit, not what is the change that you need to make. And the idea is that then you can just, with a simple interface, you can paint on and say, I want to reduce the pressure on the shoulder with a 3D interface. You can just paint there, reduce the pressure to some value that you want. And then our system figures out what should be the changes in the pattern I see. So you don't need to change the pattern yourself. It is, it is the system that figures it out. By the way, I should, I should mention that this is the work of my student Megha Shastri, who was at, at UBC at that time. And so, and this was based on her thesis research. So amazing. And the way this works, if I may, just to open up the hood a little bit, and we can go into more details if needed. So how does this work? And by the way, this happens almost instantaneously, right? How do you make that happen? And this is where the AI and the rest of this software stack comes in. I can talk more about it later. So what we do is ahead of time, once you have a pattern, we actually generate thousands of variations. I think in this case we've generated 1,800 variations of the, of the bra pattern, the basic bra pattern, in advance, right? And then we can run fit tests from all these in parallel. So the other piece of the stack you need is the scalability that you can generate this huge number of amount of fit data ahead of time. Okay, so once you have this huge corpus of good fit data, then we can train a machine learning model that can learn to predict the pattern changes that are needed to achieve the goal. So it's not a simple one-to-one mapping because the, you know, the way, for example, the pressure might be high here because of some change that you need in the bust area or something far away. So it isn't like, oh, it's tight here, so you need to fix something right there. Now, the problem may be far away because of all the way in which things are connected. So, so this is a complicated function, but with this large amount of data, the AI model learns to predict the fit pattern variations that you need. So that's how that works. And then what we did is we also, you know, we showed a lot of different examples. So here is the first one is showing how do you give a pressure goal. You can also look at ease specifically. This is like air gap ease, which is what you're seeing. By the way, I should have mentioned this earlier. What you're seeing in the in the slide is a false color representation of the fit attribute. So here I'm showing the ease, which is the air gap between the body and the garment. So red means large gap, blue means zero gap, right? It's okay, yeah, on the body. So first of all, by seeing these fit attributes in this type of color, in VitalFit, you can immediately see, you know, where's all the wrinkling happening or some such thing. So you see here, I don't know if you can see my cursor, I should probably— yeah, so do something like this. So what you see is that, okay, it's loose here, but what are the changes you need to make to the pattern to get that effect? So what we do is we just, we first say you can paint how much you want to reduce that by. And so you, you do that painting here, and then the system generates a pattern variation like this. And so once you have a new pattern, you can simulate it. And then this is the resulting simulation. Now, this is running a full-blown finite element simulation, whereas going— this was just what the the, you know, the user asked for. And so, yeah, so you got rid of pretty much the bagginess that you wanted by doing a, you know, not as simple but like a continuous variation in different parts of the pattern. Yeah, but we also did some validation. We had a— this is simple— like a tank top where again, we just had some extra wrinkles on one side. You paint the changes you want, you produce the desired change. And we actually built a garment and we— this was tested with actually a hard avatar, like a mannequin, just so that we can test it. We can scan it and things like that. So, yeah, so this is the idea of our 2023 3-year—

Yeah, amazing. So 2021 was very much a foundation to it, and 2023 was adding a layer of AI on top of it in order to kind of automate the adjustments that you need for better fitted clothing.

Yeah, exactly. So, so the, the 2021 built the foundation for, for VitalFit, and actually in between we also had some additional things that are part of the stack that we needed to fill up. So, and those two, the two missing pieces before then you finally get to the pattern adjustment are actually, they're also very important. I think maybe if I may, I can say a little bit more about.

Of course. Yeah.

Yeah. So we have the foundation, which is the, you know, can you simulate with finite elements accurately the fit of one garment on one body, right? But the second level is to scale, and a lot of the modern computer science technology is really about doing things at scale. And so, so we know we want 1,800 pattern variations tested with FIT. Another problem that we look at quite a bit is sizing. That is, you know, even if you have one, you know, it's your final pattern. You are not trying to sell this to the fit model. You're trying to sell this garment to a population that falls in a huge range. How do you know that it fits all these people, right? So, so the other thing is you may want to— you want to run a fit test on thousands of, of virtual avatars to see, you know, is it actually going to fit the tall person who's wearing a 36C or the shorter person who's wearing the same 36C? See garment, but they have different body shapes, they have different ages. Does it fit them all? So, uh, so the second really piece was to make this scalable, something you can run in a cloud in parallel. So you can run these, you know, thousands, 10,000 fits in parallel at the same time. The third piece is really the AI piece, and probably the AI piece is currently the one for which there is I would say the most amount of excitement and confusion.

Okay.

There's a lot of confusion because people just say AI as if it's some, it's a one thing, but it's really, it's a lot of different things. And, and it is a very exciting time. A lot of things are happening and you need to really understand what is going on to separate the noise from the signal.

In them.

Absolutely. But, but really at the heart of AI is that you need a huge amount of good data, right? And the, and the large language models just scraped that good data out of the internet because there are people, lots of speech available on the internet, and you could get that. But for a lot of problems, like including fit, fit testing, there isn't free data you can just scrape from the internet, and so you can't directly use those models. So, so the key really is then that to be able to generate this huge amount of data, either during simulation or also leveraging the actual human measurement data that we have, uh, to be able to then say, I have a huge amount of data, now I can train an AI model to do whatever I want. You know, it could do predict size recommendation in retail, so that you could use the same stack for that. And Megha's research was looking at one particular problem, which is, can we make pattern adjustments quickly without having to say, make an adjustment, run a fit, take a look, is that going the right direction? And, you know, people lose patience after, You know, I've done it 3 times. That's good enough. You know, it's time to go home and call it done. Whereas with, with, with computers, you can, you can try it, you know, 100,000 times and it's not—

to simulate.

Yeah, yeah, yeah.

It's not a problem. Okay. There are 3 things that have come to mind that I'd love to ask you. One was, so you mentioned obviously you can run all of these simulations in order to gather data? Has there been any element of using data already available from brands that have collected this kind of data, or is that not something they've been willing to share, or is the quality of the data not good enough?

It's really a very good point that you make, and I think that is a tremendous source of data that I think a lot of brands are sitting on and and, you know, and that can be, it can be used in this fashion to improve the fit of garments, to decide even on what else to, on new products to introduce, those kind of things. I think the brands are sitting on a lot of amazing data, and I think we can just help them connect the dots to say, okay, here's your data, but here's where you want to go, and this is, This is the path to get there.

Okay, and in terms of the actual category types of clothing that you've seen the most success in, so what you've shown us on there was obviously there was the sports bra and there was also the vest. Are there other particular clothing categories that you found more challenging than others when it comes to this kind of sizing? And fit?

So it's a very interesting question. And I think so, a couple of thoughts on that. First of all, you know, the, the initial traction for soft avatars has been in, in athletic apparel and in, in intimates category, which, yeah, because they are really the ones that are closest to the body and providing support and things like that. Well, now that said, the more I look at it, you know, having— taking the softness of the human body into account for fit, I think applies to everything unless it's like a, you know, a very loose poncho or something.

Yeah, yeah, yeah.

Because, you know, when we were doing these experiments, you look at very lean you know, a 20-year-old, when she wears a pair of jeans, I think the way it actually sits on your body depends on soft tissue deformation around the waist. And how a jacket fits, you know, whether it's comfortable or not depends quite a bit on the soft tissue. So first thing is that I think soft tissue models apply to a much larger— essentially to anything that goes on the human body, I think, needs to take into account that the human body, with very small, like, 1 Newton force, is going to deform it quite a bit. It is remarkably soft at low— at the initial forces. And it's kind of designed that way. So So then the other question is really about— it turns out actually that we find that wovens are in some ways harder to fit because without the— also without the extra stretch that you get sometimes. You, which makes you a little bit more forgiving of fit issues with, with knits. I think wovens can be sometimes a bit more challenging, and they also require— and a lot of that is to make sure that you have really good fabric and other garment material representations. So even though our I guess, um, Renown is a little bit more on the soft avatar part, but really we are fundamentally interested in making accurate digital twins of— and you have to have a good garment, you have to have a good body.

Yep, absolutely. Okay, and the third one I wanted to ask, and then I, I mean, there are more that, that I'm gonna, uh, ask you about the challenge itself, but, um, the third question that came to mind was in terms of how your system integrates with other technologies that are currently out there, because, you know, obviously there are these you know, I don't need to name them, but we all know who the big kind of 3D modeling technology providers are. Has there been any work done so far to try and integrate your soft avatar system with their 3D modeling system from the garment side?

You know, absolutely. I think we are, you know, first of all, very much interested in reducing the friction or the barrier to entry for people to start using these advanced fit prediction technologies. So we have been talking to and working with the 3D garment CAD companies, I can call that, so to make sure that once people have developed a product in some, their own favorite software, we can then get that into, into ours. But maybe just I should say, initially I wasn't as aware of the importance of making it easy for people to get it into our system because in fact we thought we initially had— we thought, okay, we can just represent the pattern with Adobe Illustrator as an Illustrator file and we can get that in there and we could simulate that. No, yeah, but people didn't want to convert from theirs to ours. So we have actually, for example, worked closely with Browzwear and we provide a plugin for Browzwear to integrate with ours. But I should say the key point is that we have our own representation called Vital Mechanics Garment Format or VMG. And, and that consists of 3 pieces of information. One is the pattern geometry, which has been there around for a long time. You know, even like there's DXF standards for it, how the pattern pieces are stitched together, the stitching. And the third is the materials. So these 3 pieces of information which you can get from any source So you can get that into, into a VMG format, then we can simulate it. So that's the, that's, that's really the, the main thing with, with BrowseWare. We have a, a plugin that we can do the conversion automatically and, and into VMG file, but we can actually work with anybody. In fact, we can take in geometry that is in, for example, a general not-proprietary format called USD, and you've probably heard of that. So if any company can generate USD, we can take that in and we can simulate that. We can even, for some advanced customers, we can, we can generate the garment automatically in code. You can generate it procedurally. So, so for example, like all these pattern variations that, that I've shown you, you know, that's generated in code, right? So, so we have a Python library called PyVMG that from which you can, if your company has that capability, you have the base pattern and you want to try lots of variations, you can generate our garment using that software. So, so then there's, you know, it can be just machines-to-machine communication to do that.

Amazing. Okay, um, well, now I wanted to ask if you can cast your mind back to 2021, which I know seems like about a decade ago. I mean, it was like 4, 4 years ago. It was. What made you decide to enter the 3DRC Grand Challenge back then?

Oh yeah, no, that's, you know, I'd actually been to PI Apparel a couple of times, and first of all, I was very impressed by all the things that I learned at PI Apparel and the community there, and that's how I first learned about the the 3DRC. And, uh, and I thought, uh, the 3DRC is doing human service by trying to help the retail industry that is, as I said before, you know, it's been very slow in moving. And I think they're really helping the industry, you know, move things in the, in that direction. And I, and I think it, they've been doing tremendous work, both in, you know, trying to categorize how fabrics should be measured most recently and even earlier. So, yeah, so I was aware of that, and then I saw, okay, you know, here's a— maybe we'll try to submit our technology to that. And by the way, we are in some ways outsiders to the retail industry, we are aware of technologies. So it was, it was with some, I would say, you know, trepidation that we are not the standard type of people who are working in the 3DRC. But so it was great to both submit and to also to win the 2021.

Absolutely. No, and look, we always like, like it when we get entries from companies that perhaps aren't the status quo, because, you know, that's where true, true innovation comes from, like cross-collaboration between different industries and between different fields. So it's great that you took part. What was the most valuable thing you got out of participating in the challenge?

Yeah, so I think there were a lot. First of all, by the way, just having a challenge itself helps to focus the mind a little bit, to be able to try to articulate and fill in the small, you know, missing gaps. So that was one. And just to participate in it, I think it's great. I think people should, should, should try, try that exercise, you know, even if you don't win. I think it's it's worthwhile. And the second is that I think having one, it provided for people like us who are outsiders in some ways a certain degree of validation that, you know, that we weren't just thinking that, hey, this would be useful for the industry, but at least some people in the industry think that this is, this is useful, that we are on the right track. And so that was great. That was really fantastic.

Amazing. Yeah. Sorry. Go ahead.

Go ahead. No, no. Also, I should say, and last but not least, I think it was great to be able to present the results at the 3DRC, though the first one was still during COVID a little bit, just at the end.

Online.

And I think so.

Yeah.

It is not— it is slightly challenging, but yeah. Yeah, but it was great to talk to the people, show what we've done.

Amazing. And through, through the actual presentation of, or through, through the win and the presentation at the events, have you been able to connect with brands that you've been able to collaborate with at all?

Yes, I think it was, uh, I think being able to get this exposure was really helpful in raising awareness. And many people reached out to us after the presentation and some are still like in the kicking the tires, you know, there's we'll do it in the future type of thing. And some people are say this, we've been waiting for this for a long time and let's get it done. Yeah. So that's amazing. Yeah.

Well, I know again, I'm not gonna mention any names 'cause I don't want to put them on the spot, but I know you being in British Columbia, there's a pretty big brand that is based there that I think if you're not, not working with yet, or if they're not working with you, they definitely should be. But again, I won't mention any names.

Actually, no, I think you can mention in the sense that I should say the brand, of course, is Lululemon. And I think I can mention the following, which is that I think I would credit Lululemon as actually helping us but direct pointing us in this direction. When the company started, Vital Mechanics started, we were actually first not sure exactly what area to focus on. We had this soft body technology, but we were actually, we initially used it in, it was used in visual effects. We did some special effects for actually a Harry Potter prequel, Fantastic Beasts and Where to Find Them. The, the, all the creatures there and things like that, or some of the creatures there. But, uh, and I can talk about this because, uh, I'm not violating an NDA because, uh, Josh Young, who, uh, I think who you know, is also very active in, uh, in Israeli industry, visionary, presented our early work that we did on a project with Little Lemon at PI Apparel in New York.

So amazing, I didn't know that.

Yeah, so, so basically we were working, we had the soft body technology, and, and then we did an R&D project with Lululemon, which basically opened my eyes and said, this is an industry that we can really make an impact in. By impact, I don't mean like just a commercial impact, but just to be able to help transform the industry is— it was my hope. And so, yeah, and so, and since Josh presented some of the results that we did with Lululemon, at PI Apparel, I can tell you about it. And they, you know, they, they, and they've been great. And, and I think in general, the athletic apparel companies are definitely much more in the future. The future has arrived more in athletic apparel than in some of the other sectors.

Absolutely, it's getting there, but yeah, I completely agree. Okay, and then the other kind of last thing that I wanted to ask you is, looking back, what advice would you give to other companies, be them a technology startup, a research institute, or whoever they might be? What advice would you give them when it comes to them submitting for the Grand challenge. You know, we are continuing now in 2025, we have no plans to stop at any time soon. So any particular advice that you'd have for them?

Um, so I, I would say, first of all, I think for startups in general and, and, uh, industry in general, and I, I think I still believe very strongly that the apparel industry is an area where you can make a big difference in terms of the digital transformation. In some ways, for better or worse, it is one of the last industries to have undergone a, like, a complete digital transformation in the sense that the, this, the whole supply chain in the industry is, is digitized. You're not moving physical things around. You, you have, you know, like this Industry 4.0 vision of having digital twins that you can, can carry through the system as much as possible until the, the final product needs to be made. So, so I think there's tremendous basic opportunities there, and I think it's, uh, it's, it's worth investing time and energy into it. Uh, the downside, in some sense, the flip side of Why it is one of the last industries to transform is that it is a very complex industry with supply chains that go, you know, all the way from where, you know, fabrics are made to where they're designed, where they're manufactured, where they're sold, etc. So it's a complex supply chain and things will take time. So the other piece of advice I would say is is to have a lot of patience. So, have a lot of patience because you might think that, look, here's a solution, this is obvious, but actually it's often on you because you haven't figured out that there is something else that's preventing them from adopting the technology, but it can be overcome. And, but that said, I think it's still also, I feel it's really worthwhile and I If I may, you know, again, this— I think that my awareness of this apparel supply chain, which wasn't there before I got into this industry, was quite an eye-opener. Because here you have something where there's so much effort and time and money put into, you know, into growing the cotton and getting the wool and then dyeing it, spinning it into yarn, making textiles, making great new designs, you know, 4 or 5 times a year, making the garment in one country, shipping it to retail. And then at the end of the day, you know, 20 to 40% of it is returned at retail.

Shocking.

And they, you know, you put in all this thing and the only thing you've got out is that you've added to the landfill and you've reduced the bottom line as a company. You know, so I think this, this waste to me just feels wrong. You know, just, it just feels wrong. And I think, I think we just need to, you know, and the environmental— we haven't really talked about it, but the environmental impact is, is huge as well, one of the largest contributors to greenhouse gas and other kind of things. So, so I think it's, it's one of those things where you can help the industry both, you know, get financially better off by improving your bottom line, not wasting money, and improving the environment. So I think everyone should work in this industry.

Get involved. Perfect. Well, look, I am really glad that we had time to talk today. I was excited to see how your work has evolved from 2021 to 2023. I'm excited to see what's next. Maybe we'll see you take part in another 3DRC Grand Challenge in the future. I think off the top of my head, you're the only one who's won twice. Granted, it was under a different company, but still, they are merged, so I totally get it. So, so yeah, so well done again, and it's been great to talk to you today and understand more about your journey. So thank you.

Yeah, thank you very much, Michael. And yeah, and I hope 3DRC, you know, continues and keeps up the good work. Thanks.