Data Standards for 3D, a Foundation for Interoperability | IEEE | PI Apparel

OK, so I'm Carol McDonald, and this is Dana Todd. And we're discussing the exciting thing of data standards for 3D. Woo! OK. So my background is I'm actually a manufacturing engineer, mechanical engineer, electrical engineer, and had 30 years of— consumer products experience in manufacturing. So if you need it to go from zero to 10,000, that's my area of expertise. And this is Dana Todd, and I'll let her take it from here.

Well, I'm gonna add that the other thing that she's done, and we're gonna really dig into this, and how I met Carol was a few years back, I noticed that I was trying to, I was really frustrated with standards. I came from the tech side of the world. I spent 25 years building digital experiences, working with martech platforms, which is very data-driven, right? Data-driven advertising, data-driven decision-making, data, data, data. And so coming into the fashion industry, I was really perplexed by the complete and utter lack of standards here. I was just like, someone introduced this concept to me of a noonstick. Anybody ever heard this? This is true story. In America, once upon a time, every township had its own version of noon which was a stick in the ground, and when it was overhead and it cast no shadow, that was officially noon. Well, you can imagine what that did to the railroads trying to set timetables, and it ultimately created what we have now as our standard time, right? So if you look at a simple kind of thing in the fashion industry, fit models are noon sticks, right? It's a completely arbitrary thing based on what township you're in, what brand you're in, what day of the week it was, what style. There are completely arbitrary pace of things throughout this system. So I'm gonna dig into it a little bit more. But yeah, my background is tech first, fashion second. For the last few years I've been, 5 years now, we've been running a marketplace where everything is made to measure. There are no standard sizes on belladonna.com. Last year we've leveraged what we've learned there to create an on-demand, made-to-measure supply chain. And that's about the time that I met Carol. So Carol's been running multiple working groups within the IEEE now since 2017. Bet you didn't know that there's stuff like that going on. We're going to tell you all about it. So standards exist everywhere, and what I'm going to say right now, we are not talking about size standards. No size 8 standard, size small standard. We're not Talking about that, so before everyone in here reacts to just like, "No! Don't take my proprietary grading chart, no!" We're not talking about that. We're talking about approaching this with a scientific methodology and rigor that would approach any kind of standard. And standards are all around us. We literally don't even think about them every day. So just in this one simple photograph, we've got your phone. Your phone now works from cell tower to cell tower and state to state without incurring roaming charges and a whole lot of other things. You can use an Apple and an Android on the same system and no problem. Wouldn't happen without standards.

It's an 802. It's an IEEE 802 standard. There you go.

She knows the law.

You can go look it up.

The fact that your automobile itself even runs and can be assembled is because someone somewhere along the way decided that a 3/8-inch screw is a 3/8-inch screw. And that's what—

metric.

Well, whatever. You know what I mean. That's why she's the boss here. Broadcast. We have broadcast standards for spectrum. All of these things exist. We take them. For granted because they've been around for so long. And yet somehow, we've never really applied them to the fashion industry. So just looking at a few of the things that the body standards working groups are working on are, is the avatar the same as the measurements in the export file? Because I can tell you, they're not in most of the scanners that they have out there. They don't. They don't match up. Should we measure your landmark of the shoulder in the skeleton for an HAnim, right? If it's animated for— they're going to rig it on the skeleton. Well, guess what? For clothes, you might need it to incorporate a little bit of fatty tissue. Do we infer landmarks? Where do those come from when the limbs are touching? And we have an amazing slide to show you just how scanners view bodies and how they render them completely differently. What's the x-axis? Is it on the floor? Is it on the side? They don't even tell you. You have to guess. You have to pull it in a secondary— there's just all these little things. It's about communication. How did we determine the height on this avatar? Did we use metric, and how far down did we go? Which waist? How many waists do you think there are officially? When they first started, they had 8. Now we're down to 4.

Yay.

Are the extremities and postures actual or modeled? A lot of scanning systems actually swap your heads and hands. And feet just 'cause, right? They may not be the same one because it anonymizes them and it saves time in rendering, right? How are the meshes and vertices? So we've got meshes that are triangles. We have meshes that are quads. All of them yield different outcomes. But if you're not telling the receiver of the information on the other side what you did, where you did, how you did, then they don't know how to interpret it. And so standards are really just communication protocols, right? It's really just you do your thing your way, I'll do my thing my way, but we're gonna make sure that we understand how we still get along and how our systems can work together. So yeah, ANKLES too. There's billions of these. So what you may not know is that, like I said, since 2017, we have been working on all kinds of things together. And there are 4 working groups now, is that right?

Yeah, they're working on metadata, they're working on quality, they're working on use case, and then they're also working on the hardware/software portion of it as well.

And if you have not— how many of you are actually IEEE members?

How many of you know what IEEE is? OK, there we go. All right.

So most people are like, well, that's for electrical engineers. It's not for me. Well, they are actually just really darn good at setting standards, right?

What standards body do— that's what they do. It's part of what you do.

Exactly. And they have dedicated staff that help organizations like our working groups work with industry who don't even have to necessarily be members, who can all collaborate together to help to create standards that help everybody else get along. So, um, you might— anybody see the Kronos, uh, group thing yesterday? Yeah, so we're actually, our teams are working with Cronus on H&M standards. Like that very important where's your heel piece, right? And you would not believe how long it takes for people to agree on something so simple as a landmark like that. But the good news is we've actually gotten a lot of momentum. So these are just some of the people who participate on the regular, every 2 weeks on most of the meetings. Some of them are working— Yeah, so it's a very active group. And that means it's about the great time for you guys to come in, 'cause a lot of the work's been done for you. You can just come in and learn from it, but still participate and still have a voice.

We feel it's a quality whole. Anyway, we're gonna get into quality and you guys can help with that.

And I'm gonna give you a QR code at the end, but just to let you know, what this is all about is to research, explore, and disseminate information specifically for the challenges that we're hitting in apparel, footwear, wearables digitization, and the spillover, by the way. So it's not just clothes you wear, but clothes you wear in the metaverse, clothes you wear in CGI. So anything that affects the human body related to an external environment or a component, we're focused around those body standards and creating information that can be transferred and methodologies for doing so for sharing. So, and again, of course, we really want to enable the use and adoption, and that's why we're here today. In this year and to next year, our biggest paper to date is— we've gotten the landmark study out.

We got the—

yes, we've got the pose and posture study out. This one's going to be all about standards, the official— the official— it's going to be the Bible, right? So it's going to be a big deal, and we're going to need you to be part of that in some way, shape, or form. So I'll hand this off to Carol now. Okay, click if you want.

I don't know, you can just say click and I'll do it. All right, I'll Try not to stare into the lights. Okay, so basically one of the first things we came up with was terminology. And what we found is we're a global industry, which means English is not the first language for the majority of people working in this industry. So to help them out, we actually need to define terms and all be in agreement. So we broke it down. Into— we have avatars, we have digital twins, we have digital clones, blah, blah, blah, blah, blah, blah. OK, what are they? They're all humanoids. They are a model. You all in this room are humans. You eat, you need clothing, you've got families, whatever, right? A humanoid is your model. So that can be 3D, it can be a mannequin, it can be a list of data that says here's your measurements, blah, blah, blah. It's all just a model. It does not require health data. Insurance. So if you want to figure out how to divide it, it's really kind of simple. Now, cover is what you're all wearing, and because you have attachments, right? This is— this would be a cover. It's on the body. My footwear is on the body, right? Nothing's internal in this. It's all what's on the body. Yeah, so watches, you name it, accessories. We're going to put it under a big category— oh, we're halfway through? I think we're about— we're on 6, we're good, okay. So then, so that's a cover, okay? Now if you have a model of it, it is a coveroid, because -oid actually means like, right? So it just means like. So, and then we actually have the interactions. So we call those transformations. We have 4 assets in our entire environment. So when you take the data from the human, you now create a humanoid, right? So you go from a human to a humanoid. Some of these were like, well, those are gonna be in the future, but some of them we definitely know how to do. Now, you can also— now think about your clothing is sized. That means your clothing relates to how big or small you are. Okay? So your humanoid is actually embedded within the coveroid because you have to have the garment to fit outside of you, right? If you can't put on what a 2-year-old wears, it's not gonna work, guys. And so, well, maybe they're, no, you can't. Anyway, so I'm like, maybe they're sock, but I don't think so. So my point is that it's actually embedded within the coveroid, and we never think that the human model is embedded in that pattern. Right? So that's how we can say the humanoid impacts the coveroid. And actually, I think I will— anyway, so we can see how they actually interact together. And that you can just kind of work around. You make a 3D model of a garment that you're then going to produce. Well, but while it's a model, it's still a coveroid. When it's produced, it's now a cover, and you're wearing it, and you're rather happy that you have your coat, right? Because you're now covered up, and so it's covered. So you can switch the slides. So how do— I mean, it's an ocean, people. You've got a— part of it is you've got a wide bit of use cases. If I'm in AR/VR and I'm making a gaming avatar, I don't really care if I'm correct on your waist measurement or on your hip measurement. I don't care. I don't need to care. But if I'm definitely making it for— on-demand and made-to-measure garments, I care, right? So there's a whole range of use. So let's start simple. We have to start with communication, security, privacy, and trust. That means I have your personal data. How do I take care of that data? Communication means how do I take care of it in transit? You're like, transit? What does that mean? Oh, from one phone to the other phone. That's— your data is officially in transit. Who knew, right? So we have that one, and then we also have the quality. We're going to start with the simple one, metadata, which is basically that human-readable text at the beginning of every file. And then we have the quality, and then we have landmarking. Quality breaks down into multiple stages. You can click. Oops. Okay, so communication security, privacy, and trust. You've got PII, which is your personal identifying information. You have a recommendation that children Children should be, have additional guidelines and requirements. There are other IEEE standards working on the rights of children. We actually have to reference those as well, because you have to make sure, for example, if you add a rig, you don't add the adult sexy woman rig when she's a 12-year-old, okay? You do not need to do that. You need to understand, and that actually goes to words about how girl and woman is intertwined. Try to tell a computer the difference between a girls' night out and it's a 10-year-old girl. Computer's going, "I am so confused." Yes, we have gotten you confused. That is the goal. So we also have rights from GDPR, which is right of access, and how you actually handle that when you have modified an existing file with somebody's measurement, when do I become— where does the right of— to be forgotten starts and ends. All of this is all really subtle, and we have to go through. Next one here. So we have some metadata examples. You conform to our standard references. Like, what type of rig are you using when you're automating it? Is it a digital twin or a digital clone? Those mean distinctly different things. And which pose are you in? Which rig are you using? Is it normalized? Is the body area— are you only giving me a hand, or are you giving me the whole body? And then also, you know, the type You have to do it. The other thing we found is a lot of problem— is there are 4 different ways of figuring out your CAD. It's a right-hand rule, it's a left-hand rule, and Y can be up or Z can be up. Now, there are 4 choices, 4 combinations. Everybody uses all different ones. But I just need to know if you're doing a right-hand rule and I'm bringing it into a left-hand rule, just tell me what you've done so I can convert it over because it actually changes your mesh from being clockwise or counterclockwise generation. You're like, way more detail than I want to know. But for your tech guys, they go, oh, it's human readable. Okay, now I know what to do. So we can—

and external systems can read it too.

And external systems can read it. You know, I'm just like, just tell me. I don't care what you do. Just tell me. Now, do you think this is the same person? Is it a trick question? OK, the answer would be yes. We took 4 phone apps, and this was the models that we gave back. Now, tell me, as apparel designers, can you make a garment for— exactly. The answer is no. So that's what I'm saying. So we need, from the scanning providers, the actual posture, We need the actual body shape. And we need to know, if you have a body rig, which one is it? Because the body rig will indicate whether you can actually do this, OK? Because most of the body rigs have one point here and one point here, but they don't allow you to roll your shoulders. I'm curious, well, like the inaccuracies in the past with 3D in the fashion industry, is there a way we we can like input things like body measurements, body fat percentage to be more specific on the outcome? That would be standards, dude. Okay, so—

Perfect.

Good segue, sorry. So the answer is yes, we can bring in, but then how much manual data? And honest to God, I can tell you, put in your height. Now if you're a male, you go, "Uh, I'm going to add an extra inch because I'm a guy." Okay? If you're female, you tend to go, And they go, "I'm not 150 pounds, no, I'm 135." Okay, but can I now make you a garment? Eh, no. So how much do you truly, you know, on the phone? You gotta be honest. And that's—

Consumers are always honest, aren't they?

Honest, always honest, always honest. So the key also is that we have to train consumers. We've trained consumers the following: any garment I can get over me, fits. Does it look good? Well, no. You know, from a trained eye, they go, "Oh, geez, okay." But you've trained consumers, if I can move, it fits, okay? That's all we've trained consumers on. We've also trained consumers that if I don't like it, it's not close enough, I can exchange it a kazillion times, okay? I mean, I can, you know, buy 3, exchange 2. We haven't trained consumers to say, "Take a little extra time at the beginning. Take 3 poses instead of just only, you know, side. Okay, now do this. Now, you know." So now I can make a garment because I know I'm standing up straight, I'm bending over. We need to train consumers why we need the information. We haven't trained them. We've just said, Here, good luck. Pick what you want. So, and then next one. So now we have landmarking as well. So landmarking is, as you all figured out, we can't define what a hip is. We can't define. We got the landing point, right? OK, we're on the heel. That's down here. So you can put on shoes. But could you actually take the waist measurement or the hip measurement from one brand to the next and bring that over for interoperability into any of your apparel CAD? The answer is probably no.

Not right now.

It's not right now. So can we set up a standard for the anchor points? From a design standpoint, you know, if you want your shoulder to land here, I don't care. I just want you to know, here's the shoulder of the body. Now do what you want. Right? I mean, you've got to understand. I'm not telling you you can't do anything. I'm just telling you, you know where the knee is. Here's the knee, right? I know where the thigh is.

I know—

now, some of the providers will say, oh, well, they should have a thigh gap, even though when you stand and you're slightly heavier, you may or may not have a thigh gap right here. But I'm just going to make a thigh gap. Now I'm going to change that dimension of that leg to admit No, I need to know the actual person. The answer is give me the person, and I'll do what I need to do. So we need anchor points so that when you model it, you know where to attach it to the body. So when you're doing all that draping and soft physics, it interacts with the body correctly. So these are just some standards that we'd sure like help with. OK, so—

And there was a Landmarking Pub paper that we published 2 years ago.

Yes, 2 years ago. So that segues into the— Ta-da. Yeah, open your minds. Help out.

OK.

I think we're supposed to be soon to be finished. And then I think the next slide, you all, I think, have entertained you enough to open your mind. So we're on an IEEE website. This is the link down there. That's the UR. I don't know if you're—

yeah, so the scan, it's the same thing, so the QR code. And you can also hit up Carol to join the groups. So, you'll be onboarding. There's this separate area called IMEET where papers and notes from the meetings are shared. So, you can read up on everything that's been done once you join up. If you are part of the Industry Connections group, you do not have to pay for an IEEE membership because if you're not one— if you don't want to be a voting member, if you just want to contribute and participate, that's better for us. There is— so, that page is actually super long. There's a whole lot of things on it. And one of those other things that is very critical. We ran it last year, but we're running it again. We have a survey asking people how they're interacting with 3D, what you're doing with it, any issues you've run into.

We ran it last year, and what, 30% of people aren't even using 3D tech. You're like, I don't know how you're going to make it till 2025.

Others use it and then just throw it away when it comes to the factory, right? So it's— there's That was last year, so we'd love to have a larger— I think we only had about 50 participants. So we want a larger data set. So please take the survey. It's very, very short. That way we have a sense— y'all are the people we really need to hear from because you're actually at the forefront in terms of adoption and utilization of some of the technologies that are most dependent. And my concern, frankly, is that we're seeing 3D run, run, run, run, run, but it's not actually tied to the reality of making garments, which is my big financial risk when I'm trying to make things to body measure.

So yeah, so do be aware that your avatar, even though you look good in, what, Second Life? I don't know what other guys you're playing with. Oh dude, I'm dating myself on that one. But, um, anyway, Roblox, whatever is current, right, is not going to be an avatar that we can make garments from. They are distinctly different. So a lot of people are like, "Oh, I have an avatar." Yeah, not really. So, but they're— and they're distinctly different, and there's a range, and you have to understand that there's a range, right? Some of the avatars are just your hands, just your head, because they're representing you, and that's the definition of avatar, which is why we had to come up with humanoid, 'cause we're like, "A humanoid, that's your hands and your head." Fought for 6 months over avatar, so.

Humanoid. Humans, humanoid. Nice. Everybody could pretty much agree on that. It doesn't preclude any particular technology.

Yeah, so you can call it the rest of it. Digital twin and digital clone actually have a technical meaning on how you have set it up and how they are different. So you have to bury down in. And these are some of the papers that we have released.

Free.

Free. So the Landmarking, the Position, Posture, and Pose. And this Position, Posture, and Pose has just been released. Like March. So this is hot off the press. And then the assets and definitions. But literally it took almost like 2 years to get that position paper. Because what we found is there was no agreement on what posture was. OK? Posture. Now if you look it up in one dictionary, it's one definition. If you look it up in another dictionary, it's another, especially if English is your second language. You wouldn't get a complete definition of posture until you looked it up in the Merriam-Webster dictionary. And then from there, blah, blah, blah. So please look that over and understand. And she says wrap up, ready for questions. So this is the last slide. So we're— all right, we made it time-wise. So any questions? Did I keep you awake before lunch? Was that— did that work? OK.

Very detailed drill down. We have a question over here.

So looking at some of the points that you have, it seems to be focused mostly on apparel.

Are you considering looking at footwear?

Yes.

Her favorite topic.

Her favorite topic. We actually do have a footwear subgroup. We have released a paper that's on that same website that you saw on the different definitions for the foot. We're working on a paper. We're almost wrapped up on the definitions for lass. So that's another whole one that's taken like almost 2 years to go through. We have a whole bunch of different suppliers. The same thing, there are 8 different definitions for the center line of your foot. So when somebody gives you that measurement, you have to know how they got it. And that's the real key. We're like, don't—

Wait, the center line is the waist of feet?

Yes, center line of your foot is your waist of feet because you have to know whether you're doing the centerline of the entire width, whether you're doing it to the second toe, whether you're doing it at an angle, whether you're doing it from the centerline. Yeah, yeah.

Go ahead. And then a follow-on question.

Are you doing anything around digital materials at all? That's 2025. Because we've introduced Cover and Coveroid, we're focusing on getting the body out the door, if you know what I mean. Yeah, we're trying to get it graduated. And then once we do that, then yes, the next— section of the standard is to go through all the covers. Because you have the covers, you have the soft physics, you have the animation, you have—

But there are textile groups within IEEE already.

Yes.

So you can go ahead and start participating now because we will be collaborating.

Because we'll need all that information anyway. So it's like, yes, we'd love to have somebody starting on that because that is the very— that's next. That's next. Yes?

I'm piggybacking off Thank you for that question. Could you give us a high level of the roadmap in terms of the graduation of this since you started it like some years ago and how it would be adopted globally in terms of the standard?

Well, the standard officially is— will probably come out in 2025. So what we have— I mean, we keep— because it's an ocean, it's a rather big ocean. It's the Pacific. It's not the Atlantic. So what we've determined And what we realized as we're doing this is we had to go back and define what is a humanoid. We had to go back and get some definite definitions so that when we bring that into the standard— because what we're able to do on the standard is literally reference the asset and transformation paper and say, this is it. We can use that definitions.

So there was an earlier draft, and there's a working draft now. So if you're participating, you can actually jump in now.

Yeah, you can actually see and—

I've been using them for my data labels to try to go back because we use a lot of different body measurements. And so what we were finding is that all of our scanning partners had different methodologies and labels. So we normalize that data to match closely to the IEEE. So our waist is defined as smallest width waist in the backend. So we know exactly where it's taken.

So, and that may— that is one waist, right? That's one. One waist. But at least we now— OK, we have one definition of waste, we can have another definition of— I mean, there are going to be multiple definitions of waste.

You can pick one.

You can pick one, or you can state it, I'm using this one or that one. Okay, 2 minutes. Yes.

I'm just wondering, we've heard from a lot of startups and some that are in other industries, in technology, avatars, video gaming, and such, and we're utilizing— this group specifically is utilizing these standards to towards fashion, but there's a lot of startups and new technologies that are coming out of like Meta and Silicon Valley and all these things. So how can we stay up to date? Not only—

and I am going to say standards—

how can we stay up to date and ensure that like not only are we setting standards and learning for our industry, but also adopting and learning and growing with other industries that are also innovating?

Yeah, so we are actually, um, We have some of the people from that. I didn't put Kelly Varros in there. But we have people from that industry that are working on VRs and ARs and all of that. So we have lots of— we are definitely cross-functional. And they're like, well, why do you need—

In fact, I would actually say that we're missing more pattern makers, more fashion production people. We're heavy on engineers. We're heavy on animation. We're heavy on people who are looking at this from different angles. Much more startup-y, future-focused. And in the metaverse space, we really do need some fact-checkers on the scene who are hardcore pattern makers and be like, yeah, that doesn't work.

How are we bringing the avatar into reality? Right. Yeah. And our whole point is, if I have metadata and I have a little thing on going gaming only, you know not to use that avatar because you know, eh, it might be that person, eh, it might not. But it's perfectly fine to go, oh, OK, I'm in a game. I look close enough like you. I'm good enough, right? So we have those people already, which is why the quality has taken this long, because they're like, I'm coming from VR, and now I'm going to fashion, right? So how to actually define— and so some of it was like, oh, why can't we just have Y is up? Because fashion uses Y is up. Because Z started because everybody was making buildings and boats, and Z is up on all of those. So, okay, we are done. So—

If you have more questions, let us know. And please join the groups.

Yeah, so join the group. Do whatever. Read the papers. If you have questions, whatever—

Give a copy to your engineers.

Yeah, if you think we're totally in left field, let us know. I really like people who go, "What are you doing?" Right? Because then you get a response, and then if you meet the needs of both groups, then you're there, right? I'm not saying compromise, I'm saying we identify what you do, do what you like, just let us know. Kind of like your mom, just let us know where you're going, when you're gonna be in tonight, you know, that's all I want.

Thank you so much.

Thank you.