"Failure Is Part of the Plan" Bonprix’s 3D Transformation Journey | PI Apparel

We're actually going to start with you doing something. So who of you has ever started something wanting to fail? No one. That's good, because that would be stupid, right? So we also didn't do that. So we don't want to talk about the ultimate failure. We want to talk about the little ones that we make in the process and on the way to actually successfully transforming our industry. For the better. Your pilots for today, it's me on the one hand. I'm with Bonprix for 17 years, been doing IT projects most of the time, and now for nearly 6 years in the technical product development and helping to digitize the department together with Solveig.

Yeah, thank you, Tom. My name is Solveig. I'm working at Bonprix since '21. I started as a 2D and 3D I'm a technical developer, and I was in former days a passionate pattern maker. Now I'm a passionate team lead, the team lead of the 3D team in our department. And we are responsible for developing all the standard patterns and in general for sizing and fitting consistency and for the product quality and to satisfy our customers. And if you have— Any questions, you can scan the QR code and put them on Slido so we hopefully can answer them in the end of the session.

All right. So we'll be speaking a little bit about our journey and generating 3D content that actually comes to life in the end. So some of you might have seen us last year here. So I'm going to do the first part of the journey rather quick. So that is not a complete repetition, but maybe for the ones that didn't manage to be here last year, that you also know where we stand and where we come from. So we started our journey on Virtual Product Island. So this means we did ask the suppliers to send us patterns and we did the 3D simulations instead of doing the physical sampling. So this was good, but it was not what we wanted to aim for in the end. So it was not the big change for Bonprix that we envisioned. So we did set up into the sea of trial and error. So for one year, we did a lot of piloting and finding out workflows, finding out job descriptions, and so on, piloting with like 50 styles during that year. And after that, we came to a good workflow for us and actually landed up at the lighthouse of growth, which already was like 600 styles done in 3D in one year. Followed by our journey to cape scalability, slightly more than 1,000 styles in the upcoming season or in the upcoming year. And of course, what we want to do is shoot for the stars and go for full 3D and lift off with our little rocket. And the way we do it, the workflow, will be presented again by Solveig.

Thank you. Yeah, our vision, a digital end-to-end production development. So it all starts with an idea, the idea from our creative team. And sometimes with the help of AI, they get inspirations for styles. And all these inspirations are collected to a collection, style collection, and sometimes also have to be edited in a tool we provide. And when everything is done and it's crystal clear how the style should look like in the end, we got briefed our team got briefed, our 3D production team got briefed by a collaboration tool. So for the collaboration tool, we provided the possibilities to give pictures, attributes, and descriptions so that every 3D technicians around the world know how the style should look like in the end. And when everything is clear, our 3D technicians go ahead in our 2D software, which is called Assist, and do the modification. If possible, with the base of a standard pattern. This is why we are very high— do put very much effort in increasing the number of style patterns so that nobody has to start from scratch and that we can assure the efficiency in the workflows. Not just the style pattern development is done then, also the simulation in our 3D software, which is Browzwear. And the simulation, all the pattern pieces has to be sewn together, and then we do the fitting in 3D. And this is quite unique for Bont Prix in comparison to our competitors, that we really aiming for fitting purposes in the 3D software. We want to save samples. We want to be even more sustainable with Bont Prix. And when everything is done, the fitting is okay, the grading is done, we can— and the style is approved, by the way, from— by our creative team— we can hand over the pattern pieces, including the grading and the tech pack to our suppliers, and the style can be produced. And what we really aiming for by naming it end-to-end is to use our files which we created in our technical product development also for other purposes, for example, for workshop purposes, and maybe in the future then for try-ons and size recommendations and other techniques which may came up in the future.

Perfect. So let's continue on the journey. So remember, we were sitting there in our little spaceship, starting off to shoot for the stars, go full digital rather quickly. But all of a sudden, smoke all around. Something was not as we expected. So as good pilots, of course, we had a look at all the red lights flashing. Finding out, OK, what's going on? Why are we currently being slowed down in this ambitious plan that we had? So yeah, we found out that we didn't make that progress that we wanted. And we had to look into why this happened. Of course, there were some things where we already knew, OK, we need to adjust. And this was our time and action calendar. So we knew that with the old way of the time and action calendar being set up, the new way of working would not be correct. But we also figured out that there were still a lot of samples requested for the 3D products, and we were missing some standards. We also saw that there were some process insecurities. I mean, we are working with people there, so of course, if you change a lot, that's always a difficulty to work on. We had a lot of changes, so maybe sometimes we didn't really get to do the trainings in time. And of course, there are still technical limitations as of what you can actually fit in 3D. So coming a little bit to the time and action calendar. So what we are doing there, and this was planned from the beginning, was to rethink about how the workflow should be, how the time and action calendar should be set up so that it allows for the product team and for the digital product creation team and for everyone in this area up to the supplier to allow all of them to work in a 3D digital workflow and still come up with a good product in the end. And we are now in our second iteration for the time and action calendar change starting with the new season. So let's see how this then supports our processes in the best possible way. And then also coming to physical samples and standards. So we found out that in the beginning, we could eliminate 20%— or for 20% of the styles, we could eliminate the samples completely. But at the same time, 80% of the styles at least had one round of samples done, requested by the 3D TBD to validate the 3D simulation, to learn on what they see in the 3D simulation and what they see actually in the physical product. Most of them were approved without any change. But still, having 80% with one sample round was not making it possible for our 3D workflow to actually come to life, come to an end-to-end workflow. We also saw that if we have new product groups that we onboard, we need to practice. We need to give the people a bit of time to actually look into a physical sample, to look into the 3D simulation, to learn how to judge what they see in 3D. And we saw that we started with some product groups where we didn't have enough standard patterns in the beginning, which meant that if you have to create a new product, it's taking more time and therefore preventing you to do more products at the same time. So what did we do in the end? We managed to reduce the sample request for physical samples. So now it's more the other way around. 80% are approved without a sample and only for 20% we still get a sample. But this could be new materials a shape that's diverting a lot from the standard pattern, something like that. So something where really the 3D TPD needs to learn. So again, at the moment, keep the samples where we still see them as a necessary thing. And then we're going to look at the future.

Yeah. What's new on planet TPDDI? So TPDDI stands for Technical Product Development and Digital Innovation. This is the name of our department. And as Torben already said, we haven't matched the numbers, so we had to increase our standard pattern library. And we have a clear picture now for our targets and which kind of silhouettes do we want to develop as a base for each and every product groups for men's, ladies', and kids'. And we started with our pilot phase for 3D in 2020. And we started with a very low, little number of standard patterns, as you can see here. And we put much effort in developing those standards, but due to projects and everything and qualification and other challenges, we haven't reached so many standard patterns by the end of '23. But last year we managed to reach 50% of the target picture already. So we qualified the people, we gave them the time for developing and being an expert in their product group and focus on the product groups. And with those qualifications, they also were able to increase the number significantly for the standard patterns. And for the next year, you see that we have the very ambitious goal to even double the number to 100%. So to be 100% without variations, not speaking of different rise heights, et cetera. And what we also realized, I mean, we started with a standard pattern development to enable our people to modify the pattern pieces for the given styles. But it was in the 3D context, but it also influences the, not only influences our 3D Bonprix collection, also influences and has a high impact of, and of consistency for the non-3D part of our styles. So it's a double match. The more standard patterns we have, the more efficient we can be, and also the more consistent we can be for all of our styles. What challenges do we face as of now? Not all the colleagues in the team are already trained for developing standard patterns by their own. So we are, or we need external support, but the availability of external pattern makers is not, let's say, easy. We have to coordinate and we have to align with timelines, etc., to make them kind of available and to reach our goals. And in the past, the development of the standard patterns took a very long period of time. Speaking of a mid-complex style or high-complex style took us about 1 year, and that's pretty much, it's too much. So, we need some solutions there. So, how can we achieve our goal? A clear target picture, which we already have. We defined the silhouettes we need per product group. We need 100% focus on standard pattern development. I mean, there are many other very interesting projects in the company, and of course, we join them, but we always have to step back and say, okay, but we focus on standard patterns. This is crucial for our benefit. And we need a faster realization, as I already said. So we aligned on a clear process for the standard pattern development, and we ensure faster sample production. So we align with the suppliers that the material is available when we need a size set for checking the new developed standard pattern. And We also have the external support. We identify appropriate external patternmaker, and within the Otto Group collaboration, we identified specialists for 3D and 2D support. So 2025 is for us setting standards, that what I already said, to focus on efficiency and preparing of— to focus of effectiveness and preparing for efficiency. And the next year, we will gain efficiency through the standard pattern development and to— and increase the productivity of each and every 3D technician we have in our international team. What's also new is the denim scaling. We put denim in our scope last year. We provided many standard patterns for men's ladies, and kids. We also prepared the briefing attributes and the glossaries, we called them, with the technical drawings so the briefing can be assured for the styles in 3D on the collaboration tool I told you. And we also provided, or tried to provide, a library and tried to provide a matrix for standardized effects. And also what we need is the digital denim fabrics we realized with Visu and have now available, have them made available in our library for, for Browzwear. Here are some results I would like to share with you. On the left-hand side, you see our simulation from Browzwear out of the system. In the middle, you can see a size that we, we checked, and on the right-hand side, the the production sample. And here the stonewash effect, for example, isn't quite the same like on the simulation, and also the fade-out effect on the pockets above. So, and the styling was originally with whiskers, but in the production it is without. And the next example, kind of like similar, also with whiskers, in the simulation, but very straight ones later in the production. So our summarize here is, although we gave the suppliers a quite clear picture of what they should produce, it's kind of handmade when we don't have a laser technique behind, and we need other ways to ensure the product looks the same like the simulation. And other challenges we are facing right now is that it is very time-consuming to provide the denim pants and browswear, to sew the pattern pieces together and also apply the effects when they aren't standardized. So next step will be standardization also for our effects and building up a library, so our technicians with a technical background are no 3D artists, and they need those libraries for being more efficient in the process of denim providings. And also, the fitting is quite challenging. In browse-wear, we have the Alvanon avatars, but these avatars are stiff, as you know, and we can't do a proper fitting for under-body measurement garments, also not quite good for jeans with elastic garments. And therefore, we need a solution, and we are working to together with Vital Mechanics, and you will add some other details here for a POC to have a kind of soft-covered avatar, hopefully, which gives us new aspects for fitting purposes and make it more reliable. And now—

Thank you.

Coming to the soft body avatars.

Yes, so something we already had in our presentation last year was the outlook that we wanted to look into soft body avatar fitting more within the last year. And we did that. And we did it actually in two ways. So the first one is probably pretty straightforward. We want to fit lingerie, underwear, and swimwear styles with a full soft body avatar. We currently work there with Vital Mechanics with their VitalFit solution. And yeah, we will have a look at that in a moment. And the second part is what Solveig just mentioned, underbody measurement garments where we can make use of a new technology that we worked on together with Alvanon and Vital Mechanics. So for the first time, it's possible to use an Alvanon avatar with a soft tissue, soft cover, and then use the Vital Fit methodology to come to a better fitting. So this is pretty new. We will be starting with a proof of concept with the training in the next months. So yeah, the technology is now available. I think they've made also an announcement in March that it's also possible for all other Alvanon avatars because we have our own Otto Group avatars. But this would be something that we will be exploring in the next months. For the LUS styles, we already did some testing and we already have some findings that we want to share here. So here you already see some promising first results from a brief and from a swim suit as to how it can look like in the Vital Mechanics software, that you see where the pressure is applied to the body and how it reacts, how it's being cut in, for example. This is our virtual representation of our fitting models. So we gave the measurements of our fit model to Vital Mechanics so that we are able to compare the VitalFit, the digital fitting of the bra, together with the model. And here you see on the right-hand side the compression, so where does the bra push in and where doesn't it push in. So that's also a very good view on how the fit is. Just a little bit to show the complexity of why it's not so easy to fit a bra in the first place. It's already when you start doing it in the 2D software and you need to stitch all those pieces together and you need to keep the order that you'd stitch them together. So it gets quite difficult in the end to put this type of garment together. But we did this, and we are constantly exploring this, currently doing our standard pattern development for bras together with Featuring, which is also a part of Otto Group, to learn more on this technology there. One problem that we faced with the briefs was that it looked strange in the simulation. So a lot of compression in an area where we didn't expect compression. And of course we wanted to look into why is that the case. And this is also something that we want to share here. So you see the result after some tweaks that we did. And this fit is again very comparable to what we see on the live garment, on the physical garment. So it's crucial to, for example, look at the grain lines of the garments and so on. So the pattern makings, pattern markings that the pattern maker has to consider needs to be reviewed. Trims, elastics is very, very important. So this is definitely something that we need to look into in digitizing them. Understand material physics, pay attention to correct use of force multipliers. This is Browzwear-specific, but probably all the other softwares have it as well. We improved the lower body of all our Vital Mechanics avatars so that you see that they have both got a bit more realistic shape. And we looked into some standing, the way they stand for better placement and so on. So this is where we are at the moment. So we are constantly experimenting with this product group. And with that, we're coming to—

Yeah, the next step on our journey is to bring our 3D files into the webshop, producing webshop content. And there we are working on a pipeline solution as a basis. And yeah, on the left-hand side is the product out of our department, so the technicians developing the style and for styling approval. Approval, and then we can hand over the style and the file to our brand and content direction, who are closely working together with photo studios and 3D artists as of now. But as I said, we are working on a pipeline. And then the next step out of this will be the 3D-based services with AI models for outfit combinations, or just with an AI model, or for virtual try-ons. Et cetera. And now the journey continues. And the headline of our speech is "Failure is not an option, it's part of the plan." And it's all about humans. And our team is out of humans. And we should be allowed to make mistakes because we are stepping right into a territory which is quite new for all of us, and this is the interesting part of it. And yeah, we are very proud of our teams full of engaged and ambitious people and really brave ones. And yeah, mistake should be a part of the journey. Thank you. Thank you.

So the first one is already something interesting. So we don't actually pay for most of the sampling. So it's more that the suppliers reduce their sampling costs. And maybe we got a better buying price in the end.

OK, I take the second one. Was it difficult to change the mindset? We do many, or did many workshops for change management, and of course it is not easy to change a mindset when you come from a technician part, an only technician part, and now wear the glasses of a pattern maker. It's not that easy for sure, but step by step it's getting easier, and yeah, I think we made a really great approach and in the last years. And what product lines are not done in 3D? Tom already talked about restrictions. We have technical restrictions for fitting, because we just put those into scope which we can fit already in 3D. So jackets, for example, are not in the scope yet. But we are going to start with non-padded ones, for example. For example, in the next year?

I think fitting standards is also— should I do that?

Mach du.

Yeah, so I think we don't really use 3D in the whole value chain. We use it, as you saw, we use it from the idea already. We then do it for the fitting, we do it for the production approval to the supplier. Then comes the supplier, they don't need to have a 3D software, they can just go with our patterns that we provide and then they can send us the final garments. And there we take off again and say, okay, we also want to put it into the webshop, so we want to use, as Solveig explained, our 3D that we created throughout this whole chain to use it also for the webshop, so that we don't have any digital waste waste because our product always progresses through the value chain there.

Coming to denim washed effects, can also be digital printed? Yes, there are many ways I think to realize the denim effects and also provider for the denim effects in 3D, so we are working on it and we are always open for new ideas.

Yeah, so a virtual try-on, we did some exploration there already, I think 2 years ago, which was not that successful. And we also tested with a customer and they were not really into it at that time. But we will be looking into it every now and then again and try to do it again.

Have you explored Geonologia? Yes, the library is also part of Browseware. But what we have in mind actually is is to automatize the whole workflow around denim. So maybe also provide our creative team with the tools to bring and apply the denim effects. So this is, yeah, trial and error right now, but what we are focusing for the next years. Any other questions? Hands up, please. We have 1 minute and 30 seconds left. No sign from the back.

All right, then thank you very much again.