Why is Digital Colour Accuracy Still a Challenge and How Can We Fix It? | PI Apparel

Hello, good morning, good afternoon, good evening everyone, and thank you for joining for another Pye Apparel Spotlight. For those of you who don't know me, I'm Michael, the Director of Production and Content here at Pye Apparel, and I will be the chair of today's event. As you all know, today we'll be diving into one of the most persistent challenges in the industry, and that is digital color accuracy. We're going to be doing it as two 45-minute panels. The first will focus on digital color accuracy still being a challenge and how we can fix it. And then we'll be moving on to panel 2, where we will discuss how to better align people, process, and technology to tackle limitations and misconceptions in the space. So thank you very, very much for joining us today. Before we get going, just a little bit of housekeeping from my side. You all should be able to see a chat box in the window that you have open for this event. If at any time you'd like to make a comment or ask any questions, feel free to pop them in there and they will be sent through to me. We will try as best as we can to tackle any of the questions that you send in during this panel. But we do have a lot of angles to get through, so please bear with us. Just so that you're aware, you can also connect with our speakers after the event by clicking on their names via the platform and you can speak to them there. But of course, feel free as well to connect with them via LinkedIn if you'd like to get in touch with any of them. I'm sure they'd be delighted to speak with you. I'd also just like to highlight our upcoming event, Pye Apparel Europe, which is going to be taking place at the Grandino Hotel on Lago Maggiore outside of Milan between April 8th and 9th. We will be talking about digital colour as part of that event too, and we'd love to see you there. I believe a colleague of mine is currently putting the registration link for the event into the chat box, so feel free to click on that, have a look at the agenda, and if you can make it, we'd love to see you there. I believe that we have an early bird campaign that is running and expiring tomorrow, so now is your chance to nab your ticket if you can make it to the beautiful shores of Lago Maggiore in April. We'd love to see you there. We'll also making these sessions available on demand. So, you know, if you don't want to scribble down notes and you want to pay attention fully to what's being said, please know that we will be releasing these on demand also, just as soon as the videos are edited and ready to go. So with all of that said, I would love to start the first panel: Why is digital color accuracy still a challenge and how can we fix it? And I'm joined today by Olaf, Silvia, Michaela, and Gabrielle. So let's start with a few intros, shall we? So Silvia, do you want to start? Please let the crowd know who you are and how you're connected to this space.

Hi, I am Silvia Simčuk. I founded— I just founded Fashionista, which is an AI tool that creates physical garments using AI. But before that, I was working for Timberland and I was working in digital product creation team where we had all these issues we will be talking about today.

Thank you very much. Gabi, over to you.

Hi, yeah, my name is Gabi Scheiner Hill. I'm co-founder of digital materials agency Bureau 555. I'm also a fashion academic, and I've got 20 years experience in product development on the supply side, working with suppliers and with brands as my customer. So I'm coming from that perspective.

Thank you very much. Michela, what about you?

Hello everyone, happy to be here. My name is Michela Jauch, I'm 3D team at Oddlaw. I'm actually coming from the design side. I've been 15 years a professional designer. In 2015, I changed into DPC, into 3D product creation, and last 5 years in the team at Oddlaw.

Thank you so much. And last but by no means least, Olaf.

Okay.

Please tell us a little bit about you.

Exactly. Thank you very much to be here. So my name is Olaf Kölling. I'm in colors over— oh, I cannot see. I would say about 30 years now, beginning in the '90s, starting from setting up design studios and then founding a lot of small companies. And at last, the company DMX with my colleague Gerd 10 years ago, and we are in color and color is our passion.

Amazing. Thank you all so much. So let's dive into the angles that we've prepared. So Michaela, let's start with you. What are the primary technical challenges causing color discrepancies between digital designs and physical products that you've seen?

So I would actually classify them into 6 points, and out of those, 3 can be addressed in advance and other 3 are a little bit more tricky to address and to tackle. So those which you can sort of get ready for is your color profile. So basically you set up your profile, your gamut, then you have to calibrate your devices as many as you can. Basically everybody working with colors should have calibrated, either software calibrated or hardware calibrated monitors. And then of course, you should also set your design software or 3D software to the profile you work with. So those are the preparation steps. And then of course, you have some challenges, which are, for example, material variability. This is something what is difficult to affect. Then the technique you are using for coloring or printing your fabrics. And then of course the lighting conditions are also very important. So the lights which you are using to judge your swatches and also the lights which you are using in your render screen, in your render preset.

Amazing, thank you very much. Does anyone else want to add to this?

So I just wanted to add that the big problem we had in Timberland actually was that The questions we had about the colors and about their calibrations were often coming from, for example, e-commerce team. So they were saying, okay, if we want to use 3D in e-commerce, the colors have to be similar to the ones of photo shootings. The problem was that photo shootings were of course Photoshopped. So the colors were not exactly true to life. And we, we worked in Photoshop or in Lightroom to try to set up a kind of— how I can say— we had a similar setup for our renders and for our photos to match somehow the color, but the discrepancies were basically coming from how they were adjusting the photos. Not how the 3D software was calibrated. That was our case.

Okay, so you kind of had to reverse engineer back from the way that the e-com team was actually positioning them. Okay, that's interesting. Olaf, did you have something to add?

Yes. So in our experience, there's always a chicken-egg problem we have with this stuff because do you have the right source? This is the main question. Is your color, is your data right? If your data is not right, you can adjust what you want. It will be a mess at the end, right? So this is our experience. First of all, getting the data right, and then it means what is right data, that we have kind of a long discussion about this, what is the right data. And we're coming to the next points, we, we can achieve right data. But anyhow, if you have the one, if you call it single source of truth at one point, then you can go from there to other steps. So maybe some shootings, or maybe simulations about shooting. So shooting can have maybe on sundowner, sundown light, completely different colors. If you have an orange glue in the, in the light, the colors are completely different from the natural ones you have in your light box or whatever.

So just on that point, because it's great, on that point, so you say that we, you know, people have to have the right data to work with. Can you talk to us a little bit about what that date data is. I mean, I'm sure that isn't something you can answer in a short thing, but just to give us kind of like a taste, a flavor of it.

Exactly, exactly. Because it comes, it comes from, from the Technicolor technology at the end. So defining data, what does it mean? It's what, what you see. And you can, you can, at the end, you need an LAB value for the data, for the color, and this represents perfectly what you see. And coming, going out from this you can maybe calculate upfront whatever you want to, but you need at one point this alignment to the natural ones, to the material you have in your hands, for example. Yeah.

Okay, nice. Gabby, did you want to add anything to this?

Yeah, I mean, it kind of moves slightly into the next question, but I'm of the opinion with the single source of truth, like that's the method that I sort of suggest subscribe to, because I've also been in the same position as Silvia, whereas I've actually digitised rendered materials and had them essentially rejected because they were then being compared to a photo studio environment as opposed to a digital materials environment set up in whatever your colour standard, your lighting standard is, D65 if that's your standard, depending on what it is, and a consistent set of lighting. So I've experienced exactly the same challenges, Silvia, I must say.

The problem then is that every single season, like, we set up the right lighting in our 3D, like we set up our tools and the season after lighting was different.

So yeah, if I started again, if I may add to it, we had the same issue and we finally got to a point that we basically give the renders together with the photoshoot images to a retoucher at the same time, together with the physical product, and he's checking everything and adjusting colors for everything. And we found that this is the best at the end.

Probably it is because 3D is much more consistent if you have your setup.

Yeah. Also the sample. Also the sample, I have to say, because sometimes when you get the SMS, it's a different batch, it's differently colored than the production sample. So then you can still have returns because the production looks different. So then he was really checking the production sample with the 3D and with the photo shoot.

It's nuts. It's nuts. But I'm glad that it works. I'm glad that it works for you. Just on the, on the note of kind of like, you know, single source of truth, who sh— who should own that within a company? You know? Yeah. Is it— does that fall with the DPC team? Does it fall within the color-specific team? Like, who should actually lead that within a company or a brand, do you think?

I mean, I know on my side it depends on what the makeup of your team is. So for us, it was the fabric technologists who always had that piece of information. We didn't actually have a full color team, but the fabric technologists gave the final approval and sign-off on all of those points. So for us, yeah, it was technologists that were working on it, not designers, not DPC.

Yeah.

No, I think color designers, I think we have some customers where color designers have their head off. So they say, OK, what I see, I I create the colours, the colour standards for the season, for instance, and this should be the colour that we should follow. And then at the end, the reality comes on the table. Oh, it's not possible, la la la. And we have different production facilities, different colours come up, and, and, and. So, but anyhow, having a good starting point to go through is very important. Not going from step to step to step, maybe having my idea, then I have my laptop, then I have my SMS, and then I have my production, and I have at the end 4 colours. Huh, which lead? And production can be different colors depending where you produce or whatever, what lot, and so on. And to having a standard in place, so it's good to have a department into— in the company which has, uh, yeah, maybe the decision of this. So this is my standard, and going from a standard, it makes it more easy for everybody else. So maybe for production, even for DPC, if I have a standard I have to follow, I say, okay, I can rely on my standard. Maybe it can derive a little bit depending on production facilities or whatever, or conditions. And but anyhow, I can define a standard, I can define a tolerance. And when I'm in the tolerance, I'm quite, quite okay, even if I go with my standard. Upfront from the first one. Now, this is our experience, and this is a good way to go so you don't have hassles with the different stages, right?

Amazing. Thank you very much, guys. Well, let's move on to the second one, and Gabby, I'll start with you. How do variations in display technologies and lighting conditions impact color perception in DPC workflows? I know you did touch upon this a moment ago, but if you could elaborate, it'd be great.

Yeah, I mean, on this side, essentially, I think, and I think members of the panel might agree with me, it needs to be sectioned down into the areas of DPC when you actually need color accuracy. Because at the beginning of your process, of your design development process, when we were obviously developing products, color accuracy maybe wasn't the first thing that we needed for a quick decision, for a quick idea, for a placement, for, you know, a little refresh of newness. It wasn't that we actually were looking for that single source of truth at that point. And so for understanding that when essentially when we were looking at digital product that was created, we were looking at it through multiple different screens. Sometimes it was on our phone, sometimes it was working with a supplier overseas who were also looking at on different screens that were not color calibrated. It really does depend on your supplier setup, whether they have the ability to to do that, resources and technical skills to be able to do all of those to deliver to you as a customer. So for me, it was almost like there was a grace period when we were allowed to work where we weren't as strict, shall we say, on that colour accuracy, which is maybe, maybe different from how other people have worked. And then it was moving into, well, actually, we would move into lab dips and we would rely on data as opposed to what we were looking at on the screen. So we were literally looking at the spectral data, which is coming a bit later, and saying we understand that this is going to be the correct colour when it's dyed in the, in the mill or in the dye house. We know this will be correct, so we can move forward with our design and development knowing that the lot that comes through is going to be essentially approved by a machine rather than a human, if that makes sense. And something in the question which is around colour perception, really it is perception if it's us as humans doing that. It's what we're looking at and it's our eyesight at that time and the time of day and the space we're in. Even when we've, you know, calibrated to within an inch of our lives, we really have to understand that if it's late in the day, different time zones— we're in the UK, the sun went down at 2 in the afternoon— everything will change that perception around us, even in the most sort of like consistent environment. So on my side, it was much more thinking about, well, what can we do right now to control? If we can't control all of the screens that we're working on, maybe we can control the capture space we're working on. So consistent lighting, consistent capturing of materials, and using that as a starting point and a basis point to work from, rather than for us trying to control every screen, because it wasn't realistic for us as a a small company to be able to control every screen that we were looking through.

So yeah, that makes a lot of sense. I hadn't even thought about people's own personal perception too, because the way you and I will see a color will be entirely different. So take the tech and the hardware out of it, even though it's obviously a huge importance. Um, yeah, I hadn't even thought about it like that. How interesting. Does anyone else want to add to this?

Yeah, completely agree with you. Oh, sorry.

No, no, go first, go first.

Completely agree with you. So having a— you need one situation you can trust, so otherwise you get completely weird. Yeah, so finding a color which is on my phone, the right one, is on my Mac and on my calibrated screen is not existing. This is the truth, it's not existing. You need it, you need a compromise at the end. But you need at one point a situation where you can trust. And then you say, okay, this is the single source of truth. This is my— we have a dedicated room here in the lab with a calibrated screen and a light box, and this is our single source of truth. And then we say, okay, everything is calibrated and I have one, and I can have a— we call it a digital transition transformation. So If this is right, I say, okay, in this environment I have a good situation, and going from then to whatever— this is a separate discussion. What is your goal at the end? Is it maybe online business on phones, or is it a designer on his PC or Mac or whatever?

So yeah, but the perception topic is, is quite a big one because I, I remember discussions with, uh, with designers with merchandisers, with folks from catalog team that this is not the right color. And then we just bring them to our computers where we have calibrated screens and yeah, does it look right? Yeah, somehow, but why on my computer it doesn't? So, you know, it's always the same kind of discussion. And then I don't know how you work with colors because you could potentially scan the already dyed material, and this is one way to go. We didn't do this, this way in VF because we simply had too many colors, too many fabrics, and we often used the same fabric for different seasons in different colorways. So we basically had 50% gray version of the fabric. Then we added the Pantone color on top of that. So this also makes a difference because fabrics are dyed in different way than you do it digitally. So we had somehow to compensate this thing, but this was the only way to go for us because otherwise we were not able to do it on time.

Michaela, is there anything you'd like to add on this?

Yes, just to add what Olaf said regarding the single source of truth, because we were quite a lot discussing before investing into high-end monitors. And we actually finally decided to go to a 10-bit monitor, which has over 1 billion colors. And of course, then we were discussing, yes, but if you are judging everything on these super fancy monitors and then the regular customer, like 90, I would say 95, maybe even more percent of population is checking the products on a mobile screen or 8-bit monitors, which have 16.7 million colors. Then of course we are obliged to check that on that monitor as well to know what we are giving them, right? So now we are actually checking on both monitors because we want to know what's going to be the final output for the end consumer.

So of course this is a real challenge. So it's so different. So if you have this high, high-end monitor on the left side and your phone on the right side, you see, oh, come on, there's such a big difference in displaying the color. But if you do this at one point, there's one advantage, because if you have a digitization port and you can, you control every color on the same, then the color consistency underneath is right, right? If you compare two products, then you see, okay, in reality, one is more bluish, And if you do this this way, it was even on the phone more bluish, maybe grayish as well, because the phone is more grayish, so it's more light. It's, it's what we highlighted, phone, so it's a little bit, a lot of, of light in. But anyhow, it underneath, it will fit. And this is also a big advantage. It's not a gambling thing that something is maybe done from one colleague displayed or controlled, and another colleague is doing it a different way, and you have both two products which normally should fit but they don't fit. But with this, it will show that we will see the difference even if we have not the right phone or not the right monitor. The difference underneath is regular at the end. So this is one advantage. Not all, but one.

And we're definitely getting— um, sorry, Gabby, go ahead.

I was just gonna say, just something I was going to add. Obviously there's a whole product areas which don't always fall into this discussion, which is denim, um, uh, things like sequins, anything with like, you know, glitter in them. So I think sometimes when I've spoken as well to industry peers and I've talked to them about digital color, they're like, yes, you're talking about lab dips, but what about Yeah, a multitude of products that maybe also don't fall into this space quite easily if we can't lab dip them, for example, like a denim. So yeah, we can't forget our good friends in that space as well.

Sure, it's a problem, right? If you have a denim, hey, it's more the appearance, even if you go for other materials, leather, for instance, Same topic, it's a living material. If you measure this, you have one point, you need the appearance of the whole. This is then the next generation of measurements or something like this.

Two who are, um, who are in with us today, one of them saying that an issue is, is that a lot of the different teams that are working across the value chain are working in silos, so it becomes very, very difficult to kind manage that, but also the fact that there's so many stakeholders related to color as well along that value chain, trying to come up with a common language and a common way of doing something is very, very difficult. Um, so yes, they, they seemingly agree with everything that you guys are saying. Um, I think we might have lost Michaela from the feed, but hopefully she'll be back in a moment. Uh, but we're all good, we'll keep on, on going. Um, Olaf, this one's for you. Um, how can spectral color, uh, measurement improve digital accuracy and bridge the gap between digital and physical mediums?

Yes, it is at the end, it will not substitute a human, right? It's for us an additional information to judge. That's it. And I cannot use spectral measurements for everything. So even for the DPC, sometimes yes, sometimes not. Sometimes you need more, even for leather or material which is like denim or so. But what I can do, with a spectral measurement is I can control and I can compare. And this is very precise, and this is a very good point if I want to control my production, if I want to control my lab dip, if I want to control— or I communicate with another one, maybe in Asia. So yeah, and if you do this with a spectral measurement, you don't need to send samples around the globe. You can send spectral information and you see instantly what's going on. Okay, it's a little bit of education. You need to know, okay, what does it mean if there is a whatever, delta L, AB, la la la. And okay, but at the end, if you're used to this, then you can really prove a fabric even if you don't see it. Yeah, this is a big advantage.

I want to also ask, you know, have you seen in general that your clients have been quite open to using spectral colour data? And if not, what have been the barriers, I guess, to them adopting it?

Barriers is, it depends on the generation, I think. If you are used to for 25 years having fabrics in your hand and go to the light box and say, okay, I'm the magic eye here, and maybe, Even if you are the magic eye in the company, right? So it is okay, they say, okay, maybe I don't need this, or may bad experience 10 years before, whatever. So, but now the technology is on a level that I will trust the instrument more than my eyes, really. It's so precise and it's okay and it gives me the right information. And this can be The younger people, they're used to technology a little bit more and say, "Let's try it out." They will try it out. And if they make good experience, they say, "Well, why not?" It's easier. It makes life easier. It's not that somebody, a human anyway, needs to do the final go or give the final go to a Fabric or to decision. But if you have good information, you can give the go.

I was just going to add something to that, that I've worked with a colourist in a mill who has used a number of different, yeah, data ways of approving. And when it started working, you know, there's the learning period, and then after about 3 months when the system was like in place and it was working correctly for him, he was absolutely thrilled because essentially his lab dips were getting approved, right? It was not— he was not having to send or having to create them and send them time and time again and then wait for the answer and all those things that we know we have to do in product development. And he said he would never never go back. Like, that was his mill experience because he's like, my job's become easier, this is great. Um, and then they were a bit like, what's he actually doing now? Like, when we were having this conversation, like, oh, he's just— he seems really happy. But it was really good, it was— it worked really well. And it was specifically for core products, right? Core products, really good on menswear, bottom weights. Like, it like just worked really well specifically for that product as well.

Yeah, thank you. Our experience as well. For, for if it's running, then the people are happy, right? So, and if it's a second advantage, even for the supplier who's in charge to dye everything right in time and so on, and he got a very precise sample to fulfill, this is a good big advantage. Yeah, can measure against and see instantly, hey, I'm on or off. So yeah, saves time a lot.

The problem is convince them to start.

Exactly.

Yeah.

Like the change management is probably 50% of work of every DPC team or anyone who is working in technology in fashion.

Yes.

We launched Pyaparo back in 2013, and at the time our focus was on PLM, so it wasn't even on the 3D stuff yet. And ever since that time, Every single event that we've ever had, IRL or online, it comes down to people and change management. It is, it is a science that I don't think anyone has got 100% right. But yeah, it's, it's, it's a challenge. I would say it's even more than 50% of it from what I've heard. Like, I just think, you know, none of this tech is plug and play, so you need to actually onboard everyone and they need to understand what the advantages to, to them are. And the annoying thing is, I think with tech, the results aren't immediately always a benefit from the get-go. Like sometimes it takes a little bit of time for people to start to understand, oh, actually this makes my life a hell of a lot easier than it ever does, but they just need to give it that chance. So I know that it's a big, big challenge for a lot of people for sure.

You know what is even bigger challenge in this case? It's that companies are gatekeeping, like they have to somehow, you know, the fashion companies. So if you don't hear the successful story stories, it's much more difficult to believe. That's why we need this kind of discussions because at least some successful stories are coming out.

But sometimes they have so much hidden treasures in their company. Yeah, they only have to have a look in detail on this and they see, hey, I can save a lot of work and a lot of time if I do this digitally or with precise measurements, for instance. Yeah, but it's, it's maybe they are used to like 20 years to do the job like, like it is, and nobody cares about the effort that you have to run to the light box 20 times a day, 5 days a week or whatever. Yeah, but at the end, if you sum it up, it is a lot of work you have to do.

It is, yes, but it's a payoff in the end, right? There is a payoff if you're willing to. There's obviously a very clear payoff here.

Exactly.

And again, I think that goes back, back to as well, the other point that people in the audience made about the multiple stakeholders, different teams are working in silos. These things are available and they're there, but it's about actually trying to onboard that whole piece across the, you know, and communicate and create the language that everyone can understand, which is difficult. As we said, you know, the human part of this is the most difficult bit. Olaf, I'll stay with you for the time being. How can integrating technologies improve color consistency? Throughout the supply chain?

Oh, yeah. So, so it is, it is, um, at the end it's a matter of network. I think it will only say— you're talking about silos before, right? In a silo you're, you're trapped. What you need is a network, is a flowing data, and this makes it very powerful. And this could be— it could improve the whole chain because you save time, 'Cause it's maybe the information is coming overnight and the next morning you say, okay, I'm fine. And it doesn't take weeks of this. And the consistency along the chain can be done even with a spectral measurement, make an example. If you start with a sample which is agreed and you go for a lab dip, which maybe you have a standard variation of whatever, Delta E 0.9, 1 point, whatever, and you go for the next level, to the next level and always go from back to the source. At the end, your whole production— and you can produce the same fabric in Turkey and in Vietnam if you want to. And this is a real challenge, to align these production lines if it needs color consistency. And with these values, with these numbers, you can control it. You can, you can instantly say this lot is out of the, out of the game, because if I, if I will produce this with the other stuff, I have a color gap here. Yeah, this, you know, upfront, this is a really, really big advantage with this data. And this is the consistency over the, over the, uh, yeah, our whole chain, if you want to. Yeah, this is one, one of the advantages of this technology.

As part of that as well, I'd love to understand, um, and maybe Sylvia at Timberland, perhaps you can, um, comment on what you, you did. So I understand for the data side, it's great that you're getting all of this data data, how might you set up a color approval process between different teams? What might that look like? Is that something you guys do?

We set up what we, what we called virtual quality standards. So everyone who worked digital had this huge PDF, which was around 100-150 pages, that was explaining exactly how how assets have to be made. And that gave us— because we started to work with manufacturers, so it was a lot of people that were delivering 3D assets for our team, and all of them were delivered within this standard. That means that it really didn't matter who made 3D assets, it was always done in the same way. And that was a great thing because it really, really helps having standards. Now, I believe that when we want to bridge the physical and digital, we have to bridge the standards between physical and digital, like have, have a kind of translation between, okay, if physically it looks like this, this is the digital counterpart. And, uh, there is a lot of ways to do that, but if you don't have standards, you won't ever have consistency.

Oh, Michaela's back, and in perfect time too, because we're getting to a quick question that I wanted to ask you. So welcome back, Michaela. Hopefully you can hear us. You're good? Maybe not. You might be on mute. You might be on mute, Michaela.

Big excuse, big apology by computer, uh, from just crashed, so I have a black screen. I had to grab my private one.

Okay, no worries at all. I'm glad you could come back. So that's, that's great. So, Mikaela, I wanted to ask you, so I'm going to put you on the spot straight away. What role does material simulation play in achieving true-to-life color representation in 3D design?

That's a great question. So of course, the material is a very complex thing. So if you look at it, it has, it has a texture. It has the reflectancy, it has a transparency. So you have to address each of them when you are creating your digital counterpart. And of course, for that, we have a bunch of wonderful channels. So I'm going to name just, just the most important one. We have Specularity, we have Roughness, Normal, Domain Map, Diffuse, Transparency, Metalness, depending what software you are using. And then of course you can have some fancy one, which most of the time we don't use because it's more for hard goods like sheens and clear coat and so on. And then And of course, each channel can affect the look of the fabric differently. So if you add a little bit specularity, you know, it becomes immediately lighter and you have to tweak them because you have to really have the fabric in hand and adjust those layers based on what you see. And then of course, we have a lot of these third-party tools. So we have Olaf with us, I believe. So that's one of the tools. Then we have— I categorize them. So basically we have procedural. So for example, that's Adobe Substance. We have camera-based like Visu, Bandicoot. And then we have AI algorithms like, like said, there are many more. So I apologize if I forgot someone. And then of course, if we decide for the tool we want to use, we go with it, but I still recommend Do a double check, check in your lighting, render. That's the most important. Don't rely on the screen, what you see in your 3D preview, render. You always need to render with your light setup and then compare on calibrated screen the final render, I would say.

Thank you. Does anyone else want to add to that?

I was just going to echo really what Michaela said. With my experience with either some of the photo capture like Visu or Bandicoot or using DMX as sampler, we would often have to also then take the material into maybe Substance if that's what the customer was looking for, if that was their digital standards for checking that material. And actually, sometimes the materials we had to create for customers, we were asked to create them in grayscale, actually. So create those in grayscale, we then deliver to the customer, and then the customer deals with the color. Quite interestingly, you know, so we actually just had to deal with really like texture, the physics, and yeah, transparency and normal maps, but the color was actually taken out of our hands sometimes, which I always found that interesting because it made life a lot more simpler for us. But that doesn't mean that— it means that obviously when the digital material then goes to the brand, they then have to control the the colour space that they're working with it in.

Thank you. Well, we're getting through, through these angles really, really well. I'm glad to say that there's a lot of engagement in the discussion. I don't see any questions yet, but great to see that a lot of people are agreeing with you and what is being shared here. So for our last kind of 7 minutes, if you do have any quick questions in the audience that you'd like to ask, this is the time. For the panelists, I'd love to talk a little bit about AI, of course. Um, I don't know how much experience you guys have with AI as it relates to color correction, but I wanted to ask the group what advancements are on the horizon for AI-driven color correction and real-time rendering in DPC.

This is a real challenge, I can tell you, and, um, I think we are not so far that AI is able to make a quick transition and you have the right color. I think this will not work. We are not— not yet, not yet. But now we are, we are now having some AI tools on board. So maybe for making a better resolution, quality of pictures at the end can be improved. And this is really great. The results are amazing, really amazing. Yeah, photo agencies will not love this, but anyhow, the AI will do the job at the end. And, and this is done right quite, quite soon. But it's not a solution for everything, I can tell. For, for colors, for maybe color transformation, it's better to control it via algorithms right now. That's our experience, but next year can be completely different, so It's maybe, who knows.

The space is evolving very, very fast. I tell you, I am now working on AI tool for, uh, for fashion, but we are not aiming to work only on images. We are mostly working in product development. So making patterns, tech packs, and, and so, uh, using AI, but it is evolving extremely fast. What I have seen 6 months ago, it's completely different from what we have today. So potentially if someone will have enough big database with fabrics dyed in certain Pantone colors, it will come. And I believe it will be very helpful in this space.

But we have to think about AI as a simulation. Somebody has to dye it at the end.

Yes, yes, absolutely. Absolutely. In fact, in fact, often what What AI tools that you can find today on the market are offering for fashion is often very disconnected from the reality. And this is a problem because they are selling themselves as something that, yeah, help you and make your life easier while it is actually adding a layer on top of things you already have to do. This is, this is something that every fashion company have to evaluate.

Yeah, I've, um, I've not, I've not actually looked at the, the color side of this, but I've been looking at within, um, for color accuracy, but looking at changing lighting scenes, um, so having set, um, images of, of, um, archive fashion pieces and then looking at the lighting and using AI tools to control the lighting, which has had some really good results. But I can't I can't get it to give me complete accuracy right now. It still adds too much variation in, but that's something I'm working on in the background. So let's see, hopefully some tools will help me soon. Please feel free to tell me if there's more out there.

Yeah, we'll do a part 2 in a few months and you can give us an update on how it's going. We have had a question come in from the audience, which is great for our last kind of few minutes. Katie has asked, I'm curious about white in both digital and visual. Visual space. How do we ensure we get the best white in our garments using a digital space? Is this possible? Do we need to tackle this differently to other shades or not? I don't know who wants to take that on first.

I think it's one for Ilan. Come on.

If I don't get it right, I think— I don't know if I get it right, the question. It is about— what does it mean? It's a digital space.

The colour white, basically trying to control the colour white. In digital product creation because whites are notoriously tricky.

Yeah, white is not— yeah, it's really tricky. So if we're talking about whites, we have to talk about, yeah, some things which are not white. This is— I don't know the word in English right now. It is the material which comes up in the UV light. Right? So white is a brightener. So the brightener is inside the white. The optical brighteners are in the white fabric, for instance. And this is something you cannot simulate in your monitor. It's not possible. It's a transformation of energy from the things you cannot see to the one you can see, normally the blue one, right? And it's more bluish then. This is the effect of white paper or a really white shirt or whatever. It's this brightener. And this is really hard to tackle in digital. You can only have your whitest white on your screen is the white, that's it. It's not more shiny, it's only related to this white at the end. But really whites, you cannot, you can only control the measurement device, then you can control whites. This is possible.

Amazing, thank you. Well, look, we'll end it there with A mic drop from Olaf. Thank you for that. This has been really great. So thank you to the entire group for spending the time to speak with me today. I think it's been really great, great engagement from the crowd too. As I mentioned, if any of you want to get in touch with any of the speakers that you've heard today, you can do it via the platform you're in now. Or as I said, I'm sure they'd love to connect with you on LinkedIn as well. So thank you very, very much. And we'll be back in about 5 minutes with our second panel. Thanks, guys.