This is the transcript of a session recorded at PI Apparel New York 2023. To view the video of the talk, go here.
So, I’m Nima. I’m one of the founders and partner at Maison Meta. So, before I get into the presentation, none of what you’re about to see actually exists. It’s all been generated through AI, and that’s the only work that we do right now. So, we’re probably the leading AI creative studio in the world. We work with a range of clients from Moncler to Zara, Revolve Group, H&M, Louis Vuitton. So, a whole list of clients.
This has happened extremely quickly. We started digging into AI about a year ago, and then realized the potential that it had. And really, all of this has happened in the last six months. What’s been really the unique component of this, and what really sort of differentiates what we’re doing with anything out there, is this custom workflow that we’ve developed. So, everyone’s seeing a lot of these AI memes on all sorts of social channels. And I think the hype of what can be created with AI is very different to the reality of what can be created with AI. And we have this challenge literally every day talking to clients who don’t know what can be done, what can’t be done, and how it works. Because if you’re a global brand, it’s one thing having super creative imagery that’s inspirational and concepting, but it’s another being able to replicate your actual products in AI, and then pull that into another image that’s also generated by AI.
So, the work that we do is not setting the background and then Photoshopping or copy pasting a product into it. Everything is done through AI. And then because we’re so deep in this space, we start to advise other organizations like Google and UBS. And actually, in September, we have a couple of courses coming out on Domestika, which is sort of like a master class competitor on how to use different generative AI tools like Mid Journey and Stable Diffusion and others. So, AI can be used for so many things across a fashion company, from customer service to product design to logistics, even to e-com or website optimization. Our focus is on sort of a slice of what AI does. And primarily, it’s about generating content and imagery.
So, we’ve worked on global advertising campaigns all the way to images for product launches on social media. We’ve also started working on actually generating fashion collections. I’ll show you some examples later of what we’ve done there. And then we’ve coined this term artificial creativity. So, this is the idea of allowing creative teams, design teams, interior design teams to be able to very quickly come to the same page in terms of what the vision is for a certain type of product, for a certain type of environment. So, rather than people traditionally having Pinterest boards and mood boards and all this, we can actually create what the vision should look like and that makes all the team work together much quicker. So, if you’re doing a photo shoot before everybody gets there, everyone has a sense of what the final output can look like and should look like. And even with one client who’s a celebrity sports icon who’s developing a cosmetics brand, we actually started to work with him on using AI to do branding and packaging product design.
So, the range is quite broad. We’ve also, because we’ve been really pushing the boundaries of what’s possible with AI, we’ve been behind a few of the first in the space. So, we did the world’s first AI generated campaign for Montclair Genius. I’ll show you some images of that. That was their collaboration with Pharrell Williams, Alicia Keys, Adidas Originals, Jay-Z, Fragments, and a couple of others. We did the world’s first ad campaign which went on our billboard. It was super, super challenging to upscale it to fit a billboard that’s sort of 48 feet by 28 feet, I think it was. And then we did the world’s first AI generated fashion collection with Revolve, which they produced and sold online. So, this is an example of the work that we did with Montclair Genius.
So, this was as part of their launch in London. I think it was around February time. So, again, if you look at something like this, I’ll say a couple of things which are really important. So, this was done in February. The technology has come much, much, much further. So, the quality of what you can produce is much, much more powerful right now, even in a matter of months. This project traditionally would have taken Montclair Genius probably four months of coordination, shoots, teams, agreements, paperwork. From start to finish, we did this in about four weeks. And that was from concepting, creative, and then actually creating the imagery.
And again, the difficulty here was the ability to actually bring their products and generate them through AI and then pull them into sort of the settings. That’s really the most difficult part of it. Here’s another set of imagery. Yeah. So, this was part of the first campaign that we did with Revolve Group, which is a big fashion tech e-com company. It was to celebrate their 20th anniversary. So, from there, we started talking with them. And Michael Mente, the CEO, is a huge believer in AI and a massive champion for the category. So, one of the advantages of AI is that we started with this sort of campaign. And very quickly, we were able to adapt it, upscale it, and sort of adapt the imagery so that it could be used on billboards. So, that’s one of the huge advantages of using AI, that you’re not stuck with whatever you produce at first. You can change it. You can adapt it. You can upscale it. You can have different people within there. So, one of the other things we hear a lot from clients is that, yes, we’re doing this campaign. We used all our budget. But our team in Japan or our team in Brazil wants it to be localized, wants it to feel more like the customer base down there. And so, in traditional shoots, that’s done and gone. You have to go back, redo everything, pay more fees, wait more time. With this, the ability for us to change the background, the imagery, the clothes, the person, the settings can be done in a matter of days.
And so, that gives the brands that we work with maximum flexibility in terms of where they can take it. So, these were some other images. And this is where they were on the billboards across California. So, everyone talks about, why should we use AI? What’s the value proposition? So, for us, the way we talk to everyone, it’s really a matter of three or four key value propositions. One is massive saving in cost. So, the budget that Revolve was able to save in generating these advertising campaigns, they were able then to put that money towards media. So, they were able to promote the campaign much more widely and on more expensive sort of territory than they traditionally would have been able to do. Obviously, speed to market. So, again, these were done in a matter of weeks. When campaigns are planned four or five months in advance, you have no control of how things change, how trends evolve, what new cultural elements come into play. So, by doing this, we’re able to go to market very quickly and really be on point with what the conversation is or what’s happening around the world. And then, I mentioned the other value proposition, which is really about customization and personalization and scaling. So, we could use this for social media. We could easily use it for outdoor advertising. And then, if Revolve wanted, we could customize it for a local audience in any country that they want. And again, all of this is done quickly, but also without all the necessary paperwork and other things that you would have to go through to get more rights, to renegotiate rights, to apply them, so on and so forth.
So, this was the collection that Revolve, in the end, produced. And again, this was a matter of how we can scale these projects into so many different places. The original idea was a campaign for their 20th anniversary at Coachella. Then, it became a billboard ad. Then, it became a wild poster campaign. And then, the idea was, well, why don’t we actually try to produce the collection and sell it online? So, you can see the jacket in the middle on the left and then bottom in the middle, too. And then, the other pieces were also from the same collection. So, this was the first time that an AI-generated collection was actually produced and sold online. And it was super, super, super exciting. And actually, it’s part of the Business and Fashion’s recent AI case study. If you haven’t seen the report, it’s actually really, really good. There’s a couple of pages about this, and then there’s a whole lot of other resources and ideas for you guys to look at. We didn’t realize how much attention this world of AI and fashion was going to get. There’s been a lot of attention across the board, and there’s a lot of questions around ethics, and are we replacing jobs, and so on. And I think these are really important questions to address. So, in terms of ethics, we always advise all our clients that they really should put up front the fact that they’re using AI to do certain things. And then, the press loved this sort of question of, are you putting people out of a job?
We look at it in the same way as how technology has progressed over time, and how it has adapted the way people do business. When the automotive came out, I’m sure a lot of carriage drivers were really scared that they were going to be out of a job. But in the end, new categories, new services, new jobs are created. And we see the same thing with AI as well. I think the main goal or the key conversation that we have with actually a lot of HR departments is make sure you upskill, reskill your audiences, because the AI on itself can’t do everything. And so, it’s the matter of how do you train your people, find the right people who can use AI to do all the things that you want to do, but be able to do them more efficiently, more quickly, and sometimes with better quality. So, besides the work that we do in the studio, there’s a couple of really other important things to mention.
So, back in January, we decided that we were going to put a stake in the ground and do the first AI Fashion Week. So, what does that mean? So, traditionally, obviously, designers have to create a collection, produce samples, hire expensive locations, try to bribe models to show their clothes. So, we decided to provide a forum. It ended up being at Spring Studios in New York, where actually New York Fashion Week takes place as well. And we did a competition across the globe. And we had submissions. We had over 400 submissions. I’ll go through some of the details later. But everything had to be done using AI. All the collections had to be purely done using AI. The three winners from AI Fashion Week, and we had a sort of a public voting and then a jury. So, the three winners have entered as the first cohorts a couple of weeks ago into the first AI Fashion Incubator that we’ve developed with Revolve Group.
And what that means is that the three winners are now able to get the support and the infrastructure and the massive sort of resources that Revolve Group has to offer for them to create their brand, do product development, sourcing, produce the collection, and then sell it online and not have to deal with anything besides creativity. And I think that’s been one of the issues around other incubator programs or designer competitions. It’s great that somebody wins a prize and they’re sort of named the next emerging talent. But then they have to go and raise money and figure out how to do product sourcing and make samples and deal with factories and set up an e-commerce shop and deal with logistics and customer service. So, we’ve sort of taken all of that pain away from the three winners with the help of Revolve Group. And so, we’re working with them to produce their collections, which will be out in end of August, September. So, just a little bit of overview about AI Fashion Week. And this is a great place to also look at the talent that exists around the world. We had submissions literally from Africa, Middle East, Asia, all sorts of places. There was over 400 submissions. About 1,500 people came through the doors at Spring Studios to look at the collections. For the public voting in the two weeks, we had over 14,000 people register on our app. Massive interest around the world to actually view what was happening. We had 1.4 million page views in a matter of a couple of weeks. And then the three winners were the ones who are going to the incubator.
So, these are some examples of some of the collections that were produced. And you know, in addition to the public voting, the last 10 went to kind of like an expert industry jury that included everyone from Pat McGrath to Matthew Drinkwater that many of you may know who’s this he’s sort of the OG of tech and fashion. We had Erica who’s the head of Adidas’s three-stripe studios, Michael Mente who’s the CEO of Revolve, Tiffany who’s the head of editorial content of Vogue Japan.
So, you know the collection the top three collections were picked from them and I’m showing you so this was the winner in the end so this is a collection that he made he’s an architect based in Portugal so another actually really another architect so all of this has been done with AI nothing is done in any other way purely through AI again the details so I was mentioning sort of the the benefits of using AI I think on one end because of the meme culture we’re in and because of social media everyone’s looking at these like crazy collaborations of like Pope wearing Balenciaga, but again I think really for the companies that we work with, for the brands that we collaborate with, it’s really about what it does for my business and you know how does it appear on my bottom line, not in one year’s time, not in two years time, but within a month’s period.
So reducing costs and being able to reinvest that or just save those costs is one of the biggest drivers for using AI. There is this sort of one-time flat fee for using the images so you don’t have to go back and do paperwork, and again the faster time to market it’s a matter of three or four weeks rather than months to do content. And, you know the sort of big halo campaigns is really attractive to a lot of people as they’re rushing to get things out of the door and there’s always last-minute requests both you know centrally but also globally once the images are created we can scale them across the board, different channels, different countries, different sizes, so that gives maximum flexibility in terms of leveraging your one-time investment to do many things.
And like I said you know, we had one beauty clients that we work with who had used all their budget to create a campaign but then their the team in China wanted a campaign that was more adapted to that audience, and so we’re able to do that for them in a matter of days once all the creative, once all the imagery, was done and we can make images we can change images all the way to the last minute literally. Traditionally that would mean going back to whoever was the talent resigning paperwork, potentially you know paying more fees. All of these issues are taken out of the door, so we don’t really need to think about those and again.
I think you know there are sort of these hard costs like budget for doing a campaign, but many people don’t really think about the amount of resources and time that it takes for one of these things to be done. We have a very small lean team but when we work with clients we start to see that you know there’s like 25 people from three different organizations on every call, we’ve no idea what some of them do, most of them have never spoken on the calls but it’s just matter of all the time and effort that it takes to coordinate between all these companies, hours spent on zoom.
Whereas with this sort of process it’s very binary. It’s let’s figure out the creative, let’s adapt it quickly, we can even workshop it with clients live, so they get a sense of where it’s going and then produce it in a matter of time. We have two clients, one is their apparel company based in the UK and the other one is a I guess a global fast fashion company, and for them reducing the carbon footprint, so the sustainability angle on using AI, is a big driver. They both have annual sustainability reports that they give to their stakeholders each year and so for them not to have to fly 20 people to Mexico and then to south of France to do these shoots saves not only significant amounts of money, but also reduces their carbon footprint.
The other the benefit is sort of speed of what we call artificial creativity. So this is where you can actually imagine what this campaign would look like and be able to get all your team to get on the same page very very quickly and get to the end point much quicker. So the real the real trick is about control. If I don’t know who’s used some AI tools like mid journey anyone mid journey darling okay has anyone used stable diffusion nice so you know with most of the AI tools out there you can have insane imagery and imagine you know the Pope on the moon, but if you want to take your product and put it on, it’s very difficult to do, so the workflow that we’ve created uses a lot of stable diffusion and what that means really is so stable diffusion allows you to train the AI to learn what a product looks like and then pull it into the AI imagery that you create.
It’s a bit like when you have a kid and you walk with them on the street, you show them the first car and show them the second car and by the time you’ve shown them the tenth car, they know what a car looks like and they even able to draw it. So, it’s the same idea of training the AI, giving it lots of references of what a product looks like so that they can so that the brain of the AI can pull it in and for that we don’t even necessarily need studio quality. For photography what we’ve done with many clients is either they take iPhone photos of a product from different angles, ideally minimum of 25, more would be better, and with those we can train the AI to recognize that product and be able to recreate it with some retouching.
It’s also important to have it in different contexts. So, in front of a brick wall, outside, again all of this trains the AI to recognize that product in different environments and then allow it to be pulled in. So, this is our workflow. How we start with a project from conception to finalize of a brief, just like you would do with any other creative team. Then we get into some AI exploration and experimentation, so getting a sense of where we can push it, where we can create the process.
Then once we have the vision in place and the direction all nailed down, then we can start to do the AI training. So that’s where we teach the AI what the product is, the color, the positions, and then from there we refine and finalize. There are always going to be some little bugs in AI. Again, don’t believe the hype that you’re going to get this pixel perfect image at the end. You’re going to need to do some retouching on some certain areas. That’s improving enormously but retouching is enormously important as part of the process.
This is a recent campaign that we’re doing for a British casual wear company. You can have these fantastical backgrounds and so on but it’s critical for the person to be wearing the actual clothes, you know the shape, the fit, the color. That’s what we can do with the workflow that we’ve developed and many clients ask for different versions of photo realism – some people want the imagery to look a little bit more exaggerated or a little bit more futuristic, which we believe should be the case with using AI especially that’s where culture is going if you look at sort of the references in popular culture coming from gaming and movies. It’s definitely that direction, but some clients like this client of ours, who’s a another fast fashion company, global one, that has a homeware brand, really wanted photo realism like I want to have a shoot as if I was having a shoot you know, what would it look like, and the products have been trained in AI and look exactly like the actual product that they’re selling in the store same here and you see how you can control the light if you look at the bottom of the glasses the shadows, the setting, everything can be absolutely controlled to the nth degree, depending on where you want to take the final image.
And, these again on the table are exact replicas of the product that they’re selling. It’s not Photoshop, it’s not copy and paste. It was trained and pulled in into the final imagery that we we created.
So, this was another project that we just completed recently. It was actually at the the Cannes Lions festival down in the south of France. This was another instance of a luxury bag company based in Paris, again the bags that you see, the white one and the other color one, they’ve been generated in AI with us training it and it looks exactly like the product that they have. This also allowed us to go down this other path of what we call generate developing AI catalogs, so that once that product is trained and the AI knows exactly what it is changing colors and textures and so on, it is very very easy and so this allows us and our clients to have this AI catalog and then be able to generate imagery very very quickly afterwards and use it for whatever purpose they want.
This was another AI generated replicas. Again nothing that you see here exists, it’s all been generated by AI and again you see the textures on the left, so you can apply textures, apply shapes, colors, shines, all sorts of stuff. This is our techniques applied to beauty but this is all part of a beauty campaign so this was the example I was mentioning about artificial creativity.
This is not the real product campaign but if you are familiar with this one you’ll see the similarities with it. So what we can do is before the shoot, develop this sort of imagery with the help of the creative director, with the help of the artistic artistic director. So when people are on location they know exactly the image that they’re looking for, then you know the angle we’re looking for, the time of day that we’re looking for, the setting that we’re looking for. So, all of these things can be done through AI because we don’t believe and we don’t want AI to replace everything.
Brands like Prada will still do their halo campaigns using famous models and famous photographers and famous creative directors, but AI can help the process be much more smooth and efficient. Again, rather than one showing up with mood boards and inspiration photos and so on, they can show up with this and and find a way to recreate it, and then you can take the inspiration in other ways. You can create your own collections with inspiration that comes from any part of the world and even take it one step further.
This is sort of a dissection of no furniture you can create all sorts of imagery as long as you know how to use the tools and sequence the work in that way.
I think I’m at the end if anybody has any questions I’m happy to answer but thank you for sitting through this you. I’m here if you have any other questions but thank you so much for your time.
