In the AI Startup of the Week, the editorial staff of ai.nl is featuring promising AI startups, their innovations, solutions and challenges. In this seventeenth episode, we are taking a look at Lalaland, an Amsterdam-based startup building AI generated humans for fashion e-commerce.
Digital fashion has taken centre stage on social media and at fashion weeks around the world. With the rise of this digital fashion, one question doing the rounds is on the effect that the virtual world will have on representation, visibility, and even identity. One company trying to bring greater representation and diversity to the digital fashion world is Netherlands-based Lalaland.
Founded by Zimbabwe-born Michael Musandu and Ugnius Rimsa in 2019, Lalaland creates artificially rendered, hyper realistic models for companies instead of the real ones. The ability of Lalaland to build these hyper realistic models that look as good as real ones makes it one of the must watch AI startups in the world.
Challenging traditional photography
The fundamental work that Lalaland is trying to do is challenge traditional photography that has remained prominent in the field of fashion and e-commerce. In an interview, Musandu told culture and fashion magazine HUNGER that traditional photography has become one of the big problems for e-commerce companies and a limiting factor for showing a diverse range of models.
“Because it is such a physical process that requires human models, make-up artists, hair stylists, venues, it can get expensive for brands to shoot a spectrum of diverse models to show customers different segments. This is why they often end up using one or two different models,” he says.
To solve this problem, Lalaland is not only doing away with traditional photography but aims to do with elements like human models, venues, etc. The startup is building cost effective models that all types of customers can relate to and one they see as the source for wider representation and diversity in the fashion industry.
“Our mission is to make sure that Web3 is a diverse place in terms of avatar creation,” Musandu explains in the interview. “We want to empower that and ensure that you have a paintbrush that you can use to create different complexions, size categories and whatever else you want to customise.”
To that effect, Lalaland introduced its newest product, the Genesis Meta Model Creator, at CES 2022. With this new product, fashion brands and retailers will be able to create highly personalised and inclusive shopping experiences for their customers.
The idea that Lalaland is bringing to the table is also already finding takers in the form of investors and fashion retailers. According to Crunchbase, Lalaland has raised a total of $2,42,000 and is being used by the likes of Tommy Hilfiger, Zalando, Steiglitz, Wehkamp, and others.
How does Lalaland work?
Lalaland’s product is all about helping e-commerce players and fashion retailers build instant fashion catalogues without having to spend a lot of money and resources on traditional photography. It does this by offering a self-service platform where users can configure the avatars they need in just a few minutes.
The users of the platform get the option to customise “size, body type, body shape, ethnicity, hairstyle, identity, pose, and even the level of happiness.” Once the avatar is finalised, customers simply need to upload images of their product SKUs in bulk. This can be either a digitally designed garment or a product packshot.
The next step requires users to style their products based on the avatars created by them and garment uploaded. By styling the product, Lalaland users will also be able to create different looks and styles.
Once the styling is done, customers can export the output images, including an option to download 8K high-resolution ready-to-use images. The startup uses artificial intelligence (AI) to develop its highly realistic digital photo models for clothing and fashion brands.
In order to generate images of artificial humans, the company also relies on neural networks, which are systems inspired by the biological neural networks constituting animal brains. These are subset of machine learning and are at the heart of deep learning algorithms needed to build such diverse yet human-like digital models.
A subscription model
If a photographer is accustomed to using Adobe Photoshop or Adobe’s Creative Cloud then Lalaland’s self-service platform could reach similar stature in the fashion industry.
In order to get there, Lalaland is not only building these hyper realistic models but also offering it as a subscription service. While the pricing is not listed, it does offer four different tiers. There is a professional tier where users can access the platform to build 150 images per month while the advanced option allows to bump up that number to 600 images per month.
Lalaland also offers a once-off tier where customers can choose to pay on a per image basis. This tier, designed for freelancers, startups, and independent designers, offers limited access to the platform. There is also a custom tier where Lalaland makes fully customised packages for the needs of retailers and high-volume apparel brands.
What are the advantages of Lalaland’s self-service platform?
As mentioned earlier, traditional photography being used extensively in the field of fashion and e-commerce requires a lot of time and resources. The photoshoots used by these brands are complex, time-consuming, and expensive since they involve human models, photographers, make-up artists, stylists, and venues.
Lalaland says its Genesis Meta Model Creator brings 90 per cent cost reduction compared to conventional photoshoots. The biggest advantage is, of course, the lower cost associated with building and using the digital models offered by the fashion tech startup.
The second advantage comes in the form of the ability for e-commerce and fashion labels to shorten their time to market. It also allows brands to sell their product even before they produce since the models themselves are hyper realistic and the return rates are thus lower than conventional methods.
“With our virtual fashion models, we enable brands and retailers to go to market faster than they ever have, increase average order values, and tackle one of fashion’s biggest problems, returns. If there was ever a time to experiment with technology like this, it’s now,” says Harold Smeeman, co-founder and Chief Commercial Officer at Lalaland.ai.
Lastly, Lalaland helps fashion retailers reach their sustainability goals. It is a known fact that the fashion industry is one of the largest polluters in the world and with a circular fashion economy taking shape, Lalaland offers the means for fashion brands and retailers to ensure fewer returns and lower their carbon footprint.