When Coronavirus hit Europe in 2020 and we suddenly and abruptly lost touch with reality because of the long lockdown months, our digital lives became extraordinary and we started craving the contact we had lost with reality.
Fashion was one of the industries that suffered most from this loss with reality, but it was also one of the industries that reacted quickly, producing digital presentations and runways.
Now that fashion weeks and trade shows are back IRL, though, you have a feeling they may have forgotten something, that connection with digital universes created during the lockdown and in particular with new trends that may be encouraged by the rise of Artificial Intelligence (AI).
Pitti Uomo opened and closed with an emphasis on real collections that at times almost verged on the banality. Yes, now we can finally touch the fabrics, if only they were used to make something a little more desirable and innovative, something that really surprised you and that genuinely pushed the limits of design, it would be even better.
AI based text generating tools may still be at the beginning, but AI image generators have become more refined and are creating mesmerising universes and imaginary products that seem immensely desirable.
Go on Instagram and you will discover entire worlds, vibrantly vivid narratives, hybrid characters, pastel colours or bright shades, surreal designs and amazing architectures created with the assistance of Artificial Intelligence tools.
Instagram account Artomaton creates scenes from an intriguing retro futurist World's Fair that never happened with detailed descriptions about fictitious pavilions and sections, including a floor dedicated to innovations in Texture Technology and an Artificial Botany Conservatory, where it is possible to explore the intersection of plant and machine. The descriptions accompanying the images are like short stories (and if you like what you see, you can also buy prints on its Etsy shop).
Neptunian Glitter Ball's account also features fictitious stories with very unique protagonists, colourful heroes and villains in extraordionary clothes and headgear, a gallery of irresistible characters, surreal and unreal, yet so incredibly credible, and bizarre locations such as the city of Cleansington, where the SoapyDuds laundromat is the destination of choice for anyone looking to rid themselves of the ghosts of their past.
AI Clothing Daily focuses instead on artificially generated clothing, posting them daily: so far, the most popular designs remain the sneakers, hacked from pictures of runway shows, they seem to sprout tiny resin flowers or they are covered in lace, beads and delicate 3D printed swirls.
Fashion-wise there's more to discover: aidesign.png creates biomimicry-inspired wearable concepts, knitwear and outdoor padded pieces, while designer Agata Panucci did a few posts about fantasy collaborations with Nike or Adidas, coming up with crocheted or knitted shoes and garments.
Blancmange is instead a digital patisserie posting freshly baked artificially generated cakes that feature dreamy ingredients, from rainbows and clouds to alien planets and Swarovski crystals.
Such accounts are mushrooming (will we reach saturation point soon?) and at the moment many of them seem to actually generate images with a similar palette of colours and quite a few ones have also started accompanying the images with retro futurist narratives.
But the most interesting aspect when it comes to accounts posting ideas for garments and accessories and in particular pictures of AI hacked sneakers, is that quite a few posts receive enthusiastic comments by the users asking "Where can I order this from?" proving that the AI is generating desirable designs that humans are struggling to make.
Yet, there are issues involved with the use of AI: the debate when it comes to Artificial Intelligence-generated images is between those who state that AI will allow more creative freedom and those who claim that it infringes copyrights and may replace one day artists, illustrators, graphic designers and photographers in professional roles. Being trained on billions of art and images created by real people collected from a variety of sources mainly found online (including images shared on social media), apps like DALL-E, Midjourney, Stable Diffusion, Deep Dream Generator (DDG) and Lensa (for mobile phones) harvest inspirations from human-made art to generate an image.
The new image is created through the diffusion process: Artificial Intelligence is fed a series of images (taken from this knowledge base formed by billions of pictures) blurred with Gaussian noise, and it is trained to denoise them to eventually be able to generate a new image that shows connections between the textual prompt and the image created.
This practice of combing the Internet with a software collecting data from a variety of sources is called "scraping". Different apps rely on different datasets: Stable Diffusion and DDG use the open-source LAION-5B ("Large-scale Artificial Intelligence Open Network") dataset, but the creators behind OpenAI (DALL-E) and Midjourney, haven't made their datasets public.
Some artists have already complained about the AI generating images inspired by their works without asking for permission and have been talking about copyright infringement (you can also ask the AI to generate an image "inspired by" or "in the style of" such artist).
So you can bet that soon there will be lawsuits and class actions against the creators of these tools and one day we will probably also have laws and legislation against indiscriminate scraping (we will obviously have to train lawyers specialized in AI matters – so this could be a new career actually...).
And while the solution to all this confusion probably stands in creating databases of works licensed by artists who are paid a fee, for the time being, image generation remains unregulated.
Yet, it would be a mistake to ignore the power of Artificial Intelligence: when 3D printing first came out, many people were skeptical about its applications in fashion and laughed at it, but the technology developed a lot in the last 12 years, pushing boundaries and helping designers experimenting with innovative applications. Chances are that the same will happen with Artificial Intelligence: teachers may be scared about AI based text generating tools used to write essays, artists are worried about their work getting stolen and fashion lecturers that their students' portfolios may entirely be made with the help of Artificial Intelligence, but AI tools are here to stay.
What you can do with these applications? Artificial Intelligence should be used as a facilitator and a tool to create variations or elaborations and develop new layers of connectivity between different disciplines, fashion included. It can be useful when you're in a creative rut and to have fun.
Looking for ideas for a fashion collection? A basic prompt such as "Autumn/Winter 2050 ready-to-wear collection grey yellow and white outerwear designs hyperreal full figure detailed sculptural silhouettes with well defined shoulders" will produce architectural outerwear designs that do look derivative in some ways (yet, isn't modern fashion largely derivative? Isn't the history of fashion full of copyright infringements and of designers stealing and borrowing from other designers and remixing tons of inspirations without being accounted for it? NB With this statement we aren't encouraging copying or infringing copyrights...), but that also feature some improbable constructions that may lead a designer to come up with more experiments. Looking for inspirations for a runway invitation? Write in your prompts "fashion model paper crafted, pop up style" and you will get a pop up model that may be reworked into an idea for an invitation.
Another interesting way to have fun with Artificial Intelligence is to learn how to use adjectives ("hyperreal", "realistic", "extremely detailed", "dynamic", "surreal", "well-crafted", "dystopian", "utopian", "unreal", "cinematic", "photorealistic", "otherworldly", "maximalist", "elegant", "intricate", "incandescent", "shimmering", "ultra-detailed" and "volumetric"...) in imaginative ways: the more intricate descriptions will indeed produce more unusual and richer images and will generate the most unexpected results.
So use AI for the time being to create fantasy narratives, to enhance your work and find your own path to come up with something more original, and always be aware of the implications behind the blurring of creative and ethical boundaries.
Comments