What happens when you go from ‘computer aided’ to ‘computer generated’ design? New web-based artificial intelligence software that creates images from text input alone has been catching the attention of creatives around the world but does it have a place in yacht design? Andrew Johansson investigates...
New creative artworks have been appearing online in recent weeks that have been created in a way that is more akin to a science-fiction movie than reality. A handful of developers have been hard at work building artificial intelligence platforms that offer the creative community new opportunities. The two main contenders at the time of writing are Dall-E and MidJouney. These platforms offer a similar service with differing results but the principles remain the same. The web-based programmes create visuals by entering a simple sentence into a search bar before returning four images in a matter of seconds. While the results can vary significantly, some can be truly extraordinary.
“It forces you to verbalise your design ideas in an efficient and articulate way. Is it an Art Deco railing or is it an Art Deco railing with ornate flower metal work and a polished brass handrail? In short, you could call it targeted unpredictability,” says designer John Cantu of Cantu Design, the studio behind the in-build 38 metre Garfield, Westport Yachts’ W35 and the refit of 30 metre Redsky.
The American designer has access to both platforms, focusing more on the capabilities of Dall-E — created by OpenAI — and its potential to change the way in which design studios collect inspirational material, create moodboards, explore themes and shapes, and even begin the design process for the exterior or interior spaces of a yacht.
“We’ve been following Dall-E since its early development,” explains Cantu who was granted access to the V2 beta version at the beginning of August 2022. “We have been impressed by how this software thinks and creates art and believe it will be a powerful tool for every creative industry and probably many more. The OpenAI platform can also write articles, movie scripts and so on. What you end up with is a creative content generator that, paired with the right person, can produce incredible results that are unique to your description. There will undoubtedly be sceptics and those who won’t see the value in it but I think the majority of people will view it as an additional source of targeted inspiration.”
Cantu goes on to explain how initially the results he was getting were mixed. However, the more searches were performed, the quicker he began to understand what terms yielded the best results. “We were able to get it to draw a wide range of yachts, from the practical to the most audacious ideas,” says the designer. “It’s interesting how you can focus on the smallest detail or zoom out to a bigger picture.”
Jana Kleine-Kalmer, a freelance interior architect and artist who has worked in collaboration with Cantu Design, agrees. She explains that the devil is in the detail and depending on the description, you can get very targeted results. The more specific the description, the better, with the same being true on MidJourney. However, you also have to understand how the AI software interprets the words and structure of the sentence.
“I was curious to see how Dall-E would react to the complexity of interiors, with the countless different elements that we take for granted in any given environment,” says Kleine-Kalmer. “When we asked for a hotel bedroom, it would focus on the most common denominator of the space, such as a bed or table within a room. It did, however, create very detailed sequences for interior design styles or architectural periods. Results got more creative when we fed it with art history influences or contemporary styles like digital art.”
Dall-E compiles results by pulling information from the web and presents it in a new format. Those with access can use the material created with limited restrictions, as OpenAI stipulates that all images must be credited to their AI but does not assert copyright status over the images so the creator can use them as they see fit. However, as the technology draws on existing content, it is plausible to think that certain creations might potentially come close to an original. It is an aspect that OpenAI appears to be aware of with the inclusion of a complaints procedure for such an eventuality. Dall-E, which at the time of writing, is in the beta stage as the developer continues to refine its artificial general intelligence.
“Designers often make use of web-based content in the process of gathering inspiration as part of a number of sources,” says Cantu. “Whether that be from old French furniture books, walking through a gallery, Pinterest or design magazines, nothing is created in a vacuum. AI art is a unique avenue to this same process."
It is a point that creative director at Thirty C Rob Armstrong agrees with and like Cantu has been exploring the possibilities of Dall-E and MidJourney, focusing on the latter primarily. Working in collaboration with Mahdi Shirzadkhan, director at MS Yacht Design, the duo has had similar experiences. However, one of the main differences between the two platforms is that MidJourney allows the user to instantly refine and further develop the four results it delivers endlessly with a single click. The AI is also being developed with the ability to use your own imagery from which to create new visuals.
“MidJourney sources images from across the web, which it draws upon to build your AI creation but the platform is also changing so you can feed it your own images. I could get it to create something new but based on my work rather than sourcing from other things,” says Armstrong. Thirty C’s portfolio includes the exterior refit of 56 metre Galaxy in collaboration with Njord by Bergman Design House and the soon-to-be-delivered Navier 27. The English designer goes on to reveal the parallel between these AI platforms and a technique he has used for many years to generate new ideas using watercolours on paper.
“I’ll get some really basic lines down and work it up from there,” explains Armstrong. “The paint is doing the AI and you are seeing what’s within it, drawing on your knowledge or imagination to make something from it.” Using this method as an example, you can quickly see how the imagery created by AI platforms could be used and developed into real conceptual projects.
“I’ll get some really basic lines down and work it up from there,” explains Armstrong. “The paint is doing the AI and you are seeing what’s within it, drawing on your knowledge or imagination to make something from it.” Using this method as an example, you can quickly see how the imagery created by AI platforms could be used and developed into real conceptual projects.
The limits of these platforms seem endless once you understand how to extract the most from them. However, how the market will view it and explore the ways to get the most out of it, will be key.
It comes as no surprise that free but less powerful platforms such as Starry AI, Dream by Wombo and Nightcafe have been released. However, the breadth of possibility and the results that you can achieve varies. With Google also rushing to join the market with their own platform called Imagen, it’ll be interesting to see how the market develops.
However, there are also those for whom this technology is not yet refined enough to be useful or, as co-founder of design studio Nuvolari Lenard, Dan Lenard explains, it is simply not a tool he’d choose to use at this time. “Software with true AI is something we are far from at this point in time,” says Lenard. “It simply cannot express the emotional component in the creative process. Our studio has always got more ideas than opportunities to build them — we have to reduce these to a final and ultimate proposal. We definitely don’t need any more ideas.”
The power of these platforms is extraordinary. Looking at the yacht sector, only time will tell if it will become a beneficial tool. Seeing how these AI platforms might be utilised once officially launched will be exciting to witness.