It only takes a few hours of using the MidJourney image generator to realize the extent of the threat looming over a series of trades and sectors. Mostly advertising illustrators, filmmakers, comic book writers, graphic designers, architects or even photographers… Anyone involved in visual creativity in one form or another will have to adapt quickly or perish.
MidJourney is what is called generative artificial intelligence. There are others, such as DALL-E, the OpenAI version, or Stable Diffusion, but they are less efficient. Like ChatGPT, which indiscriminately amassed billions of words, MidJourney was filled with hundreds of millions of images of all kinds, some public domain and others copyrighted. It has everything: landscape photos, comics, portraits of great photographers, news reports, drawings by architects or the work of famous illustrators.
To create images from this happy mixture, we use “hints”, that is, descriptions expressed in everyday language, with some technical specifications. In order to generate the image at the top of this article, which should depict an illustrator wondering what his future is (and he’s not very young, either), the hint sent to MidJourney was: /imagine: illustrator’s realistic photo, work at my desk, pencil in hand, outdoors, dramatic light, noir style.” Nothing complicated. You can ask the algorithm to suggest options, change the lighting, change the depth of field to give a photographic rendering, or change the character.
Another example: to generate the photo below, MidJourney was asked for a “realistic photo of an older Asian woman, wrinkled, with a bun, smiling, in black and white, Tri-X style. [référence à une pellicule argentique]on a black background.” A pro would add specs to improve rendering. But the result is already interesting:
This image of an Asian woman was created with the help of MidJourney.
© / MidJourney
We’ll use the same process to create a sci-fi illustration in the style of Avatar, a house inspired by Norman Foster or Frank Lloyd Wright overlooking the Mediterranean, or even a giant abandoned factory in the middle of the desert. No less impressive is the creation of a logo, where we indicate what it should embody, in what graphic style, etc. Up to the point that the question arises, who will still pay for the corporate identity of the company …
In the past, there were hinting experts like Photoshop or Illustrator virtuosos. They are fluent in the grammar of the system, either to create complex images of virtual worlds, or, conversely, to create images that resemble a photograph taken with an ordinary camera. This month, Photo-Answers published this image on the cover, indistinguishable from a professional photo:
This image was generated with MidJourney by Answers-Photo magazine.
© / MidJourney, directed by Thibaut Godet, Answers-Photo magazine.
Its author, editor-in-chief of the magazine Thibaut Godet, made about 300 versions to achieve the desired realism.
Alexandre Saudinos, co-founder of Parallell Cinéma and an expert on new film technologies, is preparing an ambitious feature film. This is a medieval saga, set out in an ultra-accurate “bible”, setting out its plan in great detail: weathered faces, equipment of the time, menacing buildings, the gloomy atmosphere of the Middle Ages. This document is necessary to sell the idea to the distributor, to formalize the work of screenwriters, director of photography and, of course, the director. “A few months ago, I would have called in specialized illustrators, concept artists, whom I would have had to instruct in detail to infuse them with our ideas, with countless trips back and forth, a complex process, long and expensive,” Saudinos says. There, in a few days, I created the entire environment for the film with MidJourney. Everything was determined by several dozen tips and hundreds of attempts, describing in detail the appearance and even the personality of the heroine to embody a free and courageous woman – with facial features, skin color … We literally shaped the characters with the help of descriptions made in MidJourney”.
Another interested sector: advertising creation. Agencies are often the first to implement these new tools. They hurried to MidJourney. “This is a profound change that is happening very quickly,” explains Ivan Bechkovsky, co-president and creative director of BETC FullSix, which uses MidJourney extensively during the design stages. But we can’t do anything with these images. them because we cannot guarantee originality. On the other hand, we use them to create models or moods”, referring to the study of the visual environment, the atmosphere of the brand. One of the clients of BETC FullSix is Disney, for example, all Marvel characters are present in the corpus of images used to train the MidJourney algorithm. “If we want to create a Spider-Man that looks up to the sky in a certain environment, that works really well. The red color of the cape won’t be right, but we’ll deal with that later.”
Honorary Badge of Intellectual Property
For now, MidJourney and similar services are admirably ignorant of any notion of copyright. They capture everything possible in their networks. And the volumes are dizzying. The public database LAION-5B (for a large-scale open network of artificial intelligence) was created by a German non-profit association. It contains 5.8 billion images, 3 billion of which contain descriptions in over a hundred languages. This type of data set is constantly expanding: in one year, the size of the LAION database has increased fourteen times!
Needless to say, LAION did not separate public domain photographs or illustrations from copyrighted works. Moreover, when we dive into the depths of these giant databases that are used to study algorithms, we find that they have absorbed the best of the world’s photographic heritage. The ability to ask MidJourney to generate an image in the style of Ansel Adams (the great American landscape painter) or a portrait in the style of Richard Avedon comes precisely from the fact that these two great photographers have seen their work, yet are jealously guarded by armies of lawyers completely captured in LAION.
“We can find elements of the style of photographers in an environment that is completely alien to them,” says Philippe Duran, photographer at Answers-Photo. “If you apply the style of Edward Weston to the buildings of Manhattan, you get the same softness as in his still lifes.”
Some photographers manage to limit copyright infringement by making MidJourney not allow their name to be used in the tooltip. This is the case of Steve McCurry, the great National Geographic photographer, whose style is basically impossible to replicate. Except, notes Philippe Durand, that it is hardly difficult to reproduce the parameters of the famous photo of a small green-eyed Afghan girl.
So we’re reversing the rule of copyright: instead of a broadcaster – like the LAION image database or a generator like MidJourney – asking artists for permission, it’s them who are asking to have their work removed from the great vacuum cleaner. When asked by the American site The Verge, David Holtz, the founder of MidJourney, admitted that at the moment there is no procedure for the author to exclude himself from the system, but that he is open to ideas is nice of him.
For Ivan Bechkovsky, the problem must be solved: “Everything will be organized: a company like MidJourney will sign an agreement with large image banks like Getty on a system that will respect copyright. In any case, this is not new: if today one of our illustrators drew a Sempe-style drawing, he would end up in court. In addition, we always ask the artist for confirmation when we want to draw inspiration from him for visual creativity; we’re not going to have fun playing with Jeff Koons without asking permission from his studio…”
But many are wondering about the future of visual creation, created algorithmically and drawing its “inspiration” from others. For Christophe Tricot, CEO of La Forge AI, which tests mobile app and website concepts for the healthcare sector with MidJourney, “it would be fun to explore the limits of a generator by teaching it its own creations; the system will inevitably self-intoxicate into a kind of chaotic mishmash”.
The head of the BETC agency FullSix remains convinced: “In order to avoid the trap of standardization, we will need the original creation more than ever. I think we need to invest in both artificial intelligence and silver photography,” sums up Ivan Bechkovsky. This originality, this authenticity will retain great value.” This is the only thing that will push a photographer or illustrator not to start their day with gin at nine in the morning.