One revolution can hide another. Unlike conversational software like ChatGPT, artificial intelligence (AI)-powered graphic design platforms have not been the subject of a planetary buzz. But the increasingly successful mastery of generative technologies in the field of images might change the world just as much as the prowess of new chatbots.
“Everything that is graphic will no longer go through a human, the AI will be able to generate the perfect image for an e-commerce site, a cooking recipe or a video”, warns Stéphane Mallard, author of the book “Disruption”. published by Dunod. Even if the technology is not there yet and it still relies on instructions written by a human.
An image created from scratch
Nine months following the first public tests of its platform, the Midjourney laboratory has just unveiled the fifth version of its algorithm. Like the previous ones and the software of its rivals OpenAI or Stable Diffusion, it creates an image from scratch from a written description (for example “a car by the sea at sunset”). Among other things, the V5 offers “significantly higher image quality”, “a wider stylistic range” and “support for seamless textures”, according to the American start-up’s Twitter message presenting its new tool. .
As with ChatGPT, the stakes are high for multiple professions and across multiple industries. Already, initial experiments are opening up immense prospects.
Generation of videos
In the world of online , the simple and rapid generation of images, especially video, is of interest for social networks. A few weeks ago, the French teams of Meta worked with Peugeot to add some AI-generated intro shots in a short video touting a new car from the brand.
More broadly, artificial intelligence might facilitate the creation of video content by users and advertisers, while social networks are increasingly promoting this format, at the expense of photos and text. Even if questions of copyright and especially of moderation arise around a technology carrying a risk of identity theft.
The front page of Time
archyde news agencies are already seizing the phenomenon. For its client Nestlé, Ogilvy thus used OpenAI’s Dall-E technology to pull off a coup: create, in the style of the painter Vermeer, a whole set around the Dutch painter’s famous milkmaid who serves as an icon for the Swiss brand. Another example, Coca-Cola’s latest video spot, mixes real shots and images from artificial intelligence throughout.
But the photos and other illustrations are not left out. Whether for advertisements or to accompany articles, the media is taking over technology. In the United States, Cosmopolitan and Time have entrusted magazine covers to Dall-E. In France, the tech site Frandroid (Ebra) is working on Midjourney to create new drawings by Elon Musk and other Tim Cooks.
Generative design in industry
The impact of generative AI on image will not stop in the world of media. In the industrial world, in construction, “it’s a new life for all visual creation”, notes Stephane Roder, the founder of the firm AI Builder. While acknowledging that his customers are only just beginning to take an interest in this part of the AI phenomenon, he has no doubt that they will come here for their plans and diagrams. “With generative AI, an IT developer saves 25% of his time, it will be the same for architects or engineers when the algorithms integrate engineering constraints and can be fed with company data”, continues- he.
Professional software publishers are already beginning to integrate these technologies. 3D modeling specialists like Autodesk and the French Dassault Systèmes have been talking regarding “generative design” for years. Adobe, the champion of graphics software (Photoshop, Illustrator, AfterEffect, etc.) also gave pride of place to AI last fall during its last conference.