SYou generate a spectacular photo or a sophisticated drawing in a few seconds without the need for human genius: image generators with AI require neither artistic nor creative skills, except that the user is imaginative in his commands to the machine. You simply enter instructions as text: “Spiderman, Donald Duck, Joker, Popeye, Superman and Wonder Woman salute the Pope in front of a backdrop of Catholic nuns. There is a church in the background.” A few minutes later, the software has generated four possible motifs and is displaying them in a preview image. You choose one, which is now displayed in high resolution. The result can be seen at the top left. Wonder Woman, Donald Duck and the nuns are missing, and the algorithm probably couldn’t do anything with Popeye.
The photo shows the strength and weakness of the image generators. Like the chat systems with artificial intelligence, such as ChatGPT or the new Microsoft Bing, the software is based on continuous training with material from the Internet. In this case, millions of images. The AI “learns” what a dragon or a dog is, what Superman looks like, and what a church looks like from the image descriptions. Learning takes place with neural networks and, as with ChatGPT and Microsoft Bing, the same applies here: The results cannot be relied on because the AI machines do not “know” anything, but work with statistical relationships.
What you enter as a command is at best generated that way
The photos shown here were created with the Midjourney software from the American research institute of the same name. The app can be used in beta software on Windows or Mac computers and it is not easy to start up. You need an account for the online service Discord and install Midjourney there as a bot. As with its siblings Dall-E or Stable Diffusion, you program Midjourney at a command prompt with text commands. What you enter as a command is at best generated that way. You can enter long, detailed phrases and specify what you don’t want to see in the picture. Anything that could become obscene in any way is immediately rejected by the machine.
Also, public figures cannot be integrated into the artificially generated works of art. We tried it anyway and said: “Robert Habeck and Volker Wissing are best friends on vacation by the sea.” The result is the two bearded men on a boat with seagulls flying over them. We assume that the AI didn’t understand either name.
Tried out
:
Image generators with AI
The command “The most popular German politicians stand by a swimming pool and smile at the camera” was also not carried out correctly. The software drew well-dressed people standing fully clothed in a pool.
In addition to the description, an art style and a mode of representation can be specified. On the left, for example, we gave “Horses are fortified at a gas station in the style of Edward Hopper”. It remains to be seen how well the painter Edward Hopper was hit here. The AI also reached its limits with the “robot dogs in a meadow, as Hieronymus Bosch would have painted them” that can be seen below.
Midjourney allows you to artistically alienate your own photos. You upload a few, and a URL is generated from the image data set, which in turn can be embedded in scenes. Because creating the photos and graphics requires a lot of computing power, you sometimes have to wait minutes for an initial result. The software can be tried out free of charge initially, but a subscription is essential for longer use. Midjourney can also be tried out on the iPhone with a limited range of functions with the “AI Art” app (for 23 euros per month).