The tool
ALL the pictures on this website have been generated by artificial intelligence:
– Aliens
– People
– Art pieces
– Museum building
– Website logo
I achieved that with MidJourney. It is an artificial intelligence program that generates images from natural language descriptions, called “prompts”. It is not an app but a bot that you can use within the server Discord. There are many other bots or AI apps that can generate images but I wanted to use only one.
The process
When you connect to the Discord server in order to use MidJourney, it looks like that:
At the bottom of the page you can enter /imagine and then your prompt:
As a result, you get 4 images like these:
Then you have 4 options:
– Ignore them if you don’t like them
– Reroll to get a new set of images
– Get some Variations by pressing V1 (or V2, V3, V4) for the selected photo(s).
– Get some Upscale(s) by pressing U1 (or U2, U3, U4) for the selected photo(s).
you want an upscale.
Here I pressed U3 to get an upscale of picture #3 and V2 for a variation of the picture #2:
Here is the Upscale #3 and as you can see under the picture I can make even more variations:
And here are the variations for the picture #2:
And you can repeat the process (reroll, upscale or variations) again and again until you get (or not!) what you want. Or try with a new prompt.
Easy peasy?
You probably think it’s easy, right? No it’s not. It is easy to write a short sentence and press enter. But it’s not easy to find the exact prompt that will give you exactly what you want. Not only about the subject of the image but also about the settings, the type of set, the light, the atmosphere, etc. Sometimes the photo looks perfect, until you see a horrible detail, like 7 fingers on a hand, too many teeth on a smile, a missing eye, a distorted part in the picture, etc. It can take dozens or hundreds of tries before you find the magical formula for a specific request. Especially when you want to get real photos of alien species that never existed!
WHAT ABOUT COPYRIGHT?
According to MidJourney Terms of Services, if you’re a Paid Member (which I am) “you own all Assets You create with the Services, to the extent possible under current law.”
An article from The Verge mentions a copyright complaint from artists, and a lawyer explains some important points: The suit claims that AI art models “store compressed copies of copyright-protected training images” and then “recombine” them; functioning as “21st-century collage tools.” However, AI art models do not store images at all, but rather mathematical representations of patterns collected from these images. The software does not piece together bits of images in the form of a collage, either, but creates pictures from scratch based on these mathematical representations. Another article from The Verge states that “The scary truth about AI copyright is nobody knows what will happen next”… Let’s wait and see…
As far as I’m concerned, if I made this website it is actually to demonstrate how problematic the use of AI generated images can be.
Text: Marie-Sophie Germain
Photos: screenshots from MidJourney