A cyberpunk ibex created in MidJourney

MidJourney: Exploring AI image generation

Ever since Dall-E Mini (now known as Craiyon), AI-generated images seem to have become the latest fad on social media. I wasn’t really a fan of the quality of the generated images until the fantastic AI-generated artwork of a fellow Mastodon user caught my attention: Through them, I stumbled upon Dall-E’s competitor MidJourney and ended up successfully signing up for the closed beta.

How does it work?

MidJourney does not have a web interface like Craiyon, but instead relies on Discord. You can use the command /imagine to enter a prompt for it to process, and additionally add modifiers to define image ratio, color scheme and more. The AI is capable of emulating a great variety of styles, ranging from sketches over Dalí-esque paintings to nearly photorealistic imagery.

At first, if you enter a prompt, it will generate a mosaic of four image suggestions. You can either ask it to reroll, create additional variants of one of the suggestions, or upscale one of them to high resolution. Here’s one of the first things I created with it, using the prompt /imagine cyberpunk ibex:

Examples

In the following I will list a couple of examples I created with more specific prompts and modifiers:

Pros and Cons

Pros

  • A great way to gain inspiration for settings, backgrounds and creature designs
  • With a paid membership, your creations are licensed for personal as well as commercial use
  • It’s just hella fun to play around with

Cons

  • An endless image-generation source for NFT scammers
  • Possibly dubious sources of the material used to train the AI?
  • Q&A sessions and detailed information for members only

Some people worry that AI-generated imagery might replace concept artists at some point, but I doubt that any AI will ever be good enough to read clients’ minds and translate their ideas into an image 1:1 – something even the best freelancer struggles with sometimes. I do think that they may be useful to generate moodboards to brief concept artists, though, to be used as a base for them to create creatures or landscapes with a particular look and feel.

Since drawing doesn’t come to me as naturally as it used to when I was younger, I’m personally grateful for the opportunity to be able to create beautiful things regardless.

I feel like this is just the beginning of AI-generated art, so I’m curious as to what the future might bring. What a time to be alive!