The Art of Midjourney

January 19, 2023

To say I’m obsessed with Midjourney might be an understatement. For those not in the know, Midjourney is an AI (artificial intelligence) art generator that allows you to input some kind of text prompt — like “underwater house” or “dog playing with an iPad” — and it will generated that image for you. It doesn’t find an image or piece together things from the internet, it creates a one-of-a-kind original artwork.

I’m not going to get into the controversy surrounding AI art and what it means for the art community. That’s not really my place. But I do like that I can get an idea for what something could look like without having the ability to create the art myself. While I am a designer, I cannot draw, paint, sculpt, or otherwise anything traditional. (You might say I can barely color inside the lines.) However, I can imagine what something could look like and now I have a way to express it, sorta.

Like what, you ask? Here are some interesting pieces of artwork Midjourney has created for me and the associated prompts:

a photorealistic wide angle shot of the back of a white 12-year-old girl basketball player with the number 10 on a blue jersey with straight brown hair shooting a basketball, cinematic lighting, 4k, shallow depth of field
cute tan chihuahua yorkie with floppy ears running in a green field, pixar style, cinematic lighting
dogs, star wars style
undersea inspired adolf loos house

None of these existed until Midjourney whipped them into existence, which is truly incredible! Go ahead, do a reverse Google Image search and see for yourself 🙂

What I’ve Learned

Okay, so I’ve been playing with Midjourney now for a couple of days, so I’m not an expert by any means! However, here are some interesting things I’ve learned.

  • The more descriptive the prompt, the closer to what you’re looking for you’ll get. I love the Star Wars style dogs above, but the type of dog was chosen for me and what “Star Wars style” is could have gone a lot of different ways. If you want to be surprised, keep it short. If you know what you want, be descriptive.
  • It’s not perfect. I get pretty good results most of the time, but occasionally it struggles. Try to tweak your prompts slightly and be extra descriptive if you are looking for something specific. Or just retry the same one 🤷‍♂️
  • I love cinematic lighting. Midjourney can create all sorts of art, but I’m drawn to very moody and cinematic styles. Most of my prompts do indeed include this term.
  • You can use camera lens settings to control the outcome! For example, I’ve done some where I set 35mm or 50mm to get a certain look and it works!
  • Play with your inspirations. Whether it’s Star Wars style, Disney style, MC Escher, Tron, Monet, Picasso, Gaudí, Frank Lloyd Wright, or so many others, add those terms to help guide the AI. You’ll get some really cool results.
  • If you pass in links to other images, the AI will use those to help generate an image. It’s mind-blowing 🤯

There’s so much more I could say here, but I want to keep it brief. Maybe I’ll do something more in-depth in the future, but I mostly just wanted to document some initial findings (and keep my 30-day publishing challenge going!).

I’ll leave you with this generated masterpiece:

Me, as a baby

Wanna chat about this article or any others? Feel free to DM me or mention me on Twitter @marcusdburnette to start a conversation!

Leave a Reply

Your email address will not be published.