Midjourney is freaking magic. On first glance it’s flawless, and only getting better. There are a lot of reasons these beautiful bits of absolute witchcraft should be used as a basis for your final work, or as inspiration. And even some ways to use the images as-is to help get your concept across. So let’s take a look at how you can harness this tool to enhance your creative process.
First off: What is Midjourney and how do I use it?
In less than a year, Midjourney has gone from a bit of a joke to HOLY FUCK. Released in March 2022 (Happy Birthday, robot friend!) on Discord with open beta starting in July. Today you can still only access it via Discord, which makes it a little bit more complicated than going to a simple webpage (like you can with Dall-E).
So let’s get started. First, you'll need to sign up for a Discord account if you don't have one already and download the app to your phone or computer. Then go to Midjourney and click the link to get an invite to join their server. Once you’re in, you can read about getting started in one of the getting-started-channels or just head over to a newbie channel, and dive in by using /imagine, which will pull up a prompt window. Then all you have to do is type in what you want to see.
I did that and my image sucks and I hate it.
Good. If your first image output is amazing then you have nowhere to go from there, right? The reason I’m not giving you a ton of prompting tips is because I would suggest you go play. Watch what others are doing, and copy bits of it if you like (or search for prompts if you really want a basis and copy them BUT DO NOT PAY PEOPLE FOR PROMPTS insert gif of little girl shaking her hairbrush here).
After you enter your prompt, you’ll get four versions of what Midjourney thinks you want to see. From there you can upscale one (or more) of them (U1, U2, U3, U4), which will give you a more detailed and higher resolution image of your selection. Important note – selections go clockwise so 1 & 2 are on the top, 3 & 4 on the bottom row. You can ask for variations on any of them (V1, V2, V3, V4), which will give you four new images based on the one you selected with slight differences. Or you can reroll the whole thing (the little arrows button), basically asking it to try again.
What I like to do if I don’t get at least one thing I like on the first go, is to adjust my prompt by adding more context or even taking some of it out, if I see it’s getting confused.
Can I get a high six...or seven?
Brb – using this as a basis for a painting.
His hair is perfect, his ears and hands and shirt are all a bit of a wtf, but still – great mood.
Besties share everything, even fingers.
K, but where do I use this in my work?
There are a lot of ways to use AI generated images, from Midjourney or any image generation AI. Here are some of the ways I use it:
Mood Boards: Trying to get that look and feel you have in your head just got easier. Instead of looking through stock and artist sites, you can now brief your friend directly. You’ll likely still have to change some colours, or touch it up a bit, but you should have a lot less to rework then before.
Character Development: Now you can create everyone from personas to show who you’re trying to reach to who you want for the shoot. It’s also incredibly creative at wardrobe and styling choices (which can be a good or a bad thing, depending on how you roll). And it’s getting better at this with each update. People are going nuts right now on how version 5 is photo realistic, surely the next one will be even better.
Design Inspiration: This one was actually a bit of a (good) surprise to me – but everything from colour palettes, to actual layout choices can be taken as inspiration for your own layouts. I wasn’t expecting it to be as good at this as it sometimes is, and maybe it’s partly luck or my own taste, but I’ve absolutely been turned on to new directions in my own designs.
Illustration and Collage: Need a li’l something to help get your idea across in your deck? Or want to sell in a look? It’s shockingly good at illustration as well, not just photographic looks.
So that’s it, I don’t need to do anything but put in some words, right?
Oh no, my friend. That is wrong. There are some serious limitations between your prompt and releasing files. One is the nebulous legal territory (but your legal department will be better at keeping you up to date on that than me). Some others are that your output will likely contain:
Terrifying smiles with an abundance of teeth and lifeless eyes.
Fingers that leave you questioning Midjourney's understanding of human anatomy.
Unfinished details, like it just got bored in certain parts and was like “yeah, that’s good enough, what’s next”.
Any kind of type. It’s absolutely wonderful to me that AI type looks like pretty much exactly how type looks in my dreams. It’s like an idea of words rather than actual letters and while I find it super charming, your client likely won’t.
I hope this has gotten you excited to go out there and PLAY. Find what works best for you – or even if it does work for you at all. Maybe you’ll hate it and dedicate yourself to using 100% human-made everything all the time. That’s fine as well, but know your enemy at least and go give it a shot.
And again as always: Midjourney or any other generative AI tool is not a replacement for you. It’s another tool for you to pour your ideas into and get results that can be a starting point to bigger and better things. The idea – your idea – is more important than ever.
Every bit of image deliciousness you see here was made with Midjourney and not edited at all. If you want to see more of this, the Robot is Friend Instagram is chock full o’ Midjourney magic.
I'm not counting them, but it seems like a lot of teeth. On top. Not so much on the bottom.
HOlCT blo!o, li'l puppy.
The bigger the hair, the closer to goddess
Yes to this styling, maybe to this hair.