My journey with Midjourney AI - Author Kevin Macklin

I didn’t know what I had planned on doing when first getting started on Midjourney AI. Seeing what it could do and learning how to do it was all I could think about. Posts showing these artificial intelligence renderings have been floating around social media for a while now and the degree of detail instantly intrigued me. Detailed images of the firmament and interpretations of gods left my mouth hanging open in astonishment. I’m sure you’ve seen them. Ultra-realistic, 3D images with enough realism to make graphics teams in Hollywood nervous. I mean, these images are amazing! I even ran across an alien that looked so life-like I was waiting on it to speak. As someone who’s horrible when it comes to translating thoughts to art, but fairly adept at entering commands into a prompt, the thought of telling a computer how to render my thoughts into images was obviously something I wanted to look into and learn more about.

After a quick Google search I found Midjourney AI’s website fairly easily. On the homepage there’s a tab to “Join the beta” as well as a tab to sign in, but since I’m a newbie and don’t have an account I joined the beta, which took me to Discord since I was using mobile. I’ve never used Discord before and didn’t have the app and I hadn’t even been sure you could use the AI on mobile. Fortunately, I was able to download the Discord app and sign up. After creating a Discord account I had to go back to my browser and click the “Join the beta” tab, and after signing up for Midjourney I was redirected to the Discord app and Midjourney’s server.

Since I had never used Discord before, I had no idea what I was doing. But as I’ve said before, I’m a fast learner. For those who have never used Discord before, there are a bunch of message threads which kind of reminded me of the chaos of Twitter. Curious by nature, I let my curiosity flow and after a bit of searching I found a thread with instructions on how to use the AI. A new account receives twenty-five renders before you have to sign up for the paid subscription and newbies are directed to the #newbie thread for their twenty-five renders.

After locating the #newbie threads, I joined one of them, and that’s when the chaos began. A rapid fire assault of images rendered by the artificial intelligence flooded my feed. It was beautifully overwhelming. Sitting there for a moment, I watched as stunning images and commands flowed across my screen. It was art created by code. People were working with machines to create all kinds of fantastical pictures, highly detailed and realistically rendered. I wanted to dive in.

I am not an artist and I do not have a lot of money. I’ve written some really good books and dressed them in some really horrible covers that I’ve had to make myself due to the fact that I can’t afford to pay a reliable, professional cover artist. My Jon Dough thriller series was in need of a serious cover makeover. I’m aware of all of the elements that must go into a thriller cover, but I don’t have the artistic ability nor the computing power to create them. And since my thriller series features a black main character it’s hard to find stock photographs of a black man in an action pose. I wanted to see if I could get artificial intelligence to create what stock photo sites don’t offer. I also wanted to create some visuals and character art for my book, Prince of Ruins.

Figuring that zombies would be easier to process than my black character in an action pose, I tapped on the box to enter my commands, but I think parameters is a better word to use than commands. You can’t simply enter your commands and expect them to run and be rendered. If you’re familiar with coding you know that you can’t type random words onto a screen and run a program. Commands are entered within or after tags that let the computer know what you’re trying to do before you tell it what you’re trying to do. For example, if I wanted to add an image to a web page I would have to type in <a href= followed by the file name or file location. The process for entering your commands on the Midjourney server is similar, but you begin with /imagine and you’ll see /imagine pop up above the text box. Tap it and it will appear in the text box. I noticed that the word “prompt” showed above the text box sometimes and other times it was automatically placed behind /imagine in the text box. If “prompt” appears above the text box, click it to add it to the text box. It’s part of the code and it must be highlighted in the text box in order for the code to run. I tried to type it in and it didn’t highlight, and as you guessed, the code didn’t process so it’s important. After /imagine prompt: you can enter the parameters for the image you want AI to render.

From watching other’s commands and the images they were creating, I noticed that some people were describing what they wanted in a sentence whereas others were entering what they wanted as parameters separated by commas. Both methods worked, so I guess it’s more of a personal preference than anything else. After trying both, I settled on a combination of the two.

The post-apocalyptic images were amazing and the zombies were rendered with a realism that was realer than reality. Thinking that I had a pretty good grasp of what was going on, I tried to create a black man. My first attempts produced a weird, dreamlike quality in which there were no details to faces. When the AI first runs your commands it produces an image that contains four separate images and you are allowed to choose one to upscale, choose one to create variations of, or you can respin the entire concept. After a series of edits to my commands, variation selections, remasters, and respins I was out of my twenty-five renders. They were gone in what seemed the blink of an eye and I hadn’t made it very far with what I was trying to create.

The potential to make exactly what I was looking for was beginning to reveal itself, but I had already drained my account. Which turned out to be a good thing. Pulling up the text box, I typed /subscribe and clicked on the link that popped up to take me to the subscription page where they offer two subscription models. The first cost $10 and gives you 200 renders and the second subscription model cost $30 and gives you unlimited renders, private renders through your direct messages so you don’t have to deal with the chaos of fishing through everyone else’s renders to find yours, and commercial rights to your images.

Making the mistake of getting the $10 subscription, my 200 renders depleted within 24 hours. But I was able to get what I was looking for. And more. Through a combination of variation selections, remasters, and respins I was able to get artificial intelligence to get me quality images that I could use as elements in my book covers. I was also able to create some amazing portraits and works of cat art. If you find that Midjourney AI is not giving you exactly what you’re looking for, just keep working with it. Edit your commands, respin and remaster, and don’t give up. Play around with it and learn how to use the ultra realistic, natural lighting, cinematic lighting, etc. parameters. You’ll find that they can completely change the feel and tone of the renders. I would suggest the $30 subscription and that you prepare to dedicate a significant amount of your life to your newly found artificial intelligence addiction.

If you’d like to check out Midjourney AI’s site you can click on the link https://www.midjourney.com/home/

Thanks for reading and hopefully my experience can offer a little help.

If you’d like to check out my book cover or read my book you can click the link https://amzn.to/3qvCHB2

If you’d like to check out some of the wall art I made you can click the link https://www.etsy.com/shop/ArtDeptbyMackMedia