As of this moment in time, if you were to mention the term artificial intelligence or AI to someone, you’d likely land on the topic of ChatGPT, as it’s been dominating headlines lately. And with good reason, ChatGPT is capable of some really impressive stuff, and it warrants discussions about its ethical uses, what it means now that Microsoft has a stake in it, and how it’s going to ruin human jobs and replace everyone with robots.
But beyond the world of text-based chatbot AI, there’s some arguably more impressive (and controversial) stuff happening in the world of AI as it pertains to image creation. And in that space, Midjourney is one of the stars.
What is Midjourney?
According to Midjourney’s website, “Midjourney is an independent research lab exploring new mediums of thought and expanding the imaginative powers of the human species.” So sci-fi. So futuristic. But what is it, really?
In a much more human description, Midjourney is a generative AI tool that turns text prompts into images. It’s like if a Google image search was able to create exactly what you searched for instead of scouring the internet for images that already existed. For example, have you ever wondered what a polar bear playing volleyball on Mars would look like? Well, now you don’t have to. It’d look something like this:
Using Midjourney is a bit different from some other AI tools, like chatGPT, for example, in that the image creation all happens within a Discord server. Within Discord, you can type the command /imagine followed by the prompt that you want Midjourney to create. For example, the input to create the image above was:
/imagine prompt polar bear playing volleyball on Mars.
Once the prompt is entered, the job is kicked off, and you can see a grid of four images begin rendering. Once the job is finished, you’re presented with four different image representations of the text prompt that you typed. Like magic.
After you have your first four images, you then have the option of regenerating four new images based on the same prompt, generating four new variants of an individual image, or upscaling one of the four generated images to a higher resolution copy.
The whole workflow takes some getting used to and can be a bit chaotic (at least in the beta version. I can’t speak to the paid plan experience) because while your creation is being created, a bunch of other users’ jobs is running at the same time in the chat. It’s definitely not as clean of an experience as some other tools, but it’s hard to argue with the results. Here are a couple of other examples that I prompted Midjourney to generate:
As you can see, what Midjourney is capable of producing in about a minute or so is pretty impressive. I’d argue, though, that none of these would pass for a commercial illustration or design due to some of the weird things that AI does (look closely at the faces and limbs in the instructional technologists’ image). They do, however, all serve as great reference points.
While Midjourney offers several paid plans with varying limits on how many image generations you can use, users can join the beta to give Midjourney a test drive for free.
Thoughts, Musings, and Ponderings
As an artist and a tech-nerd, I have mixed feelings about Midjourney, but they’re mostly positive. From a technology perspective, it’s an amazing feat that creates really cool stuff. Even from the artistic perspective, it’s a really cool tool for rapid prototyping and generating visual ideas from textual thought. I love it. And like other AI tools that have come into the spotlight recently, it’s causing me to rethink how I search for things and how I can use more specific language to get a better result. It’s also causing me to think about how I can use this technology to make my work more efficient.
On the more doom and gloom side of things, there are potential problems with technology like Midjourney. A lot of those problems simply come from the fact that this is a new technology that the rules haven’t been written for yet, but there are problems nonetheless. One of the biggest issues that have come up is that artists’ work has been scraped from the internet to train these generative AI tools, like Midjourney, oftentimes without the artists’ consent. Several artists have taken this issue to court in the form of a lawsuit. And this concern is completely valid. Art is deeply personal and takes tons of time and talent to produce. So, it makes sense that creatives are worried about generative AI tools like Midjourney. However, there was another case where images created by Midjourney that were used in a graphic novel were denied copyright due to the fact that they were “…not the product of human authorship.” So, it’s safe to say that these types of tools are in a bit of a murky area at present.
However, at the end of the day, I find Midjourney exciting. The images that it’s capable of creating are fun to look at, and I’ve personally found it to be a great tool for gathering inspiration and even generating references for my own design and illustrative work. Do I think the tool comes without its flaws? Absolutely not. The ethics and legality of it all are things that are rightfully being questioned and debated. We’re at the forefront of a new era of technology. And with that comes a lot of new decisions, rules, relegations, and controversy. But I can’t help but be excited to see what’s to come.
What are your thoughts on Midjourney?