In this post I try my hand at creating images with the image creation AI platform, MidJourney. There are hundreds of such tools and apps but MidJourney is widely recognized as foundational and lends its code to many. The process is simple: just type in "/Imagine" and then enter the prompt or the idea that you'd like the AI to create.
I started simple and asked for a "Robot Racehorse". In about 30 seconds I had 4, at which point I could choose which (or all) to have the AI further enhance. The results, for such a basic prompt, were surprisingly good!
Next, I tried a slightly lengthier prompt, asking MidJourney to imagine a log cabin in a snow-covered forest at sunrise with smoke coming from the chimney. What I found most remarkable about this set, aside from how great they look, is that while they are all similar and could have come from the same artist, each is just slightly different. It was an incredible reminder that the results of this simple process were 8 unique, never-before-seen, and never-to-be-created-again images.
Next, I tried having the AI use an uploaded image -- my face. This is a slightly more complicated process in which you point the prompt to an image file URL. I asked the AI to produce an image using my face in a Norman Rockwell-style picture. While the AI seemed to use my face in a very vague way (I'm not sure I am really there anymore!), the results were really incredible:
My takeaway is that image generation is frighteningly easy, and will surely have a massive impact on our business. Already many creative teams and designers have embraced the technology, using it as a quick way to brainstorm and generate initial concepts, as well as a helpful tool in creating concept boards, storyboards and moodboards.
I'd encourage everyone to pick an AI image gen tool and take it for a spin.