MidJourney:
Exploring Artificial Intelligence Generated Art
MidJourney, a relatively new software tool available via Discord, allows you to “prompt” an artificial intelligence (AI) Engine to generate imagery. If you have an idea, and can articulate it in words, the AI will bring it to life. The AI takes your words (prompt) and searches the web for imagery and then creates a visual representation (actually 4) of what you provide it within seconds. Once the AI has delivered the first four images you can continue to explore variations of each or upscale individual frames to increase their detail.
The results are short of spectacular. In many cases, the visual representations are gorgeous examples of digital art.
Prompts are made of parameters that help the AI determine how the images will look. MidJourney’s comprehensive manual provides a detailed description of the types of commands you can include in your prompts to alter how your images will look, including detail, size and other image modifiers.
My first MidJourney - Prompt: "A World Void of Hate"
MidJourney is great new tool for visual artists who are keen on articulating what they want to see to share with others. Or for photographers and cinematographers who want to "pre-visualize" some of their ideas before they commit to them.
The tool can also be a powerful medium for talented wordsmiths to colorfully make their words come to life.
In the short time I have been experimenting with the tool I have been mesmerized by it’s ability and the results it generates. Below are some of my first attempts at AI Generated Art.
My second attempts were a little more abstract - Prompt: "The colors of the sounds I hear"
Exploring more complex prompts, this one more in line with this site’s topic: “a single mountain biker riding away along a singletrack trail in the woods, mist, dawn, in the style of Bob Ross --aspect 16:9”
Another one more in line with this site’s topic: “a single cyclist riding along the potomac river in Washington DC with monuments visible in the background, morning, mist, --ar 16:9”
The possibilities are truly endless, even with “vague” prompts I was able to get some cool, albeit eerie, results…
beyond "BestRidesDC" my online persona is "FrescoVA", here are the results of that prompt: "frescova"
Prompt: "courtita" my wife's online alias...
Some of my favorite MidJourney results came from abstract prompts like the ones below
Prompt: "lightning wrapped in plastic in the deep pacific ocean"
Prompt: "lightning wrapped in silk"
Prompt: "lightning wrapped in silk on the edges of a nebula"
MidJourney is currently in beta and only accessible via invite; you must have a Discord account to take advantage of it. There is currently no actual User Interface (UI) and all imagery is generated via text prompts by messaging the MidJourney Bot directly. I will be posting more of my imagery directly on my personal Instagram feed, along with any bike related art on my BestRidesDC Instagram feed.