When we first heard that A.I. tools are now able to create artwork using nothing but text prompts, our ears perked up.
Not that we’re eager to have a computer replace an artist, but our curiosity into the implications of this technology was enough to want to explore.
There are several of these tools available, but the one artists and designers seem most interested in is called MidJourney, which has been making the rounds on the internet, though it’s not fully available to the public.
We got access to the beta of the product, and have been exploring how it operates, and some of the strange results.
The steps are simple: Open the tool, create a text query, and hit enter.
MidJourney takes about a minute to give you four options to choose from, and you can then “upscale” the image to a higher resolution, or try variations on a theme.
Below are some of our results, with captions showing what the query was, and what images the computer returned to us.
Our query for a “futuristic electric car” got the following result, which is sleek yet also just weird.
In this one, we asked MidJourney to create “fuzzy moss that spells out M O S S A N D F O G”. It managed the letter M, but we’re not sure it’s the right kind of fuzzy 😂
This query was for “calm forest in the fog with bright green moss”, resulting in a pretty realistic scene.
Some of our other queries were to see if the tool could create a new logo for us. These asked for “logo made from moss”.
The results are….strange. We wouldn’t call them a success, but interesting studies into what artificial intelligence can quickly gather about a subject or object.
We do think it’s only a matter of time before tools like this could become instant logo generators, which would indeed jeopardize the rolls of some graphic designers or artists.
It seems like the tool is not ideal for spelling out words, instead wanting to create strange, organic forms that incorporate some letters but not full words or sentences.
The tool also has a number of safeguards in place. For one, it’s not generally available, only by invite, limiting access. Secondly, the tool does not allow explicit or violent queries. This is for obvious reasons, so that people don’t create hate-filled artwork, or even fake news images.
Our query was “Chihuahua driving a Miata”. What do you make of the result?
This query was for “spaceship flying past Saturn’s moon Titan”. We’re generally pleased with the outcome, though there are some unfinished parts of the art.
All in all, the tool has a long way to go, and artists and designers shouldn’t fear for their careers….yet.
At this point it’s more of a curiosity or gimmick than a purposeful tool. But we imagine in the years to come, things will start to get very interesting, indeed.
[…] famous creations? Would the car embody their trademark style? We’ve enlisted AI to help us explore the ‘what if?‘. We used the program MidJourney to help visualize […]
[…] query in Midjourney was simple: “Homer Simpson in the style of ________”. We know a more nuanced and […]
[…] ingenious ideas that artificial intelligence can help bring to life. Utilizing popular AI art tool Midjourney, we had our favorite 4 year old give us ideas to feed into the […]
[…] did our own experiment this year, using Midjourney to create a series of images and logos for Moss and Fog. The results were….fascinating, if not […]