
Last remains of winter are slowly melting away.
The weather is getting warmer, yet drier and dustier, the downside of spring (later combined with pollen allergy season). But the days are rapidly getting brighter and the need to wear sunglasses is more frequent now. Soon it will be summer after all.
Letting a software application visualize text
During the Easter holiday, I’ve unexpectedly been spending my time exploring an AI image generator called Midjourney and I must say, it is incredible. Sure, I’ve read about these tools, but actually testing one really made me realize how magical it feels to use.
Playing around with it and trying out how far I can push its “imagination” has been quite addictive. Midjourney’s interface is just a text prompt where I type in what I want the AI to visualize for me and within a minute it creates four suggestions of what it thinks I was after. The speed, detail and skill to imitate different various famous artists is scary.
Sure, it is also very much a hit and miss affair, the misses being the AI understanding my query differently than what I meant or just rendering stuff wrong (holding objects is a classic blunder). Frustrating, but still early days with this technology.

My very first request. I asked Midjourney to imagine an “old sea captain, standing on ship, storm in the horizon, painted by Akseli Gallén-Kallela”. Unfortunately it doesn’t seem to know Akseli, but otherwise I was impressed.
While Midjourney renders impressive images based on my requests, there isn’t a sense I created or imagined them. I just tell it what I want and hope for the best. I can’t claim creative ownership and Midjourney’s open community approach of sharing all images and queries to the public strengthens this feeling.
Another feeling I’ve been getting is that the whole service feels like a slot machine. I try something, it gives something back, I slightly modify the query, give it another try, get again something, rinse and repeat and suddenly a few hours have passed playing around with it. Quite similar to a “just one more turn” trap I get into while playing Civilization.
What does this all mean
The obvious question everyone is asking is what sort of impact will this have on real artists and photographers. If I would consider myself a professional artist or photographer, I would view Midjourney as a tool for inspiration to create something similar myself. The joy creating something with your own hands and imagination still exists, but making a living out of it might be more challenging now (especially if the AI just imitates your style of work).
Also, out of over hundreds of query jobs I gave to Midjourney, it was surprising how few of them I thought were worthy of adding into this blog post. Most of them are just variations of a same concept I had in mind and maybe half of them never came to a satisfying result. Compared to what other users have been able to create, the ones below are rather average images to the general audience, but for me able to fiddle around locations like Helsinki and different artist styles has been fascinating.
- Senate square in Helsinki (style Enki Bilal).
- A reimagination of the image banner of this site.
- Cloud from Final Fantasy VII (style Enki Bilal).
- Jesus stepping out from a cave (style Roger Dean).
- Noble woman sitting in computer room (style mix of Masamune Shirow and Ilya Repin)
- G-Shock Robot
So rounding things up, Midjourney is fun, groundbreaking, not perfect and controversial when it comes to imitating other artists. I feel tempted to buy more rendering hours on the service, but since I really won’t be doing anything with all that imagery, I think it might be better to resist (fun).





