I’m rather impressed, to be honest.
I’ve been reading about them for a while now and really wanted to try generating images — is it art? — with an A.I.-powered software.
Couldn’t find the time to do it though… until today.
I decided to google my way toward a website that provides the feature and finally test the thing.
I ended up using NightCafé Studio — not sponsored, no referral link — which provides the users with 5 credits to use their A.I. generator for free, daily.
It’s a good way to test and see if their product is what works for your goals.
The 5 credits can go toward generating 5 low-resolution images or fewer higher-resolution ones. When you run out of free credits, you can either win or purchase additional ones. All in all, I found their pricing fair so if I decide to use this website from now on, I think it would be totally worth it.
Now, this software gives you plenty of options to generate art, but I went with text-to-image, which means it will create an image based on the words you provide as input.
Again, I was considering my general interest in this kind of product and its use, so this method fit the purpose better.
I generated 5 low-resolution images, posted below, and only made slight changes regarding the input and artistic style used to create the art.
Here’s what I got.
Text Input Version 1: Mysterious Island With A Single Tree.
Style: Oil Painting.
I think it did a pretty decent job, even though that type of tree makes you think of a rocky type of island — yes, the tree and the huge stone behind it, lol — and not a sunny, sand-filled one, which I thought of initially.
But in defense of the A.I., I didn’t say which type of island I’m expecting to see in the final image. So, just like in everyday interactions, if you want something, say it clearly to the other one. Point, the A.I.
And I find the painting mysterious enough.
“I like it. Picasso.”
Text Input Version 2: Funny Rabbit With a Bowtie.
Artsy, elegant, somewhat creepy.
Again, I’m happy with the result.
Nowhere near what I expected, but it works.
Text Input Version 3: Romanian Landscape Mountains.
Now, I wanted to see how specific it can get. Can A.I. trigger specific emotional responses?
Here I went with words that have emotional value for me — I am Romanian, and a landscape from my country should evoke a specific kind of feeling, it should feel like “home”.
I was blown away. This landscape totally screams “Romania”. It’s the kind of natural environment I expect when going to visit my grandma. It’s amazing!
I am not sure if the thing in the middle is indeed a tiny house — or if it’s just another hill and I’m making up the rest in my mind because it would totally fit the context.
Wherever this thing is pulling information from, it’s the right knowledge pool.
11/10 for this one.
Text Input Version 4: Landscape Mountains.
I wanted to see what it would generally generate as a mountainside landscape, to compare to the previous Romanian landscape.
The thing got a little bit glitchy, as you can see. Maybe the input was too vague, I don’t know what caused it.
However, you can clearly see that it was creating an environment very different from the one generated earlier.
Wouldn’t use this image anywhere, but I am still happy with how the software treats input.
Text Input Version 5: Romanian Landscape Mountains.
The last thing I wanted to test is the results’ originality.
So I placed the same input as in the third test.
As you can see, it’s glitchy again, but the image itself is very different from the previous Romanian landscape. It would pass as a new, original “painting”.
Now I am also rather certain that the red thing I thought was a house in the first Romanian landscape, was indeed just another hill.
But the thing is, if you’d decide to use this just as guidelines, a rough sketch for a piece of art you would then produce yourself, it would still be great — I could paint a house where the A.I. placed the hill.
To conclude, I am sold on the A.I. generating art idea.
I can already think of several ways I could use this type of software, so it is highly likely that NightCafé Studio and/or maybe other similar software will see me and my credit card back.
Another cool thing is that unless you use a visual input for which you are not the copyright owner, the website transfers the copyright to you for the generated image, which means you can use the art for commercial purposes as well.
Pretty cool, I say.
What are your thoughts on A.I.-generated art? Have you used such software? Let me know in the comment section below.
Thank you for reading.