My Honest Review of AI Art Tools I Used In 2023
Here’s what I think of every AI art tool I used in 2023.
AI art was everywhere in 2023 from blog post images to art contests.
Today, I want to give my honest review of every AI art tool I used and why I love/hate some of them. I’ll highlight the best features they have and how the impact they had on me as an AI artist.
So let’s start with the app that introduced me to the world of AI art.
Lensa: It wasn’t worth the price
Remember when Lensa first appeared? Picture this: an app rolls out, and suddenly, selfies of everyone become AI avatars. It was the app that had people talking and for good reason.
Most people who tried the app were amazed. Here are some of the avatars that I got with Lensa.
Unfortunately, Lensa wasn’t perfect. In fact, among the dozens of avatars it generated, only a few of them were decent, which was disappointing considering that it wasn’t free!
I feel a bit shy to show those ugly AI avatars that Lensa generated with my selfies, but I can tell you that these days you can find many other apps with way better quality.
In fact, in this list, we’ll see apps that generate so good images that will make you question whether they’re AI-generated or some kind of sorcery that’s making them look so darn real.
Midjourney v4: The first AI art tool I loved
While Lensa had its moment and offered users the chance to turn their selfies into stylized AI art effortlessly, Midjourney v4 meant a world of new possibilities. You could create anything you wanted with a prompt!
Sure, the process is a bit more complex compared to Lensa. With Midjourney, you don’t just pick photos and wait for the tool to do its magic; but you craft a prompt, fine-tune the image generation process, and, in case you want it to replace Lensa, install an external app to swap faces. But, believe me, all of this is worth it when you see the image Midjourney generates.
The images created by Midjourney looked way more realistic than Lensa. It was like comparing a snapshot from a phone camera to a professional portrait. Yes, it takes a bit more effort to create images with Midjourney, but the results speak for themselves — just look at this example (it’s me on the right after swapping faces)
Prompt: An elegant woman, realistic photo
And Midjourney could not only create avatars but anything you came up with.
Prompt: Forest, afternoon time
For me, Lensa was the start, and it was great. But what Midjourney could do just blew up my mind. It could create so many things that I felt like a child playing with a new toy.
Leonardo AI & Photoshop: These and other less-known AI art tools pushed the field in 2023
Before Midjourney v5 was unveiled (and even after), we’ve seen other AI art tools that offer features that Midjourney didn’t have.
One of them was Leonardo AI. This tool was something like Midjourney v4 at that time. A huge deal, really, because it was doing pictures for free. You had some credits to use, and that’s it, no need to pay money every month like with Midjourney.
Prompt: Forest, afternoon time
The quality of the pictures? Just as good as what you got from Midjourney v4, truly. The colors, and the details, were all top-notch (for that time).
Well-known companies also added AI features to their tools. One of them was Adobe. You know Photoshop, right? But with AI was an absolutely new game. It’s become like a magic wand for photographers and designers. With the new Firefly feature, you could create a new image or edit something in one of your existing images
What was really cool was how real the edited pictures looked. You couldn’t even tell what was added. Take a look at this picture below.
See that nature scene with the deer? I started with just a white blank page, and all of this came to life with just a few clicks in Photoshop. Impressive, right?
I have to say, that not so many AI tools come close to what Photoshop can do. It comes in handy when you need to add anything with really good details on the picture you already have or to delete the background. Even Leonardo AI, which is pretty good, doesn’t quite match up when it comes to creating lifelike shadows and details. In this area, Photoshop is definitely leading.
Midjourney v5: Good upgrade with many new features
Every time Midjourney releases an update, it’s like Christmas for AI art enthusiasts. With v5, they outdone themselves with some super amazing features.
Besides the improvement in details (and fingers!), they released new features that I loved. To explain why I love them, first let’s look at the image I generated below (again, it’s me on the right picture after using the swap face external app)
Prompt: A woman with long dark hair, hugs koala
It seems I have a photo with a koala, even though I’ve never hugged a koala in my life (though I wish I could).
Anyway, one of the new features I loved was the Zoom Out feature. You could zoom 2x, or 1.5x, make your picture square, or even custom zoom to your heart’s content.
Another feature I loved was “High Variation Mode”. It lets you crank up the variety with the variation button. You could tweak how strong the variations were and add more details … it’s like having a whole new set of brushes for your digital canvas.
Prompt: a husky puppy and a paper airplane is next to him, forest, sun — s 800
DALL-E 3: It helped me introduce AI art to non-tech people
I loved the previous AI tools in this list, but it wasn’t easy to explain to my non-tech friends how to use them and especially help them get the image they desired.
DALL-E 3 was almost like the grand prize winner in this regard.
Creating images with DALL-E 3 was a different ball game compared to other AI tools. If you’ve used Midjourney, you know you’ve got to get a bit technical with prompt engineering to get those stunning images. But DALL-E 3? It’s way more intuitive. You don’t need to be a prompt wizard. You only had to write what you wanted in plain English, and DALL-E 3 would come up with the prompt for you.
Besides, unlike Midjourney or LeonardoAI, with DALL-E 3 you didn’t need an extra subscription in case you were already a ChatGPT Plus subscriber.
Creating nice images with DALL-E 3 was as simple as this:
Prompt: superhero dog
As you can see, I typed a few words and DALL-E 3 created a robust prompt from my idea.
And something even cooler was that you could even start with a quick sketch. Here’s what I got from a doodle I created (for more details on how to do this, read this)
However, DALL-E 3 isn’t quite there yet in making pictures as realistic as what you get with Midjourney. It’s impressive, no doubt, but if ultra-realism is your goal, Midjourney still holds the crown (especially with its latest v6)
Midjourney v6 is king
And now the undisputed champ: Midjourney v6. This version is hands-down the best thing we’ve got in AI art right now. The ultra-realistic images it produces … you literally can’t tell them apart from actual photos. It’s that good.
Prompt: Sandwich with greens, cheese and red fish
The new version comes packed with features that are nothing short of groundbreaking. It can draw text, and paint hands and fingers with an unbelievable level of detail. They bring out textures in a way that makes every image pop with life.
I’ve already talked too much about Midjourney in this article, so for more details about its latest version, read this article where I show all the new features of Midjourney v6.
So, there you have it. This was my review of AI art tools I used in 2023. If I had to choose one tool, I’d choose Midjourney v6. It’s definitely setting a new standard for what AI can achieve in art. The realism, the detail, the sheer quality of the images — it’s just beautiful.
Thanks for this roundup! Do you think that the different image generation tools will continue to diversity and end up in different "niches" (for example Midjourney for realistic, DALL-E for text, etc) or do you think they'll eventually wind up with pretty equal capabilities?
Another great article!