Hey kids! How ya been? Miss me?
Been a little busy with things, and haven’t paid much attention to my own garden, but I am back with some news on AI Art.
Those of you who know me, know that I have had a passion for animation since I was a kid.
After ending an animation program at the University of Guelph when I was 12, led me on to a passion for cartooning, drawing and making silly stuff.
With the ascent of AI Art have come a holy host of AI art programs, some easy to learn and some more difficult (and expensive).
I thought I would detail some of my experiences with these programs, as well as showcase some of the art that I have created.
Controversy
With the launch of any digital art app, there are going to be those who cry foul over the perceived loss of human creativity in the process.
They will decry the theft of art from talented artists, and that how now anyone can steal and pilfer great works of art, like British soldiers in the old world.
As someone who has spent ample time creating and mixing with this tech, I can assure you that they are full of baloney.
First, creating AI art is more alchemy that anything. From learning the differences between rendering models, to knowing which keywords amplify the desired effects, all take time and practice.
I have created some heinous looking creations, while stumbling into some pieces of beauty at the same time.
Having the ability to copy/ paste other people’s prompts helps, but it also doesn’t, after all it’s up to you to put those words in order to be able to solicit the desired look.
As for the artists, I had never heard if Greg Rutowski prior to using AI art, but now that I do, I am glad that he is getting the recognition he deserves.
Computers aren’t ‘stealing’ his work, they are making adaptations to his style, and as such his name will go down in the lexicon of OGs that helped build this into the platform it is going to be.
With processors getting faster and learning more, it will only be a matter of time before AI has created it’s own variations of styles that we can pick and choose from.
Midjourney
This is considered the ‘Cadillac’ of AI Art programs, which is probably why it creates such stunning looks.
Midjourney isn’t that ‘free’ however, and users will need to be able to join Discord as well as purchase credits if they want to play around.
Midjourney uses generative adversarial networks (GANs) and other machine learning algorithms, allowing it to create very high-quality, original art.
It is the home of the ‘cool surfers’ riding the big waves of artistic change.
Many of us ‘amateur’ artists will still look in amazement at some of the cool things that the cool kids are going, hoping that they will drop a few crumbs of text-prompt to keep us hungry.
If you’re looking to get a start in AI Art, don’t come here first unless you know what you’re doing.
NightCafe Studio
Another program that uses GANS is NightCafe Studio, one of the easier platforms to pick up the process.
Nightcafe Studio also encourages users to ‘play around’, by offering credits for coming back everyday, voting, and sharing your work on social channels. While some of these credits are a little ‘buggy’, they allow you to play around and experiment.
I have been using NightCafe for several months now, and you can see the difference between some of the art I started with, and some of my more recent works.
They also just released ‘SDXL’ which ups the game in a massive way in terms of the quality it produces.
This is supposed to be an image of police arresting criminals. Eesh.
In this next one, I tried to make some Spotify album covers, and wanted to put some cool rock and rollers on it. Still quite a bit of work to do.
This Alien DJ was made using SDXL. As you can see the quality and depth of field increases dramatically.
This is a Weed Icon that I also made using SDXL. A big upgrade from some of the original works that I created. Pretty big difference.
Playground
Built by Google, this AI feels like the ‘tits on a moose’ option of them all, as it doesn’t really do much.
Yes, all the prompts are open so that people can see the magic behind the muff, but the codes are so laborious that they won’t work on other platforms.
Much like Bard, Playground seems like one of those Google products created in some far-flung division that suddenly got hip.
Not really sure if this one is going anywhere.
Adobe Firefly
The new kid on the block is Abobe Firefly, a beta version of this product, which is only available to select Abobe users. (which I am one).
Unlike other AI art programs, Firefly doesn’t trove the net looking for keywords, instead relying on Adobe’s vast creative libraries to concoct their creations.
Given that it is still testing, there are a few annoyances such as keywords being flagged that are offensive. ‘Fruit punch’, ‘covered in weeds’ all got flagged, tagged and bagged by Firefly, whereas some dirty monkeys were able to produce some porny looking shots.
Adobe hasn’t got around to licensing Firefly yet, which means that users will get in trouble for trying to use their art.
All in all, it did a good job on the following:
This is an astronaut floating in space, with a planet exploding behind:
A model who looks like a Trans Rock
A creepy dead child crawling out of an old TV
A hairless Yorkie
There will be more programs coming out in 2023, so if you’re looking to spend $30 on some App that messes around with AI…don’t. All the good stuff will be coming out soon.