AI art generation is crazy mind-blowing, to say the least. You’re now able to create the most amazing images that have never existed in this world before you clicked “Generate”, and this 100% unique creative work now belongs solely to you, its creator. And you didn’t even put any work into it whatsoever.
Or did you? AI image creation usually consists of a text to image conversion, meaning there is some work needed in order to generate an image. That work comes in the form of a “prompt”, a description to prompt the AI to create something following the description you’ve given it. In other words, feeding just the right ingredients in order for the AI to concoct something beautiful and delicious for you.
But, as with all cooking, if you’re no Gordon Ramsey, then not every attempt is going to turn out successful. The same can be said about AI image creation, alas. What you feed the machine will affect what it produces. Unfortunately, since AI imaging is still fairly new, bad inputs = bad outputs, and good inputs also sometimes = mediocre outputs.
How then, can we come up with the best and most effective AI art generation prompts so that we’re not wasting our credits (especially if you have a limited plan) on weird alien art? In this article, I share a few key steps in my own personal workflow when it comes to generating AI art.
Photo VS Art
Understanding what you want to achieve is probably going to be the first and most important thing you need to do when it comes to generating AI images.
First of all, do you want to create a realistic-looking photo, such as a stock photo for a blog article, or do you want to create some fantastic looking art that you can use on Instagram, for example?
Most AI art generators offer both options, but take note that the prompts that you’ll use for each will be completely different.
Some words that you might include for photos are: realistic, realism, hyperrealistic, photo realism, photography, ultra detailed, movie still, film scene, stock photo
Whereas for art, you might use completely different words such as: digital art, concept art, detailed illustration, high resolution illustration
Landscape VS People
Generating images of landscapes has proven to be fairly easier than generating people because, oh I don’t know, for some reason AI is still unfamiliar with the human body and thinks we have four eyes and eight fingers, or anywhere in between; a fixed number of body parts is clearly just an abstract concept for artificial intelligence. I mean, can you blame them (it??)? They are artificial intelligence after all.
Some things you can try to insert into your prompt to create less alien-looking people are things like: proportionate facial features, symmetrical eyes, symmetrical face, symmetrical lips
When it comes to hands, though, I’ve tried everything from “normal looking hands” to “five fingers”, even “no hands” and yet I still can’t seem to generate any non-alien hands. So if anyone knows the secretsauce, please do share!
Some other words you can use where people are involved: Character portrait, photo shoot, stock photo of a person, full face, portrait photography, cinematic closeup, character closeup, closeup photography, perfect composition, bokeh
Meanwhile, if it’s landscapes and backgrounds you’re after, try these: Stunning scene, beautiful lighting, extremely detailed, intricate environment, establishing shot, cinematic lighting, dramatic lighting, god rays, global illumination, award-winning photography, cinematic shot, high-quality background, HD background, rule of thirds, perfect composition, bokeh, scenery, landscape
It doesn’t matter if you’re creating AI photos or AI art, name-dropping famous creators and artists well known for a particular style can really help AI understand what you’re trying to create.
It really depends on what you want your art to look like – like something from the art masters, Piccaso, Vincent Van Gogh, Michelangelo, or something more in terms of modern pop art ala Andy Warhol, Keith Haring, Roy Lichtenstein.
For photography, you can try using some famous photographers if you know of any, or try things like award-winning photography, national geographic photography, or even something like “trending on Unsplash”.
Photographer Keith Dotson actually tried it out himself, and generated AI images in his own style of photography! His conclusion is that he, thus far at least, can still keep his day job hahaha.
Just a note about this, though, using these names helps your art generation, but it doesn’t help the artists themselves. You’re literally creating art in their style, but they get absolutely no credit or money for it. It’s controversial, and artists like Greg Rutkowski are honestly quite fed up with it.
Personally, I haven’t decided what to feel about this yet. I’m still processing the ethics behind what is possible. But I just thought I’d throw it out there so you can give it a think for yourself and know what moral dilemmas lie therein.
Make Sure to Clarify The Style
Style here is more for creating art images, and you can refer to broad, categorical styles such as surrealism, contemporary, minimalism, art nouveau, art deco, renaissance, impressionism and such. Check out a whole bunch of art styles and their examples in this article.
But you can also get even more specific by using “in the style of”, such as in the style of Final Fantasy, layering that on top of other keywords like anime, Japanese manga, or even in the style of Studio Ghlibi.
Play Around with Adjectives
This is probably my favourite part, because you’re literally just using adjectives to describe what you want to see eventually.
It’s like an artist was sitting in front of you, and you had to verbally tell them in words what kind of art you want them to create – what words would you use exactly?
Here’s where a thesaurus can sometimes be exceptionally useful. If you want to create something fantastical, for example, Thesaurus.com also gives us other relevant words we can add to our prompt, such as wondrous, spectacular, incredible, breathtaking, out of this world, extraordinary, astonishing, remarkable, and so on.
Using such words will solidify the type of creation you want to see from your AI art generator.
Phrasing Your Prompt
There are two main ways of phrasing a prompt, although I suppose it could also be a mix of both.
One is simply putting in a ton of keywords that are relevant to what you want to achieve, simply putting a comma in between them.
So for example, a prompt such as: detailed illustration, digital art, concept art, final fantasy, kawaii art, japanese manga art, cute anime eyes, symmetrical face, witch, cat, cats, spirit animal, pastel, beautiful, magnificant, ethereal, elusive, soft, floating, fantasy, trending on behance, deviantart, melancholic
Would create something like this:
Meanwhile, you can also simply put in a sentence that describes what you want to see, and perhaps the name of a famous artist, and see where that takes you.
In this scenario, a prompt such as: a cinematic shot of a beautiful scene with gorgeous landscape and stunning lighting, of a magical forest overgrown with trees, by Tyler Edlin
Will generate something like this:
I’d recommend trying both ways because you never know which one will work better for you.
Creating Effective Prompts
So yes, while AI image generators are doing most of the heavy lifting for you, I’m pretty convinced that it still requires quite some human work in terms of research and prompt development.
AI image generation is still very much a process of trial-and-error and experimentation, where you generate something based on your prompt, and then tweak your prompt constantly until you achieve something closest to what you want, and then repeat that prompt a few more times, to see what else you could get.
Just as a summary of my workflow:
- Add keywords related to photo generation or art generation
- Add keywords related to landscape creation or people as subjects
- Add artist names
- Add styles
- Add adjectives
Oh and just as a last point I guess – which is not needed in GoCharlie because they have a 4K quality option that you simply click on before generation – but in some other generators you might want to put “4k” as a keyword if you want high quality, high resolution photos and artwork.
shortcut – use your ai writer
Here’s a bit of a shortcut or hack – if your AI writer has a function such as image to alt text, then all you have to do is find the kind of image that you want to create via Google Images, and drop it into your app to get the text, then use that text as part of your prompt to sort of reverse engineer it. It won’t always work, but it’s worth a shot. It gets your brain juices started at least, so you have at least something to build up from.
Choose Your AI Image Generator
I definitely believe that unlimited credits are needed when it comes to AI art and photo generation. I personally have Nichesss, which also comes with an AI image generating feature, but people on the LTD plan from last year or whenever are limited to 25 credits or 25 runs, which is absolutely nothing when trying to generate usable images. I’ve personally seen so many people complain about running out of credits just as they were getting better at writing prompts.
On the one hand, we should be grateful that we’re given any credits at all since it is a new feature and technically not part of the LTD we paid for, but on the other hand, if we’re being offered use of the feature, it would be good to actually be able to really USE it, ya know?
Anyways, if it’s unlimited AI art generation you’re after, then I would suggest either of these AI image generators:
Jasper Art, with its $20/month offer right now, gives you unlimited generation. BUT I do hear whispers in the groups that this is just an introductory offer for people to get a feel of Jasper Art, and it seems like this unlimited-for-20-bucks deal isn’t going to last forever, but I guess we’ll see. If you’re looking to create art for a particular project and you just want to sit there and generate AI art all day long for 30 days, then Jasper Art is probably the way to go.
However, if you see yourself generating more AI art over a longer period, or for the foreseeable future, then GoCharlie is still offering their LTD with unlimited generation FOR LIFE at $435 only until 28 November, so seriously, get it before it’s gone!
The GoCharlie limited LTD plan at $99 offers you 175 credits for art generation (and 35 1-Click blogs), again FOR LIFE, so if you’re not able to afford their unlimited plan, at least just get their limited plan, because seriously, once it’s gone, it’s gone, you guys. You also get unlimited short copy generation with their limited plan, as well as limited credits for future updates, so hey, it’s really a good investment to make on a solid company that knows what they’re doing and who, above all, listens to their users.
If you just want to play around with a Free AI generator, GoCharlie also has a free plan, but you’re limited to a number of generations, obviously. I’m not sure how many but probably 30 or so as well. But just to be able to play around, sure, you can.
(Note, these GoCharlie links are my affiliate links. I am not an affiliate marketer, so this is not like my day job kinda thing, I just like and recommend the app, so please do support my little family if you like the content I’ve been putting out about GoCharlie. I definitely intend to share more tips and tricks with regards to content creation using the GoCharlie app for sure.)
Midjourney is also free via the Discord chat app, so you simply click on “Join the Beta”, visit their Midjourney bot, key in /imagine whatever you want, and their bot creates four images for you based on your prompt. You have to upgrade to a paid subscription after your free trial, but I don’t yet have any details on that at the current moment – I’ll update once I know more.
Again, I would not recommend Nichessss because 25 credits are really nothing, so yeah, sorry Nichessss.
Coming up with effective prompts is pretty much all up to your imagination, so go wild and go crazy. One thing I’ve noticed is that it doesn’t mean that the more words you use, the better. You can toss a whole tsunami of words in there, but it doesn’t mean that the AI is going to consider all of them. So it may be better to choose your words selectively, so it gives the AI more focus and direction.
For someone like me who has always loved being creative and creating beautiful things, but who can’t draw for nuts, AI art generation is simply a dream come true. Controversy aside, I am super in love with generating art and can definitely see myself using this in a ton of different ways, be it content marketing, social media, PLR content, or print on demand.
So, go, try it out, and become a believer yourself.