AI art generation — it’s a bit like magic, only better because it’s magic you can direct (well, mostly!). Over the past few months, I’ve dabbled with dedicated AI platforms and general large language models (LLMs), like ChatGPT, Grok, and Gemini, for some seriously impressive image creation. And let me tell you, these models can do wonders! I’d love to say I’m an art maestro, but it’s the AI that deserves the real credit. All I do is toss in a few words, cross my fingers, and hope for the best. But that’s the beauty of it — these tools transform our wildest ideas into visual wonderlands. Here’s a peek into my AI art journey.
Meet the Dedicated Platforms: NightCafe & Deep Dream Generator
NightCafe and Deep Dream Generator hold a special place in my AI toolbox. They create stunning visuals, and each has a unique flair. NightCafe is like the reliable friend who mostly sticks to the plan but has some quirky rules (hey, fair enough). It’s flexible enough that, with a bit of creativity, you can still explore some more unconventional ideas, though not everything can go on display in polite company. NightCafe also lets you tweak settings galore, and its “prompt magic” is top-notch — like having a little secret sauce to amp up creativity. Deep Dream Generator, on the other hand, is the wildcard. You never quite know what you’re going to get, but that’s part of the charm. Just watch out for surprise nudity, even if that’s not in your prompt!
Why LLMs Are Fast Becoming My Faves
Now, let’s talk about the big guns: LLMs. I have to admit, the general models like ChatGPT, Grok, and Gemini are edging their way to the top of my favorites list, and for good reason.
1. Iteration, Iteration, Iteration
LLMs let me play art director, giving feedback on generated images and asking for tweaks. This iteration process can lead to unexpected laughs (anyone else on a “comical quest for a celebrity group hug” as detailed here?), but it’s powerful. You can get surprisingly close to your mental vision by refining, readjusting, and even coaching these tools. Sometimes it’s perfect, sometimes it’s hilariously off, but it’s always entertaining.
2. Communication, Finally!
Dedicated platforms like NightCafe deliver beautiful images, but they’re pretty much “all image, all the time.” LLMs, though, will actually “talk” to you. Not only can you generate an image, but if it’s not what you want, you can dive back into the prompt, fine-tune it, and try again. It’s like having a conversation with an extremely talented (and somewhat unpredictable) artist who’s always open to feedback. You can even get some helpful prompt advice to really dial in your vision.
The Bonus: ChatGPT’s Magic Touch
One extra perk of working with LLMs — especially ChatGPT — is its recent ability to accurately include text within images. None of the other platforms have managed this yet, which is both surprising and a bit frustrating. Now, ChatGPT can handle those small but vital details that really bring an image together.
Curious? See the Results for Yourself
If you’re intrigued by what these tools can create, check out the galleries! Gemini and Grok, as well as the full showcase of all models on Lumaiere, are filled with everything from serene landscapes to bizarre experiments that only an AI could conjure up.
A Prompt for Inspiration
Feeling inspired? Here’s a fun prompt to try for yourself:
“Create a surreal landscape where mountains melt into the sky, with vibrant clouds shaped like animals. The scene is dreamlike, with a soft, glowing sunset that fills the sky with pastel colors.”
If you’re into this kind of creative exploration, follow me for more AI art experiments, tips, and a bit of humor along the way. Happy prompting!