HOW TO MAKE AI ART: DALL-E MINI, AI DUNGEON, AND MORE
Not all of us have the talent to whip up a piece of art at a moment’s notice. But algorithms using machine learning are learning how to create “AI art” based on text prompts—and you can use them, too. It’s fantastically fun.
Algorithms like DALL-E (and eventually, DALL-E 2), DALL-E mini, Craiyon, Midjourney, Meta’s Make-A-Scene, and more are learning how to examine publicly available art and learn what makes them art—or at least digest the various elements and style of a photo or artistic work and recombine them into something new. Sure, you can argue whether they’re, in fact, making “art,” but the creations are unique, original, and compelling.
Simply put, AI art uses a text prompt: something specific like McDonalds at the bottom of the sea, for example, or something a bit more generic like the castle of time—the prompt that generated the art at the top of this story. The AI then uses what it has found on the web and what it knows of the query to custom-create an artistic rendering that matches the description.
Because of the computational requirements of training and using the algorithms, many of the most powerful ones are still locked inside beta tests, where only a few lucky participants are able to try them out. One notable exception is DALL-E mini, a public test of the AI that’s available for you to try and is migrating to Craiyon. That’s good news; the DALL-E Mini developers are migrating to Craiyon for trademark reasons, but DALL-E Mini’s popularity swamped the site. But we’ve also found an even better one called Latitude’s Voyage, which you can try out for free.
DALL-E mini, Craiyon, and its competitors
