DALL-E 2, the artificial intelligence that draws images based on what it learns on the Internet, arrives on the Bing search engine. Microsoft is also launch of a new graphic design application called which uses AI technology. His name is Designer and you can register to be one of the first to try it. For those who remember clip art from the 80s and 90s, this is a huge improvement for creating everything from posters to websites. Instead of a few selections, you can quickly generate any image with a simple text prompt.
Microsoft was an early investor in OpenAI, the developer of DALL-E. This program recently deleted their waiting list and open registration to anyone wishing to register. It continues to learn, so the more we all use it, the better. Below, you can read more about how this AI program works and some of the features that DALL-E and its updated versions offer its users.
A free version, formerly called DALL-E Mini, now known as Craiyon, is also available for public use. It’s also an app for Android devices, making it even easier to create weird art based on a chain of random Mad Lib-like ideas. Enter any prompt and the AI spits out nine graphs. Who doesn’t want to see a giant squid assembling IKEA furniture?
How does AI like DALL-E and Craiyon work?
examples of Craigyon website include “a baby daikon radish in a tutu walking a dog” and “a baby hedgehog in a cape staring at its reflection in a mirror”. When it was first released a few months ago, the internet wasted no time putting it through its paces.
Columbo in Marvel vs. Capcom 2 pic.twitter.com/NIRcdCWjjU
– Ismael Bergara (@Calavera145) June 6, 2022
The AI’s job is to do something many of us do every day: look at images on the internet. It also reads the captions and learns what’s probably in the picture based on that. Now it can draw text-based images. The more specific the prompt, the funnier the results. The one below is “Star Wars movie poster in the style of Salvador Dali”. And they are perfect.
This one was particularly beautiful pic.twitter.com/POUL7Gejry
— Paul Adamson (@Paulywoggsvoice) June 8, 2022
Between failed attempts because the site was experiencing so much traffic, I entered “Snoopy playing with other dogs” as a prompt. The results weren’t as cute as I expected. Although it’s clear who Snoopy is, he’s closer to a Picasso painting than anything.
nerdistDan Casey’s own performed much better with “baby yoda eating pizza”. To prove that Grogu is literally cute in all circumstances.
Artificial intelligence like art with overpaint
artificial intelligence and art are great partners. He can turn terrible drawings in stunning landscapes. OpenAI’s DALL-E now has a feature called Outpainting. It allows you to expand the scope of the artwork, whether it is an original work or the mona-lisa. In the example below, the AI zooms out on the masterpiece a girl with an earring. We see the rest of the girl and also the (very cluttered) room she is in. More than just an extension of the square format of the original DALL-E, it opens up the art form to anyone wondering what’s right offscreen.
Artists creating movie posters using DALL-E provide fascinating insights into how AI works. Titles, slogans or even dialogues of the script can serve as a prompt. Craiyon developer Boris Dayma documents the progress of this program on his Twitter account. Whether you use Craiyon or DALL-E, using AI in a workout helps anyone who uses AI. It’s for the sake of data after all, but it’s also a hoot. So says the “surprised Pikachu” below.
The DALL-E 2
The more advanced iteration of the DALL-E program named DALL-E 2 is now available. DALL-E 2 aims to create even more realistic offers in response to prompts and also seeks to reduce bias and improve the safety of its creations. There is no longer a waiting list and everyone can access the DALL-E 2 beta.
According to the platform, “Each DALL E user will receive 50 free credits in their first month of use and 15 free credits every month thereafter. Users can purchase an additional 115 credits for $15. Each credit can be used for an original DALL·E prompt generation—returning four images—or a modification or variation prompt, which returns three images.
Craiyon remains free for everyone to enjoy. Then I try “manatees with Elizabethan collars swimming with rabbits”.
Originally published June 9.
Melissa is Nerdist’s science and technology editor. She also leads “science of” panels at conventions and co-hosts Star Warsologies, a podcast about science and Star Wars. Follow her on Twitter @melissatruth.