We created these weird and wonderful images using OpenAI’s Dalle-2.
This page was translated using AI and machine learning before being reviewed by a human editor in your native language.
(Pocket-lint) – Several artificial intelligence technologies have emerged lately. They are able to create interesting and compelling images thanks to artificial intelligence.
Google has one in the form of Imagen, but the best known is OpenAI’s Dalle-2. It is a system capable of creating realistic images and works of art from descriptive terms entered by the user.
Dalle-2 is now open to the public, which means anyone can use it and for free (up to a certain number of uses). So we tested the AI to see what images we could create. Keep scrolling to see what we’ve created. We’ve included the terms we used to create the images, so you can see how close the AI came to the request.
Slab-2 is interesting because you can not only tell it what to create but also what style of art you are looking for. So this is what we did:
Apple wasn’t the first to play music streaming, but its platform is very impressive.
“Child’s crayon style drawing of a monster on a wall like graffiti.”
You can also let your imagination run wild. If you are hungry, turn the food into items:
“A house made entirely of donuts with pink icing and sprinkles.”
New Google Pixel Watch
We thought it would be fun to try and get a feel for future smart devices and what an AI might think they look like. Unfortunately, she did a better job on the woman’s hair than on the watch.
“The brightly colored new Google Pixel watch is worn by an elegant woman with curly hair.”
Batman’s New Car
You can make your requests as specific as you want, down to the era of something, the style or just the subject.
“A new 1990s-style Batmobile admired by the Joker.”
What’s one of the least scary things you can see? A balloon monster? We thought it might be a fun alternative to one of the other giant creatures that have historically attacked Japan.
“A giant green balloon-shaped animal attacks Tokoyo.”
We love the concept cars that different brands offer. Designers and engineers imagine all kinds of future designs for cars. But what can AI think the future looks like?
“A futuristic concept of a self-driving BMW on a quiet street.”
What if we sent monkeys to Mars instead of humans? What would that look like? Well, according to Dalle-2, that would make some pretty grumpy monkeys. It’s understandable.
“High resolution photo of a monkey astronaut en route to Mars.”
We recently talked about Nikon’s photomicrography contest, which features awesome photos of the smallest things in our world. We wanted to see what Dalle-2 could create.
“Award winning photos of microorganisms.”
An exploding phone
No one wants this to happen to their smartphone, yet we’ve seen enough horror stories over the years.
“A cell phone that overheated and exploded on a wooden table.”
Sometimes you have to take the time to smell the roses, even if those roses are not real and were created by the AI.
“A macro shot of a rose bush in full bloom.”
dogs in a submarine
Dalle-2 seems pretty good at creating cartoon images. In this case, an amusing vision of a couple of dogs having fun under the sea.
“A cartoon dog in a yellow submarine at the bottom of the sea.”
Concept cars can often be a little quirky. And that’s the fun part, because designers often have a bit more freedom in their designs. So why not let the AI have fun too?
“The most bizarre concept car ever designed by man.”
Duck versus turtle
After seeing what Google Imagen could do, we thought it would be interesting to replicate its results using the same terms. The results aren’t as impressive, but still a lot of fun.
“A chrome duck with a gold beak arguing with an angry turtle in a forest”.
Robots in an art gallery
Another example from Google saw funky robots in a flooded gallery. What would Dalle-2 do with such a request?
“An art gallery displaying paintings by Monet. The art gallery is flooded. Robots are going around the art gallery using paddle boards.”
Corgi in Times Square
Seeing dogs wearing hats and acting like human beings is always fun in real life. We love that the AI chose to embody that joy in one image while making the dog super grumpy in the other.
“A photo of a Corgi dog riding a bicycle in Times Square. He’s wearing sunglasses and a beach hat.”
We were honestly a little disappointed with this one as we were hoping for something a little more creative when we asked to see the cheese moon. But maybe you need to be a bit more specific in your requests to get the best results.
“The cheese moon.”
A space raccoon
With this one, we actually got two quite different, albeit slightly subtle, results. A raccoon looks inside from the outside and another does the opposite.
“Photo of a raccoon wearing an astronaut helmet, looking out the window at night.”
An exploding vase
We are big fans of high speed photography and slow motion results. So we wondered what Dalle-2 would produce if we asked it to create something similar.
“A glass vase filled with flowers explodes in slow motion in front of a colored wall and the glass shatters everywhere”.
A cyberpunk-style car runs away from mounted police in Times Square.
For some reason, when we asked for this scene, Dalle-2 gave the police space helmets. Do what you want with it.
“Cyberpunk-style car flees mounted police in Times Square.”
One of the strengths of this AI technology is how it can create new works of art inspired by the style of a human artist. Even if this artist has never painted such a picture before.
“A painting of a 90s computer in the style of Claude Monet.”
The father is angry
Granted, Darth Vader looks a little weird in some of these images, but the overall theme is still what we were hoping for.
“Darth Vader angrily berates a crying stormtrooper at his desk.”
There’s something fun about teddy bears going on adventures. These two furry creatures are having crazy fun on skateboards in public spaces.
“A chic teddy bear on a skateboard near the Arc de Triomphe.”
The cat monster
We had fun experimenting with what Dalle-2 can create when it comes to cartoons. Here, the cats become sea monsters and frankly, it’s quite unsettling.
“A seamonster cat destroying a fishing boat during a storm.”
Cybernetic humans in pop art style
Even if the final results are sometimes a bit random, we can only admire the versatility of the AI. How many humans could create so many different styles of artwork?
“Cybernetic human painted in pop art style.”
Aliens in the White House
It seems that artificial intelligence has no idea what aliens might look like. We find it interesting that they can apparently look like ghosts traveling by space jellyfish.
“Alien Visitors to the White House in Mars Attacks Art Style!”
Written by Adrian Willings.