Get Premium
Dark mode theme is available exclusively for premium users. Learn more about the benefits of subscribing.
No fees, cancel anytime.
Dark Mode 
No fees, cancel anytime 
116Kviews
Midjourney AI Can Now Conjoin 2 Photos, And Here Are Some Of The Test Results Shared By A Telegram User (40 Pics)
Interview With Author
Have you ever wondered what a "cat bread" would look like? Well, it's your lucky day, because even if you have never wondered about that, you are still invited to explore this bizarre reality made possible by none other than an AI.
It has been recently discovered that Midjourney, a popular AI research lab producing a program that creates images from textual descriptions, can now mix different images.
Internet users have been testing a new version of the Midjourney neural network - V4, and the author of the Telegram channel called "AI molodca" showed his subscribers' the unique imagery that the AI has generated from his suggested photos. The user explained that instead of text queries, you are able to insert links to images in jpeg format in the search bar to get exclusive content.
And this is how, fellow Pandas, a “cat bread,” among many others, was born.
More info: midjourney.com | t.me | Instagram
This post may include affiliate links.
Bored Panda got in touch with Aleksandr, the creator of these collages, to learn more about him and his inspirations.
Aleksandr has quite a creative background. When asked to describe himself, he wrote: “For now, I'm a Georgia-based Creative Director at an independent advertising agency DADA, born in Moscow.”
RELATED:
AI being a rather new thing that's still developing, not everyone has developed an interest in it yet. We got curious about when it caught the attention of Aleksandr. He wrote: “Since ancient times (3-4 years ago). When the possibilities of generative AI became at least somewhat available. At that time - rather in a primitive, but still interesting form.”
“This year, there was a technological breakthrough, so I started a Telegram channel where I explored new opportunities.”
Aleksandr got some interesting results using Midjourney. But we’re sure that mixing photos, e.g. of two cats, wouldn’t be getting the same fun results as Alexandr's chosen pics did.
We were wondering if Aleksandr used some sort of technique that helped him to achieve these results. He wrote: “So, I really love pop and internet culture. But from an artistic point of view, I'm interested in how diametrically opposed concepts and completely different worlds will mix. In general, the combination of something old and getting something new is the main formula of creativity.”
Aleksandr named his favorite photos so far, even if that is not so easy to choose from: “Gigashrek or catbread? I can't choose!”
He also shared with us what people think about his generated images: “AI continues to surprise (and scare a little) people.” But besides that, Aleksandr encourages other people to try it out for themselves and shared some tips for newbies. “You can try it! It's very simple. Go to Midjourney, log in to Discord, go to newbie channels, type /imagine, paste two URL.jpeg that you want to blend, type --v 5 = enjoy.”
And lastly, he added: “Beware, it's very addictive.”
#12
Christian Fuller 2 years ago
Funny that a Doom uniform plus an Animal Crossing secretary outfit equals Link's tunic.
#18
#29
Doofus Jerry 2 years ago
I was going to joke that the output still looks better than the Metaverse, but after looking closer, it really does look better. Mii characters from 10 years ago look better.
#32
Fell Ripley(she, they, potato) 2 years ago (edited)
"A buh. . .a bih.. .whu?" Is what I said out loud after seeing this.
#34
#36
Monday 2 years ago
With those two raising her I can see her turning out like this no problem!
Load More Replies...
Porpoisepower 2 years ago
SpyxFamily and original TMNT... Both are classics
Load More Replies...
#37
#39
Doofus Jerry 2 years ago
Peppa Pig All Grown Up. Shame that was the only image we got to see. They were pretty far along in production too.
Add New Image
What do you think ?
Loki’s Lil Butter Knife 2 years ago
This is the artificial intelligence engine called Mid Journey. You can create three AI creations before you are forced to purchase a subscription.
Load More Replies...
VonBlade 2 years ago
Belief this is legit; 2% It isn't combining two images, nor does it know which bits to take from what. It looks far more like "here's two images, Worth1000 them together". (old website reference I know). Or 38 would have a face mask. Where did 25s skirt come from, or hair ribbons go? *shrugs*
Gor Mor 2 years ago
It isn't combining just 2 pictures. This kind of AI are trained on bilibin of pictures. And the object recognition is the core function, otherwise it would not respond to a text prompt. It also somehow understand relationships between objects in the scene and their property, although the current public AI are messy, and the only doing the job properly is yet to be released Google Imagen. What happens, the AI learns the scene on both pictures, recreates it based on billion pictures it had seen and meshed up already during training, then builds math average between recreated scenes.
Load More Replies...
Josh McJunkin 2 years ago
It's truly amazing what happens when you have an AI use it's equivalent of an imagination.
Loki’s Lil Butter Knife 2 years ago
MidJourney is an machine learning artifice intelligence engine, meaning that it “learns” and improves based on the prompts given it by the machine’s creators and users. I found that it still has a lot of trouble with facial features and often distorts eyes and does not do well when there are multiple characters in a prompt.It is also fascinating to see how AI interprets the prompts you give it. I provided the prompt “ Darth Vader knitting a Christmas sweater” and the AI engine produced an image of an ugly Christmas sweater with Darth Vader on it.
Load More Replies... Loki’s Lil Butter Knife 2 years ago
This is the artificial intelligence engine called Mid Journey. You can create three AI creations before you are forced to purchase a subscription.
Load More Replies... VonBlade 2 years ago
Belief this is legit; 2% It isn't combining two images, nor does it know which bits to take from what. It looks far more like "here's two images, Worth1000 them together". (old website reference I know). Or 38 would have a face mask. Where did 25s skirt come from, or hair ribbons go? *shrugs*
Gor Mor 2 years ago
It isn't combining just 2 pictures. This kind of AI are trained on bilibin of pictures. And the object recognition is the core function, otherwise it would not respond to a text prompt. It also somehow understand relationships between objects in the scene and their property, although the current public AI are messy, and the only doing the job properly is yet to be released Google Imagen. What happens, the AI learns the scene on both pictures, recreates it based on billion pictures it had seen and meshed up already during training, then builds math average between recreated scenes.
Load More Replies... Josh McJunkin 2 years ago
It's truly amazing what happens when you have an AI use it's equivalent of an imagination.
Loki’s Lil Butter Knife 2 years ago
MidJourney is an machine learning artifice intelligence engine, meaning that it “learns” and improves based on the prompts given it by the machine’s creators and users. I found that it still has a lot of trouble with facial features and often distorts eyes and does not do well when there are multiple characters in a prompt.It is also fascinating to see how AI interprets the prompts you give it. I provided the prompt “ Darth Vader knitting a Christmas sweater” and the AI engine produced an image of an ugly Christmas sweater with Darth Vader on it.
Load More Replies...
Related on Bored Panda
