New Illusion AI Art Trend Sends Internet Spiraling, Can Be Recreated For Free
Images by MrUgleh and featured with permission
Regardless of how you perceive the buzzing world of AI-generated art, one thing’s for sure—the technology allows people to dream up fantasy worlds like never before.
Now, a new trend transports us even deeper into these hypnotic realms. It imagines scenes that are oh-so-satisfyingly enveloped into illusions such as grids and spirals. Needless to say, the internet is in awe. Recreating it from scratch, though, takes a great deal of work. Believe us, we’ve tried.
Thankfully, a new artificial intelligence generator enables anyone to conjure up these dreamscapes in seconds, and without having to subscribe to anything too. The Illusion Diffusion model, now available on Hugging Face, only requires two key inputs: a text prompt and a shape to guide your illusion.
These AI designs are the new viral trend.
— Alvaro Cintas (@dr_cintas) September 17, 2023
And you can now create them for FREE in less than a minute.
Here is a tutorial and 8 incredible examples: pic.twitter.com/pUnPBSRKHV
Stable Diffusion user MrUgleh—a self-proclaimed fan of MC Escher’s art—appears to have pioneered the trend and technique, first experimenting with a QR code creator to attain the look. He then shared his walkthrough on Reddit to help others achieve similar results.
Pretty fun new technique to explore. pic.twitter.com/AezcK619D7
— charlesai (@HODLFrance) September 16, 2023
Inspired by this captivating style, developer AngryPenguin created Illusion Diffusion based on this workflow to make it easier for people to create their own mind-bending scenes. The program is powered by the open-source Stable Diffusion.
As explained by AI educator Alvaro Cintas, to generate one of these images, you’ll first need to upload an illusion template or mask. Preloaded shapes include 2 x 2 and 3 x 3 checkerboards, plus a spiral.
âœï¸ TUTORIAL (You can skip this if you just want to see the examples)
— Alvaro Cintas (@dr_cintas) September 17, 2023
This could be a lengthy process but @angrypenguinPNG built a Hugging Space and now it takes seconds!
1) Go to https://t.co/YnyWbp4rG1
2) Enter your image mask/illusion
3) Write a prompt
4) Click “Run” pic.twitter.com/Ltj1kAKdUg
Then, as usual, enter a text prompt describing your desired scene.
In almost no time at all, you can imagine run-of-the-mill alleyways as oddly pleasing symmetrical scapes, or turn rooms into whirlpools.
These were some contenders/experiment gens I made on the same day. pic.twitter.com/UVbijSBsu5
— MrUgleh (@MrUgleh) September 13, 2023
2. @shapiro500 https://t.co/ERN1PORwLw
— Alvaro Cintas (@dr_cintas) September 17, 2023
AngryPenguin, the tool’s creator, says the possibilities don’t end there, adding that you can even upload avatars or logos to immerse them in visionary settings.
3. @dreamingtulpa https://t.co/QZZGAR6Ued
— Alvaro Cintas (@dr_cintas) September 17, 2023
Had a play around, this is about as good as I was able to get. pic.twitter.com/MZcmsUXYtO
— robot.chihuahua (@_candroid) September 15, 2023
unnh pic.twitter.com/C89jD25b82
— 🎠(@deepfates) September 11, 2023
I won't pic.twitter.com/zghNa3kFxH
— charlesai (@HODLFrance) September 19, 2023
will this be displayed in museums? pic.twitter.com/hDXKzjgDEs
— gaut (@0xgaut) September 17, 2023
[via Alvaro Cintas and MrUgleh, cover image by MrUgleh and featured with permission]
