As best as I can determine, the following images were created by AI, in conjunction with Stable Diffusion, a neural AI network. They are pretty amazing to me. Some are easier to see than others. Back away, squint or view as thumbnails for best effect.
And my favorite:
Wow. atleast5characters
Haha! The middle one was easiest, then the bottom, and then I burst out laughing at the top one.
That is amazing. Pretty easy for me to see when I take off my glasses. Wow. I wonder if it really is AI and what the prompting was to do this.
Wow, that’s impressive.
I’m currently a web developer, have worked as an illustrator in the past, and consider myself a halfway decent amateur hobby photographer.
I’m feeling more useless every day. 
There’s a specific set of AI image tools used to make these; Stable Diffusion with Controlnet’s QRCode_Monster model. Stable Diffusion creates the main image, and Controlnet/QRCode_Monster is used to integrate an uploaded secondary image. It was originally developed to make working QR codes out of images but it works for simple shapes and text as well. Somebody noticed how well it integrates text a couple of months ago and voila, a whole new genre of optical illusions was born.
Well, they are not really “optical illusions.” They are just images with different emphasis in their high and low-resolution components.
Dali did it ages ago:
I could not for the life of me figure out what was so remarkable about three perfectly ordinary pictures, and eventually leaned back in my chair still wondering. Since I’m near-sighted and don’t wear my glasses when using the computer, things got a bit blurred and suddenly the hidden magic revealed itself! 
And yes, it was the middle one that revealed itself first.
What am I supposed to be seeing? #1 is “odd.” #2 isn’t particularly. #3… ?
I don’t wear (or need) glasses.
Zoom out, make the images really small.
Try squinting so that your eyelashes make the image blurrier. It should really pop out at you by then.
Squinting did the job.
Which fits the bill. Oxford’s online dictionary defines them as "Something that tricks your eyes and makes you think that you can see something that is not there, or makes you see something as different from what it really is.
And while amazing, Dali’s Lincoln is far from the first of this type. But the point of this post isn’t that this type art exists, or is new, it’s that AI can create it in such a stunning and realistic fashion.
I’m was an illustrator in the past as well, and I feel your pain. When AI first started creating “art”, I wasn’t impressed, or worried. Those days are over. It’s just unbelievable what it can create.
I can envision ‘artists’ of the near future having AI generate an artwork and then just painting over it. I wonder if AI will devaluate future “art”?
That worked for me. Thank you. I couldn’t see it by squinting or moving back.
I KNOW! And this is nothing new. I’ve long been angry at Adobe for making things easy.
[geezer voice] Back in mah day …
… you couldn’t say “Hey, Photoshop, can you move that person over, and remove the background?” No, it took a Serious Graphic Designer™ hours, meticulously using a pen tool to trace around the windblown hair of that person…
So, I’ve been resenting computers threatening my job for decades.
I had to hold my phone at about arm’s length and slightly unfocus my eyes, and the words just popped out.
At least that was easier than those 3D pictures that were all the rage back in the 90s. I never could see anything in those.
Yep, same here. My Photoshop skillz were pretty on point back in the day, and I could trace around windblown hair and remove and replace backgrounds with the best of ‘em.
On an amateur photography site I sometimes visit, a guy has been posting a series of AI-generated images of various celebrities, with an almost photorealistic, yet painterly style. The quality is astonishing. A bit cheesecake, but safe for work. Here’s AI Taylor Swift:
Yeah, as Dorjan said, this was a big fad that came and went in the AI art community a couple of months back. Not only can you hide text, but also other images. Here are a few more examples:
BTW, you can create your own here:
I bookmarked that weeks ago, didn’t think of anything to try, and then forgot about the page until just now.
