Sadly, an LLM might be better at this than most people.
They are always depth-inverted for me, so I’d be one of those people. I can see a web 2.0 rounded hole that I assume is the target.
Edit: it dawned on me that since my last serious attempt at a magic eye, a new invention known as “the world wide web” exists now, and answers to these questions are at my literal fingertips.
Sounds like the inversion effect happens if your eyes over converge (cross) rather than DIVERGE, which is what you wanna do for these images.
Yeah, you’re supposed to look “behind” the image.
Ironically, the word “CANCEL” is much easier to read in this instance if you cross your eyes. The button is recessed, and the word is level with the rest of the image if you cross your eyes. If you view this image the normal way with diverged eyes, it becomes super hard to read.
There are also a lot of JPG artifacts messing it up. Just look how poor the button looks when the layer is duplicated, shifted, and put in difference mode:
That makes sense but I genuinely never thought about the difference between crossing your eyes and staring out on how a magic eye appears. How interesting.
Hey so, there’s more than just 2 ways to view the 2 different kinds of images.
You can find more sweet spots in the focal depth where the image gets doubled our tripled and stacked up horizontally in your view. Like if the image is a sphere you can get it to pop out two intersecting spheres, or even three spheres in a sorta chain arrangement. In more complex images the overlaps get wild and its harder to explain but really fun to see.
