Here is an example to jog your memory

‘Magic Eye’ Pictures – Stereogram

(Can you see a horse? There is a horse in the picture!)

I wanted to see if ChatGPT could create one of these images for me. So, I uploaded the image as an example and asked it to review the image and tell me what it was:

“Certainly! The image you provided appears to be an intricate and colorful abstract pattern or design. The design consists of various shapes, possibly resembling abstract representations of fish or other aquatic creatures, intertwined in a flowing and repetitive manner. The color palette is vibrant, with a mix of warm and cool colors. Overall, it gives off a playful and visually engaging feel. It might be a piece of abstract art or a decorative design for textiles, wallpapers, or other mediums.”

It was having a little trouble, but eventually it seemed like it understood what I was asking for. I had to give it a few more details and explanation about what I was looking for and it finally did ‘understand’ the request.

Ah, I see what you're getting at. The image might be a type of stereogram, commonly known as a "Magic Eye" picture. Stereograms are designed to have a 3D image hidden within them that can be seen when viewed in a certain way, typically by focusing the eyes in a specific manner.

However, due to the limitations of my current capabilities, I'm unable to "view" the stereogram in the same way a human does to perceive the hidden 3D image. If you're trying to see the hidden image, I'd suggest focusing your eyes either behind or in front of the image (different techniques work for different people) until the patterns begin to shift and the hidden 3D image emerges. If you have a description or title for the image, I might be able to provide more context about it.

The answer came back that it didn’t really know how to view or even create an image like this. I thought that was interesting and why not? I asked if there was some mathematical or abstract way of decoding the image, which it attempted to do, but rather unsuccessfully.

However, it did begin to generate some interesting imagery based on the requested prompts. Here are a few examples of what was produced:

I continued to prompt, while poking and prodding at the system to try and make it understand what I wanted it to create and although the images and output did improve, the overall results were sub-optimal, in regards to creating a ‘Magic Eye’ picture - but exceptional at creating beautiful and profoundly unique artwork like I’ve never seen before.

Here are a few examples of the creations, after several prompts.

After several more prompts asking to improve the images to be more like a ‘Stereogram’, I decided to take a break.

Here are some of the best images that were generated at an attempt to generate a Magic Eye image:

In the end, ChatGPT was unable to create a true ‘Magic Eye’ poster, but did get quite close. With continued experiments around the prompting, I’m sure you could manage to produce some incredible Magic Eye artwork.

Perhaps someone will master how to prompt for such an image… and perhaps that person could even bring back it’s popularity to the 2020s.