Hidden Text

Your eyes are not a camera. Before anything reaches you as “a picture”, the visual system has already split the scene into coarse structure and fine detail — low and high spatial frequencies — and handed them to different channels.

This image puts two things in the same rectangle. The word ILLUSION is there as a heavily blurred shape: large, soft, low frequency. On top of it sits a field of hard two-tone noise: small, sharp, high frequency. Up close, the noise wins. It is the loudest thing in view, and your fine-detail channel reports a mess of dots.

Now increase the distance. Every feature in the image shrinks in angular size, and the noise shrinks past the point where your eye can resolve it at all — the fine grain averages into a flat grey. The blurred word does not disappear, because it was never made of fine detail. What is left is the coarse structure, and the coarse structure is the word.

Squinting does the same thing by a different route. Narrowing your eyelids and defocusing throws away the high-frequency content before it is measured, which is why the trick works without you moving at all.

The unsettling part is that nothing about the image changed. The same pixels, on the same screen, carry two different messages depending only on how far away you stand. Whatever you are seeing at any moment is not the image — it is the part of the image your eyes are currently able to resolve.

First described

After Aude Oliva / Antonio Torralba / Philippe Schyns, 2006
Source: “Hybrid images”, ACM SIGGRAPH

Two layers are combined: one carrying high spatial frequencies, one carrying low.

See it in the timeline