> "Pixels have minimal impact on the ambient vision but a large impact on the focal cone causing further delay as the brain attempts to process the increased detail, when pixelation in camouflage is done correctly (color and scale) the brain will confuse the background noise with the pixels, removing the anomaly as a threat or delaying the identification of a threat. (6)"

I read that as having more to do with the length scale of features represented by the pixels than the shape of the pixel itself.

From the biological perspective, the same phenomenon should happen with dots or triangular shapes

I think squares are easier to fit together than dots and probably triangles.

If you used dots you have irregular spaces between them. This may or may not be a good thing, but it is certainly going to be different to the dots themselves.

Hexagons could work though.

But judging by the linked site and some Googling, I'm going to say no one knows for sure because no one has tested it.

Pixelation by itself does not produce any camo effect. It just happens to be easier and cheaper to produce fractal camo patterns on clothing out of pixels - it can be literally printed. And at any range at which the pattern is likely to work anyway, the individual pixels aren't really visible.

More modern (and expensive) patterns like Multicam or A-TACS do the same thing without pixels, e.g.: https://www.camopedia.org/index.php/File:Multicam.jpg

> It just happens to be easier and cheaper to produce fractal camo patterns on clothing out of pixels - it can be literally printed.

I have never seen non-printed camo patterns. https://en.wikipedia.org/wiki/Flecktarn is printed too.

Maybe it's easier to implement the pattern generation algorithm for pixel output?