As the tile is scaled down, less and less of it's pixels will be visible on the screen, at some point there won't be enough pixels to accurately sample the original image (the nyquist rate), this causes the data to overlap itself (aliasing (in the frequency domain, very unintuitive in terms of images)), the aliasing is periodic in the frequency domain, thus it'll be periodic in the original domain too, so you get this periodic sort of artifact all over the image
15:20
You may wonder why it doesn't happen with just any image, well it does, but most images in the frequency domain don't really have a distinct spectrum, however, this sort of repeating pattern has a very distinct high magnitude high frequency component, so when it overlaps in the frequency domain you get more high magnitude periodic components
15:21
Anti aliasing methods like msaa, fxaa and such sort of blur the image at parts where there is very high frequency repeating patterns, which makes the image less periodic, which makes the aliasing less obvious(edited)
Learath2
Anti aliasing methods like msaa, fxaa and such sort of blur the image at parts where there is very high frequency repeating patterns, which makes the image less periodic, which makes the aliasing less obvious (edited)
Anti aliasing methods like msaa, fxaa and such sort of blur the image at parts where there is very high frequency repeating patterns, which makes the image less periodic, which makes the aliasing less obvious (edited)
