Is it normally the case that importance (and therefore sample density) correlates with luminance? Or is this just an example and not all images follow this pattern?

I feel like this sampling method would not work as well if we were to render an image with roughly equal luminance everywhere but with many changing colors