"In fact, people have actually made spatial pattern generators that allow you to input the frequency profile that you want, and get the corresponding point pattern out. It’s really quite neat, and I highly recommend reading this paper so you can see some other possible noise parameters, like anisotropy."<p>I had to hunt this 'custom colour' noise paper out since that link was dead. It is quite neat. Here it is on ACM in case anyone else is interested:<p>Point Sampling with General Noise Spectrum (2012)
<a href="https://dl.acm.org/doi/10.1145/2185520.2185572" rel="nofollow">https://dl.acm.org/doi/10.1145/2185520.2185572</a>
The same algorithm is described more clearly in Bridson's paper from 2007:<p><a href="https://www.cs.ubc.ca/~rbridson/docs/bridson-siggraph07-poissondisk.pdf" rel="nofollow">https://www.cs.ubc.ca/~rbridson/docs/bridson-siggraph07-pois...</a>
Something sounds odd. If points are very close, they have high spacial frequency. If I reject points that are too close, I'm excluding high-frequency points and keeping only low-frequency ones. A distribution that favor lower frequencies and filters out higher ones is red, not blue. Am I getting it wrong?
I was looking through the source code for Bevy's Screen-Space-Ambient-Occlusion (SSAO) implementation and noticed they are using blue noise as well:<p><a href="https://github.com/bevyengine/bevy/blob/56d559102858d4ce8a5b5e3bccd7053dd687a197/crates/bevy_pbr/src/ssao/mod.rs#L787">https://github.com/bevyengine/bevy/blob/56d559102858d4ce8a5b...</a><p>That link takes you to this shadertoy:<p><a href="https://www.shadertoy.com/view/3tB3z3" rel="nofollow">https://www.shadertoy.com/view/3tB3z3</a>
I would be very curious to hear if anyone knows how the color names are assigned?.. Is there any link to frequencies/spectra of light or is it kinda arbitrary?