[wolfsir]

Thanks for the link— hadn't heard of Smoothlife. Very cool. Of course the same technique could be adapted in some form.

One advantage of LifeHash using a small world-size is that the pattern inevitably becomes stable (repeats a state) after a reasonable number of iterations. I do have a hard-coded limit for the number of iterations, but it's just a backstop and large enough so I never see it hit in practice. A much larger, more complex simulation could go on for much longer, and since I really want to end up with a static icon, I want to limit the time and complexity of computing it.

Another advantage of using a small world size is that the initial 16x16 cell grid includes exactly 256 cells, which is the same number of bits in a SHA256 hash. Any hash, visual or otherwise, should be extremely sensitive to initial conditions, so a small change in input yields a massive change in output. Having the same number of cells as hash bits means every single bit counts with equal significance to the outcome.