FRACTAL-LIKE PROPERTY:

	Fractional Brownian motions (fBm) are members of the class of "1:f noises",
	that is, those signals in which the contribution of each frequency
	to the power spectrum is nearly inversely proportional to the frequency.
	Additionally, the increments of fBm are statistically self-similar ...
	Intuitively these features of fBm indicate that we may observe a
	sample of one of these functions at any scale and perceive identical
	statistical features.  A surface generated using fBm would thus possess
	macroscopic features up to the order of magnitude of the overall surface
	generated, corresponding to the lowest possible frequencies in the
	Fourier spectrum of the sample, as well as arbitrarily small surface
	detail, corresponding to the higher frequencies in the Fourier spectrum.

-- Alain Fournier, Don Fussell, Loren Carpenter,
"Computer Rendering of Stochastic Models,"
Communications of the ACM, Jun 1982, p. 371.


``According to Ken [Perlin], the Brownian motion info above isn't enough to
create the appearance of turbulence.  It only appears turbulent if it has the
multiscale discontinuities introduced by the absolute value in
the turbulence function.''


[THANKS TO Aaron Hertzmann for contributions in this page]