How to explain the picturesque patterns in François Brunault's matrix?

How to explain the patterns in the matrix defined in François Brunault's answer to the question Freeness of a Z[x] module depicted below? --

Choosing colors according to the highest power of 2 which divides an entry, we can obtain for example the following pictures:

Taking instead highest power of 3 dividing an entry, we can obtain for example this picture:

Coloring zero entries white, negative entries red and positive entries blue yields the following picture:

And finally, marking non-zero entries black and leaving zero entries blank, the pattern looks like this:

Added: Assigning colors according to numbers of prime factors of the entries, we can obtain the following picture:

Added on June 15: Choosing colors depending on entries (mod 47), we can obtain the following picture (nonzero entries divisible by 47 are colored red):

-
Have you tried plotting the number of prime factors in colors? My plot: s24.postimg.org/5rj4gkkat/rxdivisors.png –  joro Jun 14 '13 at 12:43
@Joro: Thanks. Yes, I tried also this- Though computing such picture in the same resolution (512 x 512 pixels) as the others seems to take notably more time, so I computed only smaller ones. –  Stefan Kohl Jun 14 '13 at 13:31
Factoring the matrix entries is not always easy -- the factors are sometimes not all very small. Until line 140 of the matrix, the largest second-largest factor which occurs is 278454866719523378494429. –  Stefan Kohl Jun 14 '13 at 14:52
Primes in the matrix appear very rare. –  joro Jun 14 '13 at 15:32
Thank you for spotting these nice patterns! Not a precise explanation, but it may not be surprising that the entries of the matrix have nice $p$-adic properties. Indeed, any generalized polynomial map (in the sense of the question you link to) extends to a map $\mathbf{Z}_p \to \mathbf{Z}_p$ which is continuous (and in fact, 1-Lipschitz). –  François Brunault Jun 15 '13 at 18:29