Forgot to reply to this!
Regarding the Swedish fractal 2D diffuser.
Golden horn
The things I point out below may be all wrong if it's not a prime root diffuser. If it's based on some other math, it may work out great as is.. Hope so!
If you want to do PRD's in the same fashion, here are some things to keep in mind:
The big squares use a 8x8 grid. A criteria for a PRD grid is that it should use two numbers that are coprimes. These two numbers should add up to a prime number minus one. A 8x8 grid is not coprime and 65 is not a prime. It's probably based on a 8x9 grid which gives prime 73, with the last row being omitted. If you DIY, there's no need to omit the last row. The steps on these large squares are quantized to 40, 80 and 120mm - three steps where there should ideally be 72! Having a finer quantization step can't hurt..
The smaller diffusers are very very small, number theoretically speaking. They're a 3x3 grid, which again is a square, not coprimes. I bet these are based on prime 13 ((3x4)+1) with the last row once again being omitted. With such a low prime number, the performance of the diffuser will already be quite poor as is. Omitting the last row of such a short sequence doesn't seem like a good thing to do. I don't know how much it'll affect the end result, but I bet it's going to be better with a proper grid. Recall that it's the flat fourier transform / lack of correlation, that gives the diffuser its desired properties. Omitting the last row in such a very small sequence will lead to quite disturbed fourier transform.
Quote:
|
I give you the word, and point away!! (which word did you want????)
|
Ostkaka!
