Le 16/11/2010 23:26, Samuel Benge nous fit lire :
> Le_Forgeron <jgr### [at] freefr> wrote:
>> ...in the mandel pool.
>>
>> Small picture is a classical Mandelbrot set, at usual scale.
>>
>> The other is about <-1.5099872678075,0>/2
>> with an iteration limit of 999 999 and a zoom factor of 99 999 999.
>> (just testing an exterior_type)
>  
(#declare n=399; value of fact & iter already provided)

pigment {mandel iter
	exterior 7,n
	interior 0,0.5
	pigment_map{
		[0.0 Blue]
		[((n-1)/n)*1/5 White]
		[((n-1)/n)*2/5 Red]
		[((n-1)/n)*3/5 Black]
		[((n-1)/n)*4/5 Green]
		[((n-1)/n)*5/5 White]
		[1.0 Red]
	}
} translate <1.5099872678075,0,Radius>
scale fact





> Did you use log smoothing for the exterior in the larger image?

No, I used my own patched version to get a value based on the iteration
number at exit(e):  (e % n) / n
With e the number of iteration at exit
and n the factor of the exterior

(instead of the classical type which return n / max iteration)

It does not fully cover [0..1], so I also made a type 8: (e % (n+1))/n
Still wondering which one is best.
(7 is easy to reserve the 1.0 for the interior, 8 has more coverage...)

> If so, do you
> know if it's possible to bridge iterations perfectly, or are there always going
> to be slight inconsistencies? I've gotten *pretty* close, but sometimes there
> are still visible artifacts. I had to adjust the size function to (x*x+y*y)-2048
> in order to get the best results. The coloring function reflects that
> adjustment:
> log(log(iters)/log(x*x+y*y))/log(2)/(i-1)-(i-1)/iters+512/(i-1)/iters
> 
> Rather ugly, I'd say, but it's the best I have come up with :\

Euuuuhhhh, I do not understand the previous block.

But it might be an interesting idea to have a type based on log(iter).
Thanks for that.

> 
> Also, what kind of system are you using to navigate the fractal? Are you taking
> parameters from other programs? What I'm doing now is placing a numbered, static
> grid over the entire image and alternating between scale and translate
> statements (using the grid as a reference), but there must be an easier way.

I tried Xaos & Xfractint, but I suck at finding / displaying the actual
coordinates, so I end up finding an easy to spot area of interest then
using increasing zoom within povray.

Post a reply to this message