Nabble has removed Mailing-list integration.
Posts created here DO NOT GET SENT TO THE MAILING LIST.
Mailing-list emails DO NOT GET POSTED TO THE FORUM.
So basically the Forum is now out of date, we are looking into migrating the history.

# Using recursion to create smoothed random variation ("red" noise) Classic List Threaded 7 messages Open this post in threaded view
|

## Using recursion to create smoothed random variation ("red" noise)

 Hello, I am trying to use a recursive function within a list comprehension to add some random component to both x and y coordinates of each point in a polygon (a circle actually). However I want each random component to be based on the previous random value in order to smooth the random component (and get some king of brownian/red noise). Obviously there are a few things which I don't understand because my code doesn't provide the expected results : the noise component is not smoothed. For instance, I am not sure if "n-1" in the function definition means using the results of the previous iteration in the current one or substracting 1 from n. Any help would be appreciated! / polygon_steps = 100; random_range = 10; function smoothrand(n = 0, range  = 1) = ( n == 0 ? rands(-range, range, 1) : smoothrand(n-1) * 0.9 + rands(-range * 0.1, range * 0.1, 1) ); start_points = [for(i=[0:polygon_steps])         let(a = smoothrand(1, random_range), b = smoothrand(1, random_range))         [cos(i/polygon_steps * 360) * (0.4 * a + 10), sin(i/polygon_steps * 360) * (0.4 * b + 10)] ]; polygon(start_points);/ -- Sent from: http://forum.openscad.org/_______________________________________________ OpenSCAD mailing list [hidden email] http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org
Open this post in threaded view
|

## A feature I'd like to see... import png or jpeg

 Have no idea how difficult this would be, but it would be helpful for some projects to be able to import an image file like a png or jpg as a 2 dimensional object, be able to scale and rotate and set luminance. It could then be used for more rapid prototyping of an existing object, or getting things like thread pitch correct… that kind of thing. Anyone else think this worthwhile? _______________________________________________ OpenSCAD mailing list [hidden email] http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org
Open this post in threaded view
|

## Re: Using recursion to create smoothed random variation ("red" noise)

Open this post in threaded view
|

## Re: A feature I'd like to see... import png or jpeg

 In reply to this post by boxcarmib On 10/20/2020 5:34 PM, Hugo Jackson wrote: ```Have no idea how difficult this would be, but it would be helpful for some projects to be able to import an image file like a png or jpg as a 2 dimensional object, be able to scale and rotate and set luminance. It could then be used for more rapid prototyping of an existing object, or getting things like thread pitch correct… that kind of thing. Anyone else think this worthwhile? ``` Are you expecting it to contribute to the geometry, or just to be sort of a backdrop that you could then line things up with? One wild-assed thought along those lines would be to import it as the texture of a 2D object, or even as the texture of a face of a 3D object. _______________________________________________ OpenSCAD mailing list [hidden email] http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org
Open this post in threaded view
|

## Re: Using recursion to create smoothed random variation ("red" noise)

 In reply to this post by amundsen Adrian gave a good idea on how to accomplish your goal. Here are some thoughts on why what you tried didn't work, in the hopes of helping you to better understand recursive functions. On 10/20/2020 5:22 PM, amundsen wrote: Obviously there are a few things which I don't understand because my code doesn't provide the expected results : the noise component is not smoothed. For instance, I am not sure if "n-1" in the function definition means using the results of the previous iteration in the current one or substracting 1 from n. ```/ polygon_steps = 100; random_range = 10; function smoothrand(n = 0, range = 1) = ( n == 0 ? rands(-range, range, 1) : smoothrand(n-1) * 0.9 + rands(-range * 0.1, range * 0.1, 1) ); start_points = [for(i=[0:polygon_steps]) let(a = smoothrand(1, random_range), b = smoothrand(1, random_range)) [cos(i/polygon_steps * 360) * (0.4 * a + 10), sin(i/polygon_steps * 360) * (0.4 * b + 10)] ]; polygon(start_points);/ ``` "n-1" means "subtract 1 from n".  Always. "smoothrand(n-1)" means "call smoothrand with one less than n".  The fact that it's being called from inside smoothrand() is irrelevant, except that you have to ensure that the recursion eventually terminates.  (Which you did.) First, you're calling smoothrand with n=1, which means that you'll always get exactly one recursion:  once with n=1 and once with n=0.  You might have been thinking of calling it with n=i, so that each call would recursively calculate the previous value and base its result on that value. However, second, you're talking about random numbers.  If you called smoothrand(0) and it gave you some number, and then called smoothrand(1), the two numbers would be unrelated to one another because when the smoothrand(1) internally called smoothrand(0), it would be using different random numbers than the first call. Also, it would be really expensive.  I believe you'd have 10,100 calls to smoothrand.  The first point, i=0, would be one call: smoothrand(0).  The second point, i=1, would be two calls:  smoothrand(1) and smoothrand(0).  The third point, i=2, three.  (That's six so far.)  The 100th point would require 100 calls to smoothrand.  The total, the sum of all of the numbers from 1 to 100, is (n²+n)/2, or 5050.  Doubled, because you call it twice.  (That's (n^2+n)/2 if you can't see the superscript properly.) _______________________________________________ OpenSCAD mailing list [hidden email] http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org