Populate geometry with single point through iterations (Anemone) produces a pattern instead of randomness

Replies to This Discussion

Permalink Reply by Yoann Mescam (Systemiq) on November 30, 2015 at 11:42am

The seed is made with the iteration's counter.

Seems to work, perhaps you should upload your modified script for inspection.

Permalink Reply by martyn hogg on November 30, 2015 at 2:09pm

Could it be that because the initial points seem to have a pattern the algorithm will tend to place certain size rectangles at certain points because of the way it checks inclusion, intersection etc.?

i.e. it is randomly placing the rectangles but after the inclusion / intersection rules have been applied it will more likely leave certain sized rectangles in certain places so the pattern of the initial points is reflected in the final placements.

If that makes sense?

Permalink Reply by Yoann Mescam (Systemiq) on November 30, 2015 at 4:39pm

No, it is really a stupid algorithm:

 iteration i  : seed = i+S (S=slider) ;

                  PopGeo(seed, boundary) -> point p

                  choose randomly (seed) one of the objects to pack, generate rotated and scaled versions at p, check which ones fit ok, keep the first one;

Repeat

Permalink Reply by Tim on December 1, 2015 at 1:46am

Hi Systemiq,

Thanks for sharing the packing example back then, it really helped me grasp the looping components (although I prefer anemone in the end for its accurateness with the wires, you will see in the attached file))) 

Could you please show the centroids of the rectangles (without the rectangles), at first glance rectangles seem to be placed randomly, however pattern seeking brain can kinda spot that some of them are being placed in diagonal fashion. Again, it may be that I'm just seeing things.

I would also like to ask you about these bad guys (while I have a chance)))

I can't quite get what are they doing there? The definition seems fine without them but I'm clearly missing something here.

P.S. I've placed a piece of definition under the question, it should work just fine.

Attachments:

• Patterned Randomness.gh, 37 KB

Permalink Reply by Yoann Mescam (Systemiq) on December 1, 2015 at 10:58am

You forgot to take the preexisting population of points into account, in the PoPGeo component. (Thats what the D1 in anemone is made for).

 The bad guys are probably just old stuff i forgot to remove when i adapted the code from hoopsnake to anemone, or unnecessary intialization check/debug.

Permalink Reply by Tim on December 2, 2015 at 2:03am

I love you man,

Yes, indeed, somewhere in the process I unplugged it and completely forgot about it, thanks for pointing this out. 

Now I kinda understand why it was patterned when it didn't know about the pre-existing points.

This is for someone who is new to this:

Permalink Reply by Tim on December 1, 2015 at 2:08am

Hi there Martyn,

I get your point, but, ah, I wish I had that awesome tool for making GIFs just from your screen, it would have been so much easier.

The thing is that they start out with plenty of space around and never intersect, it is when the area is filled it starts to check stuff. But from the beginning you clearly see how definition kinda jumps randomly around but with all these rectangles being place (like a history) you notice the pattern.

Permalink Reply by martyn hogg on December 1, 2015 at 3:48pm

Oh, you mean it does the same thing every time? It always will unless you run it with a different set of random seeds. So, for example if you make the seeds equal to the Anenome counter + a parameter you change at the start of the run then each run will be different. If the random seeds are always the same each run then the results will always be the same!

Permalink Reply by Tim on December 2, 2015 at 1:47am

Hey Martyn,

Yes, I know, this is why, if I'm not mistaken, David called it "pseudo random" somewhere on the forum. 

However fiddling with the seed count locally won't change the picture globally, the filling of the area will still obey the pattern.