Diffusion Limited Aggregation

[Tesla's fan]

 

http://paulbourke.net/fractals/dla/

 

I do this in pop just to make you convenient to learning principle.In fact I think just add a new random point near by the already existent points is a more efficient way.According to the time of their existence, the tree points more longer existence the feed(walker) points being addad probability more smaller,but I haven't tried.But pop is more controllable，you can use popvop/wrangle etc.and interact with other dynamic module.About efficiency,I also thought,my algorithm is a little different.I not use tree points for pointcloud, but use feed points for pointcloud. Because the number of feed points basically gose to constant, but tree points will gose to more and more and explode. You can check and test my file. My algorithm can also be improved.My algorithm is each line generated a new primitive, but in fact this can seriously reduce the efficiency with the increase of tree.The improved algorithm is unnecessary every line generated a new primitive.When a feed point near the tree point and before connect it,we can check the number of edges has been connected of this tree point,if edge > 1 then generated a new primitive, else use the existing primitive of this tree point.But I haven't tried this algorithm. 

dla.hipnc

Edited September 3, 2016 by aty84122

[Farmfield]

Such an awesome setup. I'm so gonna dig into this. Much appreciated.

[Tesla's fan]

3 hours ago, Farmfield said:

Such an awesome setup. I'm so gonna dig into this. Much appreciated.

I'm glad u like.

[Tesla's fan]

I tried many ways.I have no idea how to reduce the primitive and vertex.I
found if connect points in dop can really slow down the process,so I do
this in sop. Here is dla_v2

dla_v2.hipnc