Populate 3D very very slow: help!

Replies to This Discussion

Permalink Reply by ng5 Alex on April 11, 2015 at 9:28am

80000 are lots of point

take a look at this post

cheers

alex

Permalink Reply by Luca Borro on April 11, 2015 at 9:38am

Thanks, I've seen but in post are 2.000.000 points...I need "only 80.000" points!
Is there a suggestion for obtain. It?

Permalink Reply by ng5 Alex on April 11, 2015 at 10:03am

unfortunately not.

since i am struggling to develop a workflow in gh for point clouds the best, but not optimum, solution was found here.

but i don't think its usable in your case since you are dealing with random (parametrically controlled) points.

cheers

alex

Permalink Reply by peter fotiadis on April 11, 2015 at 12:02pm

Hi Luca (are you still after ... bones? he he)

My 2 cents on that matter:

1. I assume/suspect/guess that David had masterminded some method to yield "even" random distribution on these GH components.

2. See this? It's a abstract (for the moment) layout of some WIP thing: Imagine a region where "evenly" random points are placed (and then a random zNoise is added) - then a ball-pivot/delauney triangulation is applied ... then ... :

2. I use my method to create "even" points (I suspect that David's is way better/faster/cooler ... but anyway): after a random point is found (inside the region) an additional check is performed: think of the point as a "candidate" that must "pass" a 2nd constrain: if the min distance from all the already found (random) points is smaller than a user defined one > reject and try again (the "try-again" thing [call it: min distance "loops"] is also user controllable). Thus that C# captured attempts to place 122 points but due to (a) the min distance constrain AND (b) the low (about 8 in this case) amount of "try-again" loops it finishes with "only" 59 (not a big deal for this case). The interesting part is that the attempts required are 1573 (~30 times the random points returned). Of course there's a lot of factors affecting this 1573 (variable) thing ... but don't stick to that.

3. So if David uses a "similar" culling method (add some " " more, he he) ... for 80K points ... well we are talking about a BIG number of attempts.

I can provide you with a "no-even" random points C# that (I assume/guess/hope)  can speed things a bit (after all:  who's gonna notice an "even" random distribution of 80K points within a micro cube?).

best, Peter