Lots of particles with Toxiclibs

janwillem..

Lots of particles with Toxiclibs

in Contributed Library Questions • 1 year ago

Hi all,

I am using VerletPhysics from Toxiclibs to create a particle system. For each particle I use a spring that is attached to a fixed point (a locked particle), so that all particles are attracted to this point. At the same time I use a VerletMinDistanceSpring for all the particles to give the particles a minimum distance from each other. When I click the mouse, new particles are added, and the result works nicely and the particles are nicely stabilizing in a somewhat circular form around this target point. This is what I want to achieve.

However, it is not very effective to have a spring between all of the particles of course. I notice that at around 200 particles things start to slow down.

I have tried various things, like adding and removing springs if particles are a certain distance apart. Drawing to an offscreen buffer (PGraphics), etc.

But I haven't found a good solution to working with many particles this way yet. So, has anyone a good suggestion for a strategy for this situation? In short, I want to optimize my code by not having each particle have a spring connection to every other particle, while still make sure it has at least a spring connection to the target point, and also to the minimal required number of particles to maintain a 'force-directed layout'.

Thanks!

Replies(6)

amnon.owed

Re: Lots of particles with Toxiclibs

1 year ago

Why not use a negative AttractionBehavior to keep the particles apart (force field around each particle) plus a string to keep each particle attached to the fixed point?

toxmeister

Re: Lots of particles with Toxiclibs

1 year ago

correct! :) also see bundled demo(s) for usage (check in /verletphysics/examples folder)

http://openprocessing.org/visuals/?visualID=17191

http://openprocessing.org/visuals/?visualID=22516

http://postspectacular.com | http://toxiclibs.org

janwillem..

Re: Re: Lots of particles with Toxiclibs

1 year ago

of course, why didn't I think of that!! Thanks guys!!

toxmeister

Re: Lots of particles with Toxiclibs

1 year ago

Hi Jan, just saw your tweet, but thought it's better to comment here...

The fact is that these kinds of problems have O(N^2) complexity and without any further added intelligence don't scale well at all. If you want to have that many particles interacting with each other you'll need to do some culling of particles which will certainly not have any (noticeable) effect on the current one. Question is how you figure out quickly what might have impact and what doesn't. That's what spatial subdivision trees are for, e.g. quadtree/octree, or spatial hash maps, binning systems etc. To plug something like this into toxiclibs, would require your own AttractionBehavior implementation (easy) and integrate some of these concepts into a custom VerletPhysics (spatial selection logic only need to be plugged into one or two places). Might be a good addition for the library in general... wink :)

http://postspectacular.com | http://toxiclibs.org

janwillem..

Re: Lots of particles with Toxiclibs

1 year ago

Hi Karsten,

thanks for elaborating more on this!! At the moment I have found an alternative solution where I require less particles, so it works for me now. But it totally makes sense what you say, and it would definitely make a good exercise for me, so I might give it a shot ;-)

Will something like this be covered at your upcoming Resonate workshop? I'll see you there for sure!

Thanks again!

toxmeister

Re: Lots of particles with Toxiclibs

1 year ago

Great to hear you'll be @ Resonate too!!! Always good to meet IRL. Am only planning 50% of the workshop and leave the rest open to the group's interest and/or pain points.. we can definitely talk about this topic in more detail, but won't have time to go into actually trying to implement it there & then (it's only for 1 day after all)

Best, K.

http://postspectacular.com | http://toxiclibs.org

Top Reply