forces on particles with goal

[dvelasco]

Hello!

I am still very much getting acquainted with Houdini, but every time I need to use it I have a very specific purpose. This one has me stumped for now, but I am sure that someone versed in Houdini dynamics will resolve it very quickly!

What I want to do is: fill a volume with spheres, make said spheres bounce off of each other on contact (like snooker balls), but also try and accommodate themselves onto the original shape (created by the volume). I have tried this with grains, but didn´t get too far. I am betting on doing it with particles:

This path seems a lot more promising! I can get the ball´s pscale and id attributes, which will be a boon later on, when I need to swap the balls for different emojis (yes, it´s for a TV ad!). 

What I need to do:

1. duplicate the current particle so I can use it as a "goal shape", with the same ids. This I can do.
2. add an initial offset to the particles from the surface, ideally from the normals of the original geometry (or a smoothed copy). I can somewhat get it by tweaking the iso offset of the vdbtospheres node, but I worry that the IDs will be messed up. Ideally, I would add an offset to them, but not add any velocity.
3. add a small amount of turbulence to the particles. This will eventually be animated to 0.
4. make them collide with each other, based on their sphere geometry. The eventual geometry is basically spheres as well.
5. make them try and reach their goal, and eventually settle down (a slider for this would be great!)

Seems very achievable, but I need some light on how to make these connections! 

After that, I still will need to swap the spheres for 4 different groups of geometries, but I believe that I can glean information from tokeru.com (Emit_packed_prims_into_RBD_sim), and make it happen. 

Any help with this will be GREATLY appreciated!  

This is the scene file: C03_008_DVA.hip  (if anyone is interested in taking a look at it!)

 

cheers, 

 

Diego

Edited August 18, 2020 by dvelasco

[dvelasco]

Just an addendum - I the isooffset node would work perfectly for the "offset from surface" part of the problem, except that it resamples the volume if you change the offset, which foils that path. 

[dvelasco]

Got a great path from this link:

Also, I will probably use the isooffset node, even if it´s just with an average position, as opposed to ids. Will try and work that in with the solution above! Seems very promising  Who knows, maybe I´ll figure this out by myself... but still, any help would be greatly appreciated! I will update this thread with my findings as I progress.