Collision of fluid particles?

   7877   5   2
User Avatar
Member
9 posts
Joined: Oct. 2008
Offline
Hello everybody,

I just started with Houdini, so I'm still quite confused about a couple of things. Here is my first big problem.

Fluid particles are squirted from a standard emitter against a wall. I'd like to have some control over the impact, i.e. change the angle at which they bounce off the wall so the fluid spreads out evenly over the wall instead of splashing around. Now I already figured out several ways to do that with particles, for example changing the Gain Normal in a Collision POP, killing the particles on impact and generating new ones with Split POP and so on. So I created a fluid particle network via the shelf and thought I could exercise some control over the particles in the externalpopnet that's created with it. But only very few things I do in externalpopnet have any impact on the fluid particles. In the attached file I added a collision node which doesn't seem to do anything.

The interesting questions would be:
1) Why does this setup not work?
2) Is it actually possible and advisable to influence fluid particles in this way?
3) If not, how else could I do this? I couldn't find anything like the collision and group controls of particles for dynamic objects, so it doesn't seem to be possible to have this kind of control over a fluid on the level of the DOP network.

As I said I'm new to this, so it might be some very obvious mistake.
Any help would be greatly appreciated!

Attachments:
fluidCollision.hipnc (241.5 KB)

User Avatar
Member
9 posts
Joined: Oct. 2008
Offline
Bumping this topic since I have a similar problem again (got around the last one in a different way). Does really nobody know anything about this? What do I have to do to get collisions in the external popnet to work? Right answer might win a washing machine… :wink:
User Avatar
Member
258 posts
Joined:
Offline
I took a quick look at you file…

There is no collision object even in your autodop network, the only thing that is in your autodop network in the particle fluid emitter and the various stuff that goes along with it.

In your case create your collision geometry at the object level, select it and under rigid bodies select static object. Once done dive into the autodop network and you should see a new branch with a static object, a rigid body solver and a new merge node.

Under the collsion tab of the static object click on “show Guide geometry” and you should see the collision representation of the geometry. If it is not what it should look like up the resolution of the collisions until it is done.

Notice in this example I have just disable gravity by selecting the node above it and siming from that. If you want gravity just turn it on and rotate the geo so it collides.

Attachments:
sdf_collide_01.hipnc (243.5 KB)

User Avatar
Member
127 posts
Joined: Nov. 2008
Offline
also my guess would be Since simulations are different than cooking particle data, you might or might not get full control over the Particles. just like what happens when we try using an animated geometry in RBDs.
(again I might be wrong, i would love it if any one clarifies it).
Bhavesh Pandey.

https://bhaveshpandey.io [bhaveshpandey.io]
User Avatar
Member
1145 posts
Joined: July 2005
Offline
That's correct Bhavesh, there is essentially no “art” direction with simulations. You can set conditions/parameters for a simulation, you can even animate some of them, but that is done at the beginning of a simulation. During a simulation run, what you get is what you get.
You can increase you control by breaking your simulations into components and layers.
“gravity is not a force, it is a boundary layer”
“everything is coincident”
“Love; the state of suspended anticipation.”
User Avatar
Member
9 posts
Joined: Oct. 2008
Offline
@sl0throp:
Thank you, but you misunderstood my intention. I know how to make collisions work inside the dop network. What I was trying to do is to make the particles collide in the externalpopnet inside the particle_fluid node. Since it is possible to use forces, groups and other particle stuff in there to influence a particle fluid, I guess collisions should also work.

@Bhavesh, probbins:
I've read several comments on how great Houdini is because it allows exactly that kind of art direction even with fluids/RBD, and there's an example around somewhere here or on odforce.net where a particle fluid is following a path which is done via the external popnet. Another example was a fluid forming a letter, accomplished in the same way.

I figured you should be able to do everything to particle fluids that you can do to particles. But what doesn't seem to work at all is collision in the popnet. That would be a pity if this wasn't possible.
  • Quick Links