Particle Fluid Collisions

[zglynn]

Hi everyone,

I am trying to make a goo-like effect for a student film using Houdini's particle fluids. I have the consistency of the goo how I want it. I ran some tests where I sourced particles from a sphere and let them poor over a static object(tube). It worked great. The collisions were fine and everything worked as I expected it to. Then when I tried to use the goo in the real shot, things went really wrong.

I need to fill a sphere, (that is made up of two pieces which are animated to pull apart) with the particle fluids, let them settle, then pull the sphere open to poor out the particles.

The Problem:

When I try to fill the sphere with particles, the particles immediately explode out.

I first checked my collision surface. No matter what I did I could not get the two halved spheres to have a nice collision surface using ray-intersect. I changed it to minimum and set the offset up a little. This gave me a nice collision surface, with no holes or weird volumes, but the problem was the same; the particles would immediately explode out.

Some Settings:

particle separation = 0.1

bounce for particles and static objects = 0

friction for particles and static objects = 1

static friction for particles and static objects = 1

Any ideas would be greatly appreciated.

Thanks,

Zach

[mightcouldb1]

Increase your substeps on the DOP network. Depending on the speed of your geometry you might have to go really high. I was up to about 14 at one point.

[zglynn]

Jason

Thanks a lot. That's all it took. I cranked up the substeps and all is well. I will post progress later.

Zach

[zglynn]

OK

So increasing the substeps (up to 20) worked fine to get the goo filled in the two egg shells and settled. Then when I try to sim with animation on the egg shells (using a cached out .sim file from the settle as my initial frame), the particles explode out. I try it at like 75 substeps and they still exploded. At first I thought it might be my internal pressure, which was set at 500. (that's in part how I have been getting the particles to "goo togther") I set the internal; pressure down to 10 and they still exploded. I've attached a hip of a test goo that I did(which works great) and I've attached the hip of the goo exploded out of the eggs.

Any help would be greatly appreciated.

Zach

goo_test.hipnc

Goo.zip

[stevenong]

Your egg shells are not closed for the SDF generation so the particle fluid sees the egg shells objects as a "ball".

Turn on "Show Collision Guide Geometry" under the Collisions > Volume tab of the egg1 Static Object DOP and you will know what I mean. Before you turn it on, I advise you to lower the number of divisions first as the settings you have makes the generation really really slow.

Cheers!

steven

[zglynn]

Steven

Thanks for the reply. I am little confused about what you mean by "Your egg shells are not closed for the SDF generation..." I looked at the collision surface again and it looks fine.

I am using "Minimum" as the mode instead of "Ray Intersect" and it actually calculates fairly quickly. Now I've never used "Minimum" as the mode before,(I've always just left it on "Ray Intersect") but I was having trouble getting the collision geometry to be accurate with the egg shell so I started experimenting and found that changing the mode to "Minimum" and adding a slight offset gave me a nice looking collision geometry.

So to me the collision geometry looks okay, but is there something I'm just not understanding?

Thanks,

Zach

[phong]

For collision a field is generated, which defines for every point in the field whether it is inside (negative) or outside (positive) of a specific geometry. This field is the SDF and to compute it most algorithms require the geometry to be closed. If it is not, errors might occur which might cause your particles to appear inside the collision object to the solver (which in fact is not true) and thus result in crazy collision behavior. I don't know what the collision guide geometry actually displays, maybe it's not showing the true SDF that the solver gets?

Phong

[zglynn]

Thanks for all the help I'll just keep plugging away at it, and if anyone else has any ideas I'd love to hear them.

Zach