flip fluid morph and settle

[whodini]

I'm trying to do a simulation with flip fluids that involve morphing one object into another and settle quickly. I have an rbd simulation of a sphere hitting the ground and breaking into a few large pieces. I have the pieces turning into liquid and create a nice splash. I want the liquid to attract upward and take the shape of a goal object as closely as possible and then settle.

I did a bunch of searching but I cant seem to get this sort of thing to work. I tried using the sdf of my goal object in combination with a field force but no luck. I tried creating a bounding box of my scene and fill it with points and have each point's velocity pointing toward my goal object but that doesn't seem to work either. Any help or ideas would be appreciated.

[cubiccube]

I've played around with this in the past and, if I remember correctly, my solution was to use 2 separate simulations: one sim for particles as they move toward the goal and another for particles once they enter the goal. Basically, the goal in the first sim acts like a sink (removing particles) and, in the second, as a collision object. I'm sure there's a more elegant solution to this..hopefully some pros will chime in.

[Annon]

So you have created your velocities pointing towards the goal?

In a vopsop (sopsolver) add the velocity to Position.

[ninjump]

Also workig on something like this - the problem being that for a believable effect you need to have the candidate object points pull the closest points of the target object, then REMOVE these points from the list of available target points so that the next candidate particle grabs something else.

As far I know (limited Houdini experience), popnets evaluate the system as a whole before moving on to the next node, as opposed to going through particle by particle. Any ideas on how to get around this?

Edited April 12, 2013 by ninjump

[Annon]

I don't think you need to be that pedantic about it, but If you are doing like for like, 1k points source, 1k points target, then you could look up just the pos of the corresponding ID... ie:

pt332 in input 1 is looking for pt332 in input2.

[Annon]

Also, I generally don't use pops and do the majority of my particles in SopSolvers.

Same sh*t, different context

If I need proper sourcing, I'll have a dopnet with pops handling sourcing/killing and sops doing everything else.

[eetu]

There is a nice thread on the SideFX forums about directable fluids/gases, where a couple of people implemented a related paper with volumes in DOPs.

I can never find the thread when I need to, though.. Anyone remember the title?

[Annon]

This one?

http://www.sidefx.com/index.php?option=com_forum&Itemid=172&page=viewtopic&t=21558&highlight=

[eetu]

Ah found it, descriptively titled "bullet time, "matrix effect"", http://www.sidefx.co...ewtopic&t=14844

[whodini]

I finally got this working. I have a sop solver plugged into my flip solver and inside the sop solver I'm taking the "Geometry" flip particles and assigning each one a velocity that points toward a nearest point on my goal object. I had to fill the goal object with points in order to get some decent results. Then I have those goal points deleted based on distance between a flip particle and a goal point. I then added some basic velocity cutting into the mix to get the velocities to cut by half as they start to go wild (works like drag but keeps the nice behavior of the sim rather than affecting everything at once. I'll share more soon. Still working on it.

[Annon]

It would be quicker to do it at the volume stage rather than per point I would imagine.

[whodini]

hmm. I actually tried that and I do think it was a bit quicker but the result was much different at volume stage rather than at particle state. I'll have another look at it though. I am getting some cleaner better details out of the particle stage

Edited April 16, 2013 by rhussain