A couple fluid (liquid) questions

[bandini]

Hi,

I have a couple questions about liquid fluids in Houdini. I am hoping someone with experience can help answer my questions.

First, we have two types of fluids in Houdini, Particle Fluids and Volume Fluids. I have read the help card on "How to choose between liquid simulation types," but I would like to hear from people about real-world situations where one might be more appropriate to use than another. Also, more advantages/disadvantages of one method vs. the other.

Second, I am trying to simulate a specific kind of fluid effect, and I am finding it very hard to get the look that I am after. I am trying to get an effect of an object emerging from a honey-like substance. So the object should come out of the honey, then as it moves up the honey clings to the surface, finally slowly dripping off from a collecting point at the bottom of the object. I have tried both fluid types and tried all kind of viscosity settings. I basically get the object breaking through the fluid surface, but the fluid will not stick to the object. I have used a Field Force node to try to attract the fluid to the surface of my object, but that doesn't really behave quite how I would like, either. Is there a way to achieve this kind of effect in Houdini?

[Nibbler]

i can't answer for the first part of the question, but for the second i think it's better to paint paths on velocity field to move stuff how you want, fields are good for some unpredictable kind of stuff IMO

[mightcouldb1]

For the first part, I think that level-sets are really useful for lots of things, but I'd say that the best time to use them is when you have a contained area. You don't want to have to build a massive fluid container and have to crank up the resolution really high to get good detail. Unless you are have some kind of proprietary fluid simulation tools like R&H has the LSTK and Scanline VFX has flowline which can simulate fluids in the billions of voxel range. Particle fluids are versatile and can be simulated without worrying about where they travel. They are useful for simulations in open areas. There are plenty of ways to use both, and both can give similar results depending on what you are doing.

For the effect with the honey... Peter Claes did a simulation with a guy crawling out of jam in a sandwich. He used SPH I believe. I remember seeing a PDF with a breakdown of his process, or maybe it was something on the SideFX site. Anyway, you'll have to do a bit of hunting but it's out there.

Jason

[bandini]

Thank you both for the replies. From my tests I have found the Voxel fluids just not suitable for my shot. They don't seem to capture the fluid motion or object interaction quite the way I would like. Particle fluids seem to work better, but are very slow. I am trying the same shot in Realflow now, as well. I seem to be having better luck with it there and getting results faster, although I really wish I could keep the whole thing in Houdini.

[pclaes]

Hey there,

since it was mentioned above, here is a link to that jam project:

http://www.peterclaes.be/main.php?category=5

It was basically sph, combined with metaballs and particle fluid surfacing. Some magnet forces to make some particles stick to the surface.

One huge tip I would give you is to do some of the sims separately: such as the points dripping of the surface might need different settings and or scale. Almost as if you would do "an adaptive simulation". First bigger particles with the basic motion, then smaller particles influenced by them. Then normal particles (non-sph) for extra detail where you need it. Give them slightly different "pscale" and merge them all in one huge pointcloud. The meshing will take a long time, so either you cache that to disk, or you do it with the metaball procedural at rendertime (you have a LOD control).

Getting viscosity has mainly to do with "blurring the velocity". Which you can do with a sop-solver and pointclouds if you wanted to. Or with the settings the sph provides. The wedge rop comes in handy to try different quick tests. I had plenty of "exploding" simulations. So there is quite a bit of trial and error involved unfortunately.

If I had to do it again I would actually consider programming it with cuda and modify the balls example. Realflow will probably get you what you want faster and so many people have done similar liquids with it, so there should be plenty of info available.

[bandini]

Hi Peter,

Thank you so much for responding to this. Your jam project really is impressive. I've seen it several times before this.

I've done a couple other projects where I've blended in multiple sims, like you are suggesting, so I am familiar with that workflow. For this particular instance, it's just a bit too much for me. I'm on a pretty tight deadline, and I've seen already that Realflow gets me where I need to be very quickly. I'd love to be able to render it through Houdnini, but at the moment I can't get the RF plugins to work on my Mac. (The DSOs don't load at all. Been talking with Nextlimit tech support this week to see if we can diagnose the problem)

I haven't delved into programming yet in Houdini, but it sounds like an interesting task. Wouldn't even know where to start there, although I do have some C++ skills and program plugins for other software.

About exploding sims, I've run into that before... Been a little frustrating because I'll get a low res sim the way I like it, then when I raise the settings (higher density of particles) and let it run for the night I come back in the morning to see a big explosion. I found out that substeps need to be raised quite a bit in the fluid solver as density of fluid particles increases. Also changing the fluid solver to the most accurate one helps there, too.