Jump to content

Efficient FLIP workflows (mechanical animation)


brasco

Recommended Posts

Hi All,

 

I'm fairly new to H, been using it on and off for the last 4 months, I'm almost overwhelmed by the versatility and potential!

There's a few questions I have about FLIP workflows related to a project I'm working on.

 

The task is to simulate a valve in different states allowing different flows and fluids to mix or be kept separate and I'm finding the required resolution and substeps to be really high to achieve a nice looking fluid and also avoid leaks.

 

Here's some screenshots of the product to give you an idea of what I'm dealing with, they're concept fluids (ie. the fluids are just some static meshes I threw in for look dev):

 

Xpeyw0h.png?1

 

pXWlxOV.png?1

 

 

The top image shows the valve in a closed state, so it's the easiest one to test out, but eventually I'll have liquids all over the place whilst the valve opens and closes.

 

My current collider workflow is import Alembic > VDB > Volume collision in the DOP. That seems to work well, but can be slow in itself.

 

So I have two main questions:

 

1. Is there a way to have a lower res sim drive a high res sim so the times overall are kept much tighter? I'm only running on a single hexacore, so I'm seeing crazy long sim times (12Hr+ for 400 frames).

 

2. The chambers need to fill up and I'm finding it hard to get enough particles flowing through to actually reach the top! Here's the crazy geo I've had to create to get this working:

 

vZembuc.png

 

cGHUgMS.png

 

The "funnels" are filled by FLIP fluid emitters (dark blue cylinders) and that drops into the pipe and runs across the valve, it then pools at the blocked off pipe and it fills up.

Is this the only way to get this to work or am I missing something really obvious that would make everything way more efficient?

 

(Sorry for the mega long first post... :))

 

 

cheers

brasc

Link to comment
Share on other sites

if you need a smooth clean aspect , i would also try to use simple pop solver that i would then post process in sop to give them a nice liquid aspect.

if you have thin object i would use basic geo collider with low poly count and clean normal.

 

FLIP + VDB collision is overkill imo for this scenario, where you want (i guess) a clean aspect

  • Like 1
Link to comment
Share on other sites

Using the default source method you won't get any kind of pressurized behavior, which I think is what your looking for.  Source FLIP basically fills the source area with fluid (surface) and particles, but if that area is already full of fluid you get no additional quantity being added to the simulation.  Try adding some divergence to your sourcing.

  • Like 1
Link to comment
Share on other sites

Thanks guys, I very much appreciate the input.

 

 

if you need a smooth clean aspect , i would also try to use simple pop solver that i would then post process in sop to give them a nice liquid aspect.

if you have thin object i would use basic geo collider with low poly count and clean normal.

 

FLIP + VDB collision is overkill imo for this scenario, where you want (i guess) a clean aspect

 

 

So on your advice, I went off and explored using a POP solver to get a liquid like stream, seems pretty promising, although it seems tricky to maintain the volume in the flow. Is there a good way to have particle to particle collisions without jumping to the FLIP solver?

 

You may be right that it's overkill to be using FLIP/VDB but I found it was the only combo I could get to without leaks and exploding fluids.

 

 

Using the default source method you won't get any kind of pressurized behavior, which I think is what your looking for.  Source FLIP basically fills the source area with fluid (surface) and particles, but if that area is already full of fluid you get no additional quantity being added to the simulation.  Try adding some divergence to your sourcing.

 

That's definitely one of the main issues I was experience, it simply stops filling the volume, thanks for the tip on divergence that's exactly the sort of titbits I need to start exploring.

 

 

Thanks again

brasc

 
Link to comment
Share on other sites

Thanks guys, I very much appreciate the input.

 

 
 

 

So on your advice, I went off and explored using a POP solver to get a liquid like stream, seems pretty promising, although it seems tricky to maintain the volume in the flow. Is there a good way to have particle to particle collisions without jumping to the FLIP solver?

 

You may be right that it's overkill to be using FLIP/VDB but I found it was the only combo I could get to without leaks and exploding fluids.

 

 

 

That's definitely one of the main issues I was experience, it simply stops filling the volume, thanks for the tip on divergence that's exactly the sort of titbits I need to start exploring.

 

 

Thanks again

brasc

 

Well jonathan i'm not sure it was a good advice after all. In my old Maya workflow i use nParticles in collision mode to generate those kind of effects where you want an object

fill with a smooth water surface.

 

But as you rightly mention you need to be able to keep cohesion in volume beetween each particule , to make it work.

 

I've try to play with POP but i haven't find a simple straightforward way to make.

- particle collide together according to a radiusPP

- by keeping cohesion beetween each other after collision

 

I try the pop interact stuff which looks to be the easy way to fake collision in POP , but my result sucks.

the new grain stuff looks more interesting in this regards , but it add spring connection and we don't need that.

 

the closest worflow to nparticle in Houdini would be to use bullet with sphere object.

 

but at the end it looks that FLIP could be the most straightforward way.

if you post a scene it would be easier to help you.

 

sticking with flip with a small viscosity to keep the particle together + Pavel Tricks could be a better idea

http://forums.odforce.net/topic/18111-flip-smorganicsheeter-effect/page-4

Edited by sebkaine
  • Like 1
Link to comment
Share on other sites

I'd second the divergence thing.  Default emission in FLIP is basically binary - "if this area isn't fluid already, make it be fluid".  With a source-and-divergence emitter in the center of a pipe, you should be able to continuously force fluid into the system without needing the big gravity-feed tanks at the top... that would save you quite a bit of sim time.  You may need to play with the Reseeding parameters on the FLIP solver... expansion in the fluid can cause it to become under-resolved if you don't have it adding particles back in.

 

On sim times though... I'd say expect to see times like 12h+.  FLIP sims are heavy beasts, and for a sim like this you're going to need a fairly high level of detail to get the precision it requires.  If clients/producers want to see daily updates... well, I find I sometimes need to explain it just doesn't happen working with water sims.

  • Like 1
Link to comment
Share on other sites

i have done a very quick test to show you what i have in mind.

https://www.dropbox.com/s/vowa8ie1acgfgm6/nParticle.mov?dl=0

 

it's Maya nParticle, basically you have

- collisions beetween particles

- a simple gravity

- NO SPH properties

 

the setup time is around 30 seconds.

the simulation time around 5 seconds

 

i bet that if i export this to houdini

- clean the sim

- mesh it with vdb

- add a flow noise displacement in SOP

 

i have something very close to what you expect.

i'm sorry i can't replicate this yet in houdini , but my H skills are still limitated ... :)

 

but if you have also Maya , this could be a B-Plan for you ...

 

Cheers

 

E

Edited by sebkaine
  • Like 1
Link to comment
Share on other sites

I'm actually quite pleased to see C4D being taken more seriously as a VFX tool these days and not just for the motion graphics work that has traditionally been its domain, as we can see with the introduction of things like Arnold for C4D, Krakatoa C4D, and of course Houdini Engine. Anything to help break Autodesk's stranglehold on the VFX industry is a-ok by me. 

Link to comment
Share on other sites

Well here is a test of liquid sim by using bullet solver, it kinda mimic nParticle.

With some:

- extra tweaking + clean up

- liquid refractive shader

- displacement in SHOP

- 3d motion blur

i think it could be a useful technics for smooth liquids. well maybe you can give it a try and tell me how it works for your case !

 

this could also be use as prepass to drive FLIP sim if you need more detail.

 

Cheers

 

E

bullet_liquid.hipnc

Edited by sebkaine
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...