Hi guys,

I'm needing to fill a logo with liquid and choose particle fluids as they seem to give me the nicest splashes. As it is a logo it needs to be represented possibly accurate.

Luckily on this one I should get away with a fairly low-res particle count. I am having two problems with this:

1. if I use a fairly low particle separation of e.g. 0.25 the particle fluid seems to be blocked from falling down through my e.g. “S” (see attachment) as if the staticRBD collision wasn't set up right. However looking at it (by displaying) it shows the collision geo just fine. Also if I use a higher particle res e.g. particle separation of 0.1 the flow is just fine.

Can anybody tell me what that is and how to fix it? Or do I just have to use the higher particle res?


2. as shown in attached image, the particles don't seem to be colliding right on the border of the collision geometry rather then having something like a buffer zone (marked bright green in a few areas in the screenshot) I have looked into various options to fix this and came up with:

- the VOLUME OFFSET under the collisions TAB of the particleFluid-DOP, but I cannot seem to b able to set a negative value in order to get it closer to the actual RBD geo borders

- the OFFSET SURFACE setting in the collision->volume tab of the static RBD I am using, but looking at the font and the fact that then the letters would cut into each other doesn't make me think that's the way to go either

As a last back up I guess I could have a relatively wide (e.g. high Point Radius Scale or Surface Distance in the particle fluid surface SOP) mesh and then subtract over-standing geo with a cookie node and the original font. However this does seem to be more of a “Maya-solution” rather then the logical work flow I am used to from Houdini.


Is there a “reduce collision distance” - slider somewhere I haven't found? Or what other ways can be used to do this?


cheers for ur help
rodpacker

Hi,

The collision code for particle fluids looks at the “pscale” point attribute.
However, the amount of separation between the particle centers and the geometry seems to be double the amount that is required here.
This is logged as bug 37138.
Until this is fixed, the simplest workaround is to reduce the particle separation.

Hi Michiel,

thanks a lot for having a look into this for me! Highly appreciated.


I will give the particle increase a go over the weekend and let the 8Core churn through it while I'm gone to see how much closer this can get me

thanx again for your help and it would be very interesting to know if there is any bugfixes

cheers
rodpacker

A fix was included in Houdini 10.0.410.
This fix reduces the buffer zone between the particles and the geometry,
so you shouldn't have to decrease the particle separation.

Michiel

Hi Michiel,

very very impressed!!! I'll update today and test it.

cheers
rodpacker

Hi Michiel,
I ran a bunch of tests with 10.0.410 on MacOSX and it doesn't seem like much (if not anything) has changed for my file?

I guess the only option I have now is lowering the particle separation?

Thanks a lot for your help
cheers
rodpacker

I suspect that the coarseness of your SDFs may also be contributing to the buffer layer between your particles and your geometry.
There are two other things you can try to reduce this buffer zone:
1. increase the number of divisions in your SDFs even further.
2. collide directly with geometry rather than SDFs
To try the second option, you will need to disable “Use Volume Based Collision Detection” in the Collisions tab of your Static Object or RBD Object.

Michiel

Hi Michiel,

thanks for your helpful tips. I am currently simming a version whithout using SDFs.

However, I did do another test overnight using the exact same file in H9.5.379 and H10.0.410 and it looks like H9.5 gives me a way more “accurate” particle interaction as shown in the two images below.

I guess I'll be using 9.5 for these sims again

cheers
rodpacker

Hi. Thank you very much for posting your images.

When you use spheres to visualize your particle fluid in 10.0, the particle fluid spheres should be touching the surface exactly.
This is because the particle centers are kept at a
distance PSCALE/2 from the surrounding geometry.
(Please let me know if this is not the case for your file).

It seems that it would be good to have a slider that allows you to control
the buffer distance relative to PSCALE. RFE 37215 has been added for this.

In the meantime, I've disabled the buffer distance for particle-to-SDF collisions in DOPs. In tomorrow's build of H10.0 (build 416), the buffer zone will be removed for SDFs. So you will be able to continue your work in H10.0.

Michiel

Hi Michiel,

thanks for your reply.

Yes, having looked at the different set-ups I thought that that's what has been changed. It totally makes sense and setting things up a bit quicker for most cases as well.

However especially for filling up logos with pointy corners it caused some issues for me (in combination of needing to do lowish-res particle sims that is).


I highly agree a slider that enables users to decide themselves is a VERY good idea.


I will enable the spheres as guides and post another screen shot of that later on today.

(BTW using non SDF's as collision objects didn't work for my setup in H10 the way I needed either.)

Again, I highly appreciate your efforts and fast response and will be downloading H10.0.416 as soon as it's online.

cheers
rodpacker



(sorry my boss was still logged on to the forums on my machine hence the repost under my name now)

THX! 10.0.416 works perfect for me again.

I hadn't had time to get that spheres screenshot alst night - will get it up today.

cheers
rodpacker

Hi Michiel,

am happy as with H10.0.416. Just since I promised to post a screenshot with sphere guides turned on, see below.

As I mentioned before, especially in pointy corners (e.g. left bottom bit of the A) I didn't get any filling.

So yeah that slider u mentioned would be a great improvement to have both worlds.

cheers for ur help again
rodpacker