Foam goes through the geometry

[nicolas_fuminier]

Hello there,

I'm trying to push a little my basic understanding of Phoenix.
Currently working on a cocktail simulation.
I have a basic issue but I can't find the answer.
The foam goes through the glass geometry. Not that much but werdly no issue at all with the liquid particles.
Glass geo is perfectly closed without normal issue.
Increasing the substeb doesn't seems to change the problem (i'm currently at 24 substeps)
Changing the Phoenix geometry property to a precise collider type doesn't seems to change the problem
I know I'm missing the solution but I don't know which menu to look for.

Phoenix 5.10.00

Thanks for your time,
Best,
Nicolas

[slavina.nikolova]

Hey,

Could you download and try the latest Phoenix Nightly - https://nightlies.chaos.com/main#/ph...ghtly/20230525. We've made some changes and fixes in the latest nightly builds that may be related to your issue.


Thanks!

[georgi.zhekov]

Hey Nicolas,

Besides the new nightly, how about if you increase the grid resolution a bit? Currently your glass is something like 2-3 voxels thick, which might not be enough to get a correct result. How about if you increase the grid resolution 2 or 3 times more.

Cheers!

[nicolas_fuminier]

Hello,

Thanks both of you for the fast and accurate answer.
I've just downladed the latest nightly build and will test it during the afternoon.

georgi.zhekov thanks for the tips. Is there any "basic" rule to better understand the links between the grid resolution and the mesh polygons resolution. As far as I understand your tips, the voxel grid resolution should be 8 to 10 time smaller than the polygons resolution. Do the math work both ways ? If I could lower the mesh resolution by a factor of 2 to 3, my actual voxel grid would be enought ?
I'm asking because the issue with your solution is more computing by quite a lot since increasing the grid resolution by a factor of 2 to 3 is adding x2³ to x3³ particles to the simulation time and the simulation output file weight. Not that I can't do that, but it's more easy to understand the need behind this increase to be able to choose well according to the constraints of the project

Thanks a lot again for your time and knowledge,
Best,
Nicolas

[nicolas_fuminier]

Hello again,

New test with the lastest nightly build and an voxel grid small than before (from 0.05cm to 0.035cm grid).
It look like my glass polygons are about 0.1cm for the smaller to 0.3cm for the bigger.

The result is obviously better than the previous version but there is still some foam escaping the geo (no other parameters have been changed)
Sadly I can't really afford increasing the grid since I'll probably reach 3 Millions of particles in total at the end in the simulation for a total of ~70Mo/frame. I know this is only time and storage space but in bigger production / longer movie / faster expected result, everything count.

The look and the animation of the result of the simulation totaly suits me (even with the first grid resolution).
I know I can cut the foam geometry for the render to remove the bad particles, but in some case cutting them is impossible (due to animation setup or 3D layout of the simulation)
That's why I'm trying to understand all the solution there.

Thanks agin for you time,
Best,
Nicolas

[georgi.zhekov]

Hey Nicolas,

Usually when you have very thin or small objects - the simulation won't be able to distinguish them if the voxel size is larger than the object. If the voxel size is a few times smaller than the object you should be fine.

Can you send over the scene so that we can take a look? If it's something confidential you can use the support form here (there is a submit a request, top right) - https://support.chaos.com/hc/en-us

Both the grid resolution and the steps per frame should be more than enough in this case.

Thanks!

[Joelaff]

In some cases I have made a duplicate of my glass and thickened the outside of the glass, and used that copy for the collisions. In other words, the collision geometry can be made thicker, but still match the hero glass on the inside where it matters.

Also setting the collision geometry to kill all particle types inside of it in the Phoenix properties can be helpful.