Hi,

I am having an issue with a mesh whereby it seems to pop / bubble as it's settling.

I've played around with the min/max substeps, particle radius scale, surface oversampling and smoothing but none seem to have any effect on the popping.

Any advice would be greatly appreciated.

MP4 preview and HIP project file attached.

Thanks in advance,
Gareth

Hey,

I'm quite new to Houdini and FLIP, so take everything I say with a (huge) grain of salt...

I see you're using voxel size = particle separation everywhere. Inside FLIP voxel size = particle separation * grid scale, 0.005 x 2 = 0.01 in your case. Now it only appears in the Flip Source and the VDB from particles, both of which should not matter, but maybe I'm missing something. Is the popping happening in your sim as well as after creating the fluid surface?

Have you tried turning reseeding off? The community seems very torn on of its usefulness, some say it's a must, some say it causes weird behaviour.

Also have you tried particle separation?

I'm probably just grasping at straws, but I'd start with checking if these effect the popping.

Hi Madorjan,

Thank you so much for taking the time to respond.

I'm fairly new to Houdini also so any help / advice is greatly appreciated.

Are you saying the grid size should be set to 1 instead of 2 so that the voxel size remains at 0.005?

I'm just trying out another sim with reseeding turned off. I have also seen lots of arguments for and against this.

I did try decreasing the particle separation to 0.001 but Houdini kept crashing mid sim. I will try 0.004 next.

Thanks
Gareth

My understanding of the grid size matter is the following: FLIP works with both a particle system and a volume that keeps the 'pressure' of the fluid. These two together are calculated to give you the final result. If you decrease the grid size to 1, you make that volume 8x bigger than it already is. Now your PC might be able to handle it, but the simulation time will probably increase significantly, since FLIP will have to calculate 8x as much 'pressure' data. The more accepted solution (at least from the tutorials I saw) is that you make your voxel size = particle sep * grid scale. I'm not sure about the dimensions of your simulation, but a voxel size of .01 still makes a million voxels per cubic meter (or Houdini unit or whatever). So if your scene is big, I'd go with that rather than decreasing the grid scale. (Also it is set up like that by default, and as a beginner I'm prone to lean on default settings, cause I have no idea what I might break changing them.)

Regarding particle separation, I meant the option in the flip solver -> particles -> separation tab. There is an option there that tries to keep particles a certain distance away. From https://www.sidefx.com/docs/houdini/nodes/dop/flipsolver [www.sidefx.com] : "Despite the velocity projection stage, particles can end up closer together than their pscale attribute. When this happens, internal forces can’t separate the particles because the velocity projection will remove those forces. This results in the fluid compressing over time." I wouldn't think the bubbles are caused by collapsing, but may worth a shot anyway...

Be careful of lowering your particle separation option in the flip object though, 0.001 will give you 125x the number of particles as 0.005, so it will take up 125x RAM and computational power, that might just crash Houdini. You can always check your particle numbers by middle clicking on your output. Depending on your machine, 1-10 mil are good targets, then you can go to a higher number for your final sim. Just be wary that FLIP gives you very different results on higher particle separation, as I learned recently.