Can't wrap my head around flip surface tension

[Gianni333]

I'm trying to understand the relation of the flip parameters to the real world and from my understanding density is kg per cubic meter of the liquid and viscosity is a real world value that can be looked up, which is always nice to get a good base for realistic effects.

What I just don't understand is how the number in surface tension is related to any real world value.

I'm pretty sure it does, because most things in houdini seem to be based on real world values.

 

I thought maybe somebody in the forum can shed some light on this?

[Atom]

I found this tutorial quite helpful at understanding  surface tension and the benefit of using higher substeps

https://www.sidefx.com/tutorials/efficient-techniques-for-realistic-small-scale-tendrils-droplets-and-sheets-in-houdini/

 

[Gianni333]

Thanks, I have seen this before and now rewatched it again.

It is really helpful, but what I'm trying to understand is the relation between houdinis surface tension value and real world surface tension.

Right now it feels like just randomly picking a value and make it work by trial and error.

[paranoidx]

From what I understand: If you dive into the FLIP solver's Surface Tension feature, you'll find that the key parameter is the scale of curvature.

The solver is based on the Navier-Stokes equations. It uses divergence to measure whether particle flow is shrinking or expanding, and then applies pressure to push or pull the fluid, ensuring incompressibility. With surface tension enabled, the solver applies surface pressure based on curvature. This pressure is summed and submitted during the pressure update step (FLIP solve), which then influences particle motion.

In FLIP, the surface tension force is defined as:

Code

F = scale_ST × mean_curvature × surface_normal

Where:
F is the force density from surface tension

FLIP Surface Tension parameter: scale_ST is the surface tension coefficient, typically based on particle separation or voxel size

mean_curvature represents how much the surface is expanding, greater expansion leads to higher curvature magnitude

surface_normal indicates the direction of the shrinking flow

For more background, you might want to check out the paper “Continuum Surface Force Method.” It's important to note that the standard Navier–Stokes equations do not account for surface tension directly, they only solve for inertia, pressure, and viscosity. The Continuum Surface Force model is a technique used in computer graphics (CG) to inject surface tension behavior into fluid simulations. It’s designed for visual realism in animation, not for physically accurate modeling of real-world water materials.

The blur radius feature acts as a smoothing filter to reduce volume noise and jitter. It uses a kernel radius to apply the blur. Increasing the blur radius reduces precision, which in turn diminishes fine details such as tendrils and thin films.

The higher the scale ST, the stronger the surface tension pushes inward. Small droplets quickly transform into bulk spheres, and tendrils disappear faster. Reducing the scale ST makes films and tendrils harder to hold, causing them to scatter more quickly into floating dots.

Maybe someone else could help make it easier to understand from an art direction or VFX perspective, rather than just explaining the dry technical details of the solver. 

Edited 16 hours ago by paranoidx