Hello, I'm having an odd issue where my viscous FLIP sim works perfectly well with slip-on-collision turned off, but loses all of its volume when slip is turned on:
The way it loses its volume seems strange too - like it is rolling itself into a tighter and tighter ball.
I've tried everything I could dig up on FLIP volume loss in various configurations:
Significantly increase substeps (I tried up to 60)
Lower particle separation
Thicken the collision geometry (no difference and the collider works perfectly with slip off)
Lower grid scale
Raise particle radius scale
Tried with reseeding on and off
Tried higher oversampling values in reseeding
Lower surface extrapolation
Tried different slip scale values
Tried different friction values with "Use Friction and Bounce" turned on and off
Experimented with different Velocity Types and Volume Fraction Methods (I have no idea what either does, but what the heck!)
Tried with Surface Tension on and off
Tried Normal and fast moving extrapolation modes
Prayed to the FLIP gods
Prayed to Jeff Wagner
Laid a sacrifice at the altar of SideFX
I'm trying to learn as much as I can about FLIP through experimentation, so would appreciate any pointers. I'm more on the artist than technical spectrum, but enjoy trying to understand the "Why?" of these kinds of problems.
It's hard to conduct tests when so many defaults have been arbitrarily changed. If you alter a default and see no change in simulation, return it back to its default value before conducting another test. If you are concerned about volume loss, just set grid scale to 1.0.
NOTE: In shots like this where you want the flip to come to rest, RBDs too, it's not uncommon to simply apply a timeshift at the end to make the final frame static.
I reset a lot of values and seem to have established a stable result for both slip on and off.
The settings changes I made were to get the look I wanted for the liquid itself. I unfortunately didn't notice the volume loss issue until later. For testing the volume loss, I was definitely trying one thing at a time.
I'm seeing similar volume loss in your example, but point taken that maybe it is asking too much of FLIP to stabilize and time shifting would be a better answer.
