Day 54

I suppose my encounter with pyro fx has paid off even though it's not going to be used for the project in LA.

I dived into the AutoDopNetwork for Tanner's flip fluids file, tweaked a couple of parameters and nodes here and there.

I have a rough idea of how the file works. The smoke solver is actually driving the flip fluid particles to flow nicely. The particles are somewhat there for the details.

Here's my documentation of what I've learnt so far:

AudoDopNetwork

Particle Fluid emitter:
Velocity changes speed emission rate of fluid.
Variance increases randomness of particle motion.

FLIP Solver:
Force Scale increases amount of strength used to drive the particles.
Smoothing controls how much friction is applied to particles. Affects bouncing off primitives.

Gravity:
Y-axis tweak, nice to tweak this along with velocity and force.

flipfluidobject:
Particle seperation affects space between particles.
Physical Tab -
Bounce affects particles bounce sensitivity. Tweaked this along with smooth.

Smoke Solver's Apply Source:
Noise Tab -
Frequency tweak?
Turbulence tweak?
I turned it off as it didn't make much difference in the end.
Emission Amount -
Doesn't affect look much.
Affects rate of emission of smoke but doesn't affect particles much

OBJ/ particlefluid


particlefluidsurface:
Surface tightness controls connectivity between particles in volume state.

Basic Liquid Shader

Achieve looks by tweaking:
Attenuation Density,
Attenuation Colour,
White water &
White water colour.






===================================================

Flood Simulation

Here are test simulations to determine the flow of how the flood is going to pour down the terrain. The simulation of the particles may differ when the emitter is scaled up to match the size of the terrain. Thus I'm prepared to increase the emission rate of particles and re-tweak the settings if needed.

Test Renders 01 to 08 showcase the simulations of the particles not in their fluid volumetric state:

After getting the look I want from the flow of particles, simulating them in their volumetric state didn't go quite well.

Note: For comparison with render timings, total number of frames flipbooked is 50 for all simulations.

testRender_09: Simulation from testrender_08 with volume state.
Simulation Time: 11 min +

testRender_10: Very active noisy look.
Force Scale: 2.5
Smoothing: 0.25
Bounce: 0
Simulation Time: 10 min +

testRender_11: Particles still scattered around but less active.
Force Scale: 2
Smoothing: 0.25
Bounce: 0
Simulation Time: 10 min +

testRender_12: Particles extremely active. Particles scattered, seperated and isolated everywhere.
Force Scale: 2
Smoothing: 0.1
Bounce: 0
Simulation Time: 9 min +

testRender_13: Lack some splash details. Liquid flows somewhat nicely though.
Force Scale: 1.25
Smoothing: 0.5
Bounce: 0
Simulation Time: 10 min +

testRender_14: Particles flow very smooth and nicely, probably not ideal for a violent flooding scene we're working on.
Force Scale: 1.5
Smoothing: 0.5
Bounce: 0
Simulation Time: 10 min +

testRender_15: Particles flow slightly more vigorously, other than that not much change.
Force Scale: 1.5
Smoothing: 0.5
Bounce: 1
Simulation Time: 10 min +

testRender_16: Some particles flying around.
Force Scale: 1.5
Smoothing: 0.35
Bounce: 1
Simulation Time: 10 min +

testRender_17: Looks decent if it's going to be scaled up?
Force Scale: 1.5
Smoothing: 0.4
Bounce: 1
Variance: 5 in x, y & z
Simulation Time: 10 min +


These are what I've observed so far

In simple terms,
Force Scale controls how nice and smooth the flow is.
Smoothing controls how sensitive the particles react to geometries.
Bounce controls how vigorous the flow is.

Possible Outcomes?

Increase emission rate of particles if scaled? (To achieve and maintain fluid details.)
Somehow decrease the size of each particle in it's volumetric state?
May have to bump up tweaked values to maintain detail on scaled fluid?