car skidding through puddle

Discussions about the use of Realflow 3
mrotman
Posts: 24
Joined: Tue Oct 16, 2007 6:51 pm

car skidding through puddle

Postby mrotman » Tue Oct 16, 2007 7:01 pm

hey everyone,

this is my first post, although ive been using RF for a few years. my question pertains mostly to internal and external pressure settings. Shaun Michael suggests having a very low internal pressure for water. While this may work well in a confined glass of water, im not sure it works so well in a widespread puddle. The fluid seems to separate and bead up like mercury, even at very high resolutions (500 at a scene scale of 1.0). Are there any general rules to follow if we are dealing with splashing in shallow ponds or puddles as opposed to a glass of water being poured?


shaun_michael
Posts: 4766
Joined: Sun Sep 10, 2006 8:04 am

car skidding through puddle

Postby shaun_michael » Tue Oct 16, 2007 8:34 pm

Hi,

Thanks for joining in the "discussion".

My feelings about internal pressure are such that a significantly lower internal pressure than external will enable your particles to sit closer together. This works for contained and uncontained fluid. What you may be experiencing is a combination of other factors such as the lack of friction on your surfaces, or overly high surface tension.

One of the downsides of low pressure settings is that the lower they get the more lilke dumb particles the particles behave, and can sometimes lack a bit of "life".

Surface water needs to be treated in a similar way in my experience. You still want your particles to sit closely to one another so that you can mesh with good detail. Puddles generally congregate in depressions on surfaces, but for depressions or just sitting water held together by surface tension you're still going to want friction to stop your particles continuously drifting apart (and probably beading as you suggest it does).

Friction will help back the fluid up so that it doesn't thin-out too much and also rid your sim of that "sliding" appearance - the default friction isn't enough.

Other general pointers would be looking at the roughness and bounce settings to add life into your particle reactions. Don't ignore these settngs.

On a wider basis, I think in your experience with RF you know that there are many ways to skin a cat. A quick look around at people's finished work here shows a plethora of settings that produce similar results, and I think we all have our preferred starting point with a simulation.

What are your thoughts?

Shaun

mrotman
Posts: 24
Joined: Tue Oct 16, 2007 6:51 pm

car skidding through puddle

Postby mrotman » Tue Oct 16, 2007 9:56 pm

thanks shaun, that makes perfect sense. I think i will try increasing friction and bounce. I will also narrow the gap between internal and external pressure settings. There are many ways to skin a cat, but it would be nice to come up with some rules for water behavior at this scale so we dont have to reinvent the wheel every time.


also, you have recommended using very high compression values (8-10k) what effect does this have on the fluid behavior?

thanks
Check back here for updates.

shaun_michael
Posts: 4766
Joined: Sun Sep 10, 2006 8:04 am

car skidding through puddle

Postby shaun_michael » Wed Oct 17, 2007 7:41 am

Hi,

There are many ways to skin a cat, but it would be nice to come up with some rules for water behavior at this scale so we dont have to reinvent the wheel every time

There is a thread currently active where we're discussing the age old RF utopia of having some sort of "preset" that gives you water at a set scale. You can find it here.

Perhaps you'd like to get involved - the more the merrier.

With regards to your question on the compress parameter, the higher the setting the more elasticity is removed from the particle mass as a whole. If you set this to 200, fill a container for instance and allow it to settle without the use of speed daemons etc, you'll notice it bounces up and down for quite some time. Pushing the value of the compress parameter removes some of this "bounce" and seems to exert an overall compression tot eh fluid, not in volume terms, but in behavioural terms ie it wont compress the fluid down causing heat build up and instability, it'll just make the whole mass of particles, behaving as one, less elastic in it's behaviour.

You can find some documentation in the help files.

The 8k-10k value I suggested from my testing is rather arbitrary in that that was the value I found worked fine for me but I didn't then go on the examine what 20k, 30k, 40k etc would have looked like. I did run one with 100k compress value once and noticed little difference in the behaviour of the sim I was doing when compared to 8K. Generally I leave it at 10k but I'm sure you can see where they may be situation when having that elasticity in your fluid mass might be beneficial.

Shaun

mrotman
Posts: 24
Joined: Tue Oct 16, 2007 6:51 pm

car skidding through puddle

Postby mrotman » Thu Oct 18, 2007 10:11 pm

hi

so i got my first full meshed playblast out, and it looks very good, but there are a few minor issues. Mainly it looks like slushy water like melted snow with ice in it, not really like rain water. my settings were


scale 1.0
resolution 1500
int press .1
ext press .5
viscosity 1
density 1000
particles 500k

surface tension 5

the characteristics i disliked were that when the splash landed on the surface, it piled on itself of spreading out into a smooth surface

i am beginning to settle a new starting point

scale 1.0
resolution 2000
int press .05
ext press .5
viscosity .25
density 800
particles 800k

surface tension 1

as you can see, im trying to attack this from the standpoint of the surface tension and viscosity seemed to be way too high. Does anyone know what i'm talking about? And is dropping the viscosity, int pressure, and surface tension, while raising the resolution a good way to correct the problem?

shaun_michael
Posts: 4766
Joined: Sun Sep 10, 2006 8:04 am

car skidding through puddle

Postby shaun_michael » Fri Oct 19, 2007 9:00 am

What's your friction like? Its an incredibly sensitive value, so if you have raised it slightly, as we discussed earlier in this thread, you may find that you've brought it up too high. If I do raise it, I tend to raise it marginally higher than the default and run some low res tests.

How's your gravity? I tend to use 15-20 as a default rather than the 9.8 you get when adding the daemon.

Bringing your density down, and your resolution up is going to make each particle very light, which in turn should make it react better to external force, but this can come at a price where they have simply too much life in them - but then that may be the desired effect. Couple this with such a low viscosity value, you're likely to see some busy particles :)

I would persevere with the surface tension. I tend to leave it at about 8-10 at this scale for "water". Post 44 in this thread shows the results of the sort of settings I've mentioned. In this one I don't think I pushed the friction quite high enough though, but i was relatively happy with the water action (although not with LWs motion blur at that time). maybe it'll give you some food for thought.

Shaun

mrotman
Posts: 24
Joined: Tue Oct 16, 2007 6:51 pm

car skidding through puddle

Postby mrotman » Fri Oct 19, 2007 4:46 pm

in my last test i raised the friction from default of .001 to .005, and i think it was too high, as the fluid didnt want to fill in areas that had been swept away by the tires. So i lowered it back to .002, the one thing i have noticed in my current test is that its taking about 5 times longer to simulate than the previous test. i hope its not too busy as you suggest. the problem before wasnt the livelyness, it was more the weight felt too heavy to be water and the viscosity seemed too muddy, i would assume that lowering the surface tension and friction is the way to get the fluid to spread out more evenly, maybe i did not need to lower the density.

mrotman
Posts: 24
Joined: Tue Oct 16, 2007 6:51 pm

car skidding through puddle

Postby mrotman » Fri Oct 19, 2007 5:07 pm

you were totally right, the particles fly off way too fast now, i tried to split the difference now

density 1000
viscosity .5
surface tension 3

hopefully this will get us there. why do you add extra gravity? isnt it best to stay real world as much as possible?

shaun_michael
Posts: 4766
Joined: Sun Sep 10, 2006 8:04 am

car skidding through puddle

Postby shaun_michael » Fri Oct 19, 2007 6:56 pm

why do you add extra gravity? isnt it best to stay real world as much as possible?

Ideally, yes it would be best to add real world gravity. However, it seems that nearly all sims done with this setting tend to be kinda "slow-mo", so I add extra gravity. It would seem that with all the variables we can change that affect the way a particle behaves and also how it reacts to its neighbours makes enterng a real world constant, like gravity, somewhat redundant as the water isn't necessarily "real-world" in its make-up.

Shaun

mrotman
Posts: 24
Joined: Tue Oct 16, 2007 6:51 pm

car skidding through puddle

Postby mrotman » Mon Oct 29, 2007 6:26 pm

so far so good with controlling the simulation attributes, i just wanted to ask about meshing... does blend factor affect the "thickness" of the final mesh? i read somewhere to try lowering the blend factor to 30-40 for thinner fluids, if this is true, why is this the case?


Return to “Realflow 3”

Who is online

Users browsing this forum: No registered users and 1 guest