im considering options for a project in which a car will thrash about in a muddy environment.
can phoenix handle thick claylike mud? lumpy stuff that leaves tyre prints, squeezes out from under the tyres, sticks to stuff, breaks into chunks etc…?
can it deal with fluids with variable viscosities throughout? maybe controlled by a procedural 3d texture? this would be great as i could get lumps of more solid claylike stuff and more runny elements that splash about, all interacting together… i expect this would be better as a number of sims, rather than all in one.
never tried, but phoenix is still grid based simulator and can’t produce the static friction effect , that is needed to simulate mud. the viscous liquid in phoenix always have some residual slow movement, that slowly changes the shape. we are planing to upgrade to grid-particle core and then such simulations would be possible
well to be honest if i can get relatively close it should be fine, any residual movement can be minimised, or i can simply make the whole scene more “wet”
do you have any examples of a really really thick fluid in phoenix that could give me an idea where the limits lie?
or alternatively a rough set of parameters i could play with in the demo?
we have a chocolate sample that have viscous setup. you can use it as starting scene, upgrade to nightly because important bug related to the viscosity was fixed. use the sample from the nightly builds.
i think in your case the surface tension may be switched off, you don’t need smooth edges, but better try both the variants.
the maximal viscosity is 1, but actually the maximal possible value is 1.01. you can try to set viscosity above 1 using the listener. don’t use 1.01 it may cause program hang. keep in mind that the closer to 1.01 you are the slower is the calculation. for example 1.09 will be 10 times slower than 1, 1.099 will be 100 times slower. i suppose artifacts may appear too.
anyway i made some tests based on the chocolate scene (using the latest demo version only) i changed the grid to partially filled, removed the chocolate bar and source, and added a wheel rolling through. i see it would be hard to get an effect close to my needs, even with viscosity maxed out i only get a “thick tomato ketchup” kinda consistency.
one thing that really bugs me though, however slowly i roll the wheel into the grid, it still intersects massively with the fluid. i get a reaction from the fluid, it is partially displaced by the object, but there is a good portion of the fluid still inside the wheel geometry (a simple cylinder, rotating and translating) its like the fluid cannot move fast enough to get out of the way of the wheel. i assumed i could fix this by either slowing the wheel, or increasing the substeps. ive tried slowing the wheel right down, and also had the substeps up as far as 128, doesnt seem to make a blind bit of difference.
i dont think its a simple case of not choosing the right object voxel type either.. the intersection is much bigger than a single sim cell and ive tried all 3 types anyway.
ive just tried increasing the quality from 20 up to 200. the grid is 200 in its longest dimension. this made it slower to calculate but didnt benefit the issue at all. was gonna attach a couple of screengrabs showing the issue but i cant upload in the public forum it seems.
if you hide the wheel geometry you can see that the surface level of the mud inside the wheel is almost at the same level as the original mud surface. there is lots of spray and nice reaction around the wheel, as expected, but inside its full of mud. this would be ok if it wasnt clearly visible around the sides of the wheel.
unfortunately with the official build you have no chance at all, it has a bug related to the simultaneous action of the viscosity and gravity, the bug leads to inability of the viscous liquid to resist gravity, regardless of the viscosity factor.
We can get you a full evaluation version, and then you would be able to use the nightlies. If you are interested, please email us to support@chaosgroup.com and mention this thread.
ok all done, ive got my shiny 30 day evaluation and nightly build access. since im only on vray 2.4, it looks like the latest nightly is 24174? does this build contain the improvements mentioned? looking forward to playing!!
ok first thing to report, the intersection is much less.. still some but ive not tried super high substeps yet.
second is, whereas in the demo, the surface of the (filled) fluid stayed quite static until the wheel arrived, with the new nightly i get a massive “bounce” of the whole surface in the first few frames.. some kind of settling issue i guess, but it causes the liquid to move about a helluva lot, id imagine id need at least 100 frames for it to calm down.
seems in the new nightly build i need to set the z axis to -jammed to stop the fluid falling out of the bottom when the sim starts! (it does come back again later in the sim, the issue is a large area sim with a shallow fill.. the sudden drop in level at the start of the sim causes the whole lot to dissapear out of the bottom of the grid.
unlike the official build in the nightly build the initial fill up does not produce solid bottom automatically, it produces supporting pressure, but the liquid can step back if you push it downward.
ok so i tried your tip of increasing the viscosity above one in the listener, in the official demo this possibly worked, i didnt get many comparison tests out, but with the nightlies, any values above 1 seemed to calculate really fast and have much lower viscosity.
on a related note, static surfaces correction does a good job of reducing movement in static areas, but it would be great if this could be adjustable.. id imagine using the method here it would be possible to approximately simulate much thicker stuff, as you could essentially stop any slow moving fluids from settling? as you said, its so far impossible to do really thick mud. i got a lovely effect by mapping the fluid channel with a cellular map and a gradient for cutoff.. got a lovely blobby field of mud, but of course, even with very high viscosities, it never stays still and after 25-50 frames its settled to a flat surface.
yes, the slow movement is as expected.
i have another suggestion, you can try to simulate mud using the foam. actually it has all the needed properties, use non solid source to create it.
the bad thing is that it is much slower.
ok having a crack at using foam, however im a still a bit of a noob. few questions:
1. if im using a source (in brush mode on a box set to non-solid) do i leave liquids enabled in the simulator or turn them off?
2. when using a source for the foam, what controls the particle density? ive tried upping the discharge of the source but i still get the same sparse collection of particles.
3. im assuming the foam parameters in the simulator control the behaviour of the foam generated by the source? i also assume some settings will be irrelevent since im not generating the foam from a liquid.
4. from what i saw in another post about foam fluids (same discussion really) i cannot mesh these foam particles with phoenix. since splashing out on frost is highly unlikely in my scenario, can i get a reasonable mesh out of these things?
5. does the fact im using particles mean i can lower the grid density or does it still have an effect on the simulation?
6. wha governs the behaviour of particles that leave the grid? and do they get picked up by the sim again if they re-enter the grid?
i have a sneaking suspicion this is gonna be way too slow and fiddly (considering im on a 30 day trial!)
