me too.. its basically just a shelf tool with some imported collision geometry.. youll pick that up in about 20 minutes. although im having trouble increasing the res as you can see.

hm, actually they now shoot out of the sand vertically at extremely high velocity.

edit- i dont think its related to density/mass. ive tried density levels from 1 up to 500,000 and the stones still shoot up out of the grain sim on the first simmed frame.

ive also increased the outer bandwith of the vdb that removes the particles in the volume of the stones, so there is a larger hole around them, but that doesnt seem to help either.

im having difficulty scattering rigid bodies in my sand sim.

ive got them working without the grains,

and in the grain sim, ive converted them to vdb and used that to select grains and delete them, to create holes in the sand where the stones start.

so far so good, but when i sim they sink immediately to the bottom of the sand.

im not sure what density i should give the stones so they have “neutral buoyancy” in the sand?

here is a quicktime of the typical result.. this is with 1500 constraint iterations and 2 substeps.

i have a sand sim on a plane with a central indentation for the area where my collision object will rise up from below, so there is much more depth there.

when i crank the detail (0.003 is the lowest (highest) ive gone. ) , the central area completely collapses down to about 1/20th its original height.

there are no particles falling through the ground object.

ive tried with iterations up to 5000, and dopnet substeps up to 10, with those settings its taking about half an hour per frame, and it doesnt seem to improve. this is a sim with about 8 million particles.

i cannot believe for a minute the pile is anywhere near 5000 particles deep even in the centre.

ive tried disabling mass shock, ive tried disabling constraint averaging. ive played with scale kinetic.. nothing seems to help.

attached is the file and source geometry. its set at lower settings than the maximum i tried, but still quite slow to sim, and really collaps-ey.

any help most appreciated.

i have a uv mapped pointcloud, and im exporting as alembic and trying to load it into vray/Max using the vrayproxy loader.

it does not appear the point uv mapping is recognised, and i read that to get mapping out in alembic you have to use attribpromote to convert the point uv data to vertex data. this was referring to geometry however.
mine is a particle cloud, so this solution doesnt work.

is it possible?

im trying to do a grain sim and i wish to get s much stronger clumping effect (much larger chunks, less breakable.)

ive used a procedural noise to define chunks in a group, and am applying the attractionweight attribute only to those points.

i get the effect im after by cranking the weight up to 10 or so, but obviously this causes quite bad instability since the weight should be 1 or below. - i get chunks jumping around when they fall.

that leaves me with stiffness. does not seem to be exactly the right parameter to play with, but anyway, cranking it up to a high value ( i think, no idea what is a sensible range) of 500,

causes the grains to collapse when hit by a collision object ( i can see them sinking down in the area where the collision object is rising from below. -obviously they should not do this)


ive searched around but i cannot see any detailed information about controlling clumping beyond the basic “wet sand” effect.

so

a) any way to get a high weight value stable?
b) any other method to get larger clumps?


- ive played around with explicit constraints, but they are slow and use loads of ram, i never got far enough to get a decent result.

ive got a deforming alembic object that i wish to collide with a grain sim.

ive tried both the “ray-intersect” based “static object” (with deforming mesh enabled) and vdb based “deforming object” methods.

the vdb based option is much,much faster, even with the voxel size turned down so the collision is (seemingly) very precise.

so obviously id like to use this method.


however, the results are terrible compared to the ray-intersect based setup.

im colliding with a grain sim, and i get very strange grain behaviour around the collision object when using the vdb method.

it looks ok initially, although my collider (growing from below the sand) doesnt seem to lift as much sand up when using vdb,

but later on in the sim, once the object has grown through the sand layer, and the “stem” passing through the sand is more or less static ( its still moving vertically a bit and twisting slightly, but is a smooth cylinder at this point)

with vdb i get a huge gap around the collider with the particles swirling around it.

note this is with a very precise vdb representation.

with ray-intersect based collisions i do not get this effect.

any way to get sensible grain collisions with VDB of a deforming alembic?

hm, reducing the popsolver substeps to 1 and increasing the dopnet substeps to 10 has bumped my sim time from 10 mins a frame up to 1 hr+ a frame.

if that is an unavoidable price to pay for simming on top of a cache, i think ill have to find another way!

hi, thanks for the reply.. i dont have that checkbox ticked. anyway im now trying saving the cache using an output node in the dopnet instead.

and another question (showing my inexperience)

is a bgeo.sim file the same as a sim file? do i just use the same dop output node ive been using to cache my sim, and set the name to bgeo.sim.sc instead of .sim.sc ?

Thank you very much for your help! this sounds very promising.

i do not need 2 way interaction at all. in fact 1 way would be much more controllable as i can get the first layer looking right then add layers as i go.

i understand the workflow you propose, and the first part is no problem, although the second part has left me a bit confused (im not an expert in this stuff.. never used a wrangle before, amongst a million other things)


any chance of a simple example scene for the setup to add a second layer? ? there are no particles created after first frame, and only particles deleted outside the dopnet, post sim.


many many thanks.

ok so i set my simulation to “cache to disk” and ran it..

i can see its all cached.

i go to my grain_particles net, select my ROP output driver, and hit “render to disk”

instead of exporting the cached simulation, it clears the cache and starts resimulating.

this is extremely annoying!

any help most appreciated. i need to export a cached simulation, as i need to export different particle groups with different rop output drivers. not exactly ideal to have to resimulate each time.

please help?

in my case it would be ideal to split my grain sim by group rather than slice, since i have layers of grains with different properties stacked on top of each other.

is this possible? any tips on how to achieve it?

on a related note, is it possible to simulate a lower layer, then load it as an alembic or sim file, and sim another layer on top, and have it behave well?

it would allow me to bump up my particle count significantly within ram limits i think..

odd one this.

when i reduce my particle separation in the grain solver past a certain level, i get squares of particles missing from my simulation.

it seems this is (hopefully) a viewport issue, as i can see other grains are colliding with -something- in the areas with missing particles.

ive tried changing the instancing limit in the display options (closest i could find to a “particle count limit”

but it does not seem to have any effect.

im not maxing out my ram ( i have a whole 2gb free of my 32 gb!)

i have “cache to disk” enabled, and a low (2.5gb) ram limit for my ram cache. looking at the timeline seems to allow 2 frames to fit in the ram cache.

im still simming, so ive not tried an alembic export yet to see if the particles are actually present. it does make flipbooking rather ineffective though.

its odd because if it was a simple particle count limit, id have seen it before, as ive done “upres” on grain sims before with at least 5x more particles, without viewport issues. also, all particles are shown if i look at the “grain source” ..

this is without any upres, but its the highest ive gone with actual DOP particle count.

hopefully somebody can advise.. ive not had a lot of response to previous posts on this forum. either it is a quiet place, or im asking lots of noob/obscure qeustions that people do not wish to, or cannot help with!


thanks, Robin.

doing a sand sim, and id like to export grouped particles to individual alembic files.

since i have caching disabled due to ram constraints, is there any way to render all the alembics simultaneously as i sim?

so far, ive just been using the “render to disk” button on the ROP output driver, but obviously, that only exports the selected alembic.

thanks!

ive tried both the standard “rigid body” tool and the “deforming geometry” one for my alembic imported,animated collision object.

with rigid body, i get decent collisions, but its so damn slow when i have the volumetric subdivisions turned up high enough.

with the “deforming mesh” collisions, its much much faster, even if i set the vdb grid to be extremely finely divided, so it looks almost identical to the source mesh in the viewport.


however, i find the vdb based collisions with grains to be quite odd looking.

ive tried enabling “calculate angular momentum” but i get odd movement from particles in a fairly large radius around my collider.

any tricks to get vdb collisions to work more like the standard rbd one?

ok im stuck here. trying to scatter some stones in my sand simulation.

i created the stones and imported them as obj. circa 300 stones. using the fractured object shelf tool, i drop them onto my sand.

with the sand disabled, i can get them to react relatively well to the ground plane using either the bullet or rbd solvers, although if i choose concave for the collision shape, they tend to fall through the floor geometry with bullet.

anyway, i wish to use the rbd solver, as i need to use a decent volumetric collision object with my main “collider” in the scene, as polygon based collisions seem to cause sand explosions.

i have no idea wether you can use grain solver, bullet solver and rbd solver in the same dopnet and have them all play along.. i do know i need my ground plane to be on the same rbd solver as my stones or they fall through the floor.

ive not tried adding my main “collider” to the mix yet.. but i do know it only seems to work with the rbd solver.

anyway my main problem is as follows:


when i use the RBD solver, the moment the falling stones touch the grains, they fly off at high speed. they are falling from a very small height.

if i use the bullet solver, i dont get rocket stones, but i do get odd effects with stones refusing to come to rest in the sand.. suddenly moving after a few frames, or rotating randomly, even though the grains are settled and unmoving.

ive tried cranking up the substeps and iterations on the bullet solver and min and max on rbd, and ive tried different collision types.

how to get this to work?

if anyone is interested i can package up a file.

could you elaborate on why this is the case, and possibly some pointers on how to do it? it all seemed so logical and straightforward.

on a related note, i get many, many times faster performance when i turn off “volumetric collisions” (even with a cached sdf, as mentioned that doesnt seem to help)

however without volumetric collisions enabled, the particles explode all over the place, when they are hit by the collision object.