hello forum,
Lets say you have an emitter few units over the ground, which is emitting particles with gravity force and collide with the ground, the result will be, all the particles will be land in the same level on the ground but how can make them create a pile of particles, like a sand?? Probably they need to be aware one with each other and each one of them have attributes like friction and mass, is this using “property” pop node? and how?

here is an example:
http://vimeo.com/2582144 [vimeo.com]

Also, when I am applying geometry to each of these particles using the copy sop (like small rocks), how can you make simulation be aware for the volume of the assigned geometry for collisions between the ground and between them?

any help discussion on that or hip files would be really appreciated.

cheers
Christos

does not seem there much interest on that subject.
Anyone? Any help or ideas?

8 hours isn't really all that much time to wait for a response on the forums…
anyway - what you're looking for is a sand solver, Houdini had one I think but removed it…there is an improved one coming IIRC…

as for the idea of copying geometry to particles…have a look at the interact POP

arctor thanks for your respond. I know 8 hours is not that much, but I have noticed past question about sand have been without respond for years.
I just wanted to make it clear that I am not looking for direct solutions but for a discussion with ideas, I forgot to give that in my initial post

hat you're looking for is a sand solver, Houdini had one I think but removed it

About the sand solver, I know used to be one in Houdini and has been removed, although has not been removed from the Houdini 11 documentation.

there is an improved one coming IIRC

by coming, do you mean in Houdini 11? How do you know about that? That will be great.

I am preparing sand and fluid solver too, but via GPU(OpenCL), but I will release in January-February, so you must wait , but Its a beast :shock:

cybermax:

I am preparing sand and fluid solver too, but via GPU(OpenCL), but I will release in January-February, so you must wait , but Its a beast

I am gonna wait , I hope not forget to post your results here..

arctor:

as for the idea of copying geometry to particles…have a look at the interact POP

I went through all options in interact POP, none of them give me what I want. It seems this node assist you to create an area (copying geometry or any radius) for each particle in order to avoid each other. But this works in some level because particles stuck on next to other but never on top. O change the ground in order to assist them to create a pile or particles but no luck.

Sand is a very difficult problem and continues to remain so. The vimeo reference posted above is not indicative of, well, anything really.

Sand represents the classic stacking along with the viscous shear problem. For efficiency you need to sleep particles and then have them wake up again meaning that sleeping particles still need to be evaluated based on some logic.

And then there is the physical sand behavior. Many sand simulations tend to have too much of a fluid behavior and is difficult to get that granular flow. Is the sand wet? Dry? Dry for the first 6 inches then wet inside? Oh it gets complicated real fast.

All aggregates have an angle of repose where given enough piled up sand and a given angle to the ground, invariably a big chunk of sand will break away and slide down. I have not seen anything close to physically simulating that yet. I know all about angle of repose. As a kid I used to work for a mason and the gag was to try to bury your buddy under a sand slide as you were digging in to the sand pile to make mortar. Lots of fun.

If you need to do lots of flowing sand as in a real shot, you can simulate sand flow with a smoke DOP solver and then advect particles through the generated velocity field. I believe there are a couple recent Houdini sand tests done this way on Vimeo/Youtube?
I believe the sand done for Prince Of Persia used this approach along with Bullet Solves to generate lots of collision SDF's for the sand to interact with.

To do a pouring accumulating sand effect you can do the time honored animated collision surface moving upwards to control the particles as they stack upwards but is a hack and remains somewhat effective but re-awaking sand is difficult. I can see using sdf's as an improvement over this approach.

Instead of punting around waiting for something that may or may not get you there, get off yer duff and start hacking away and get creative! Pick a real shot with a real purpose and get going. Chances are your “real” shot will require a bunch of different approaches to pull it of.

Jeff`s right, it's complicated!

few more information, which I used for my solver: http://graphics.cs.uiuc.edu/~wnbell/publications/2005-05-SCA-Granular/BeYiMu2005.pdf [graphics.cs.uiuc.edu]

according my few early tests the best shape(quality/speed) for granule is tetrahedron from spheres like in this video http://vimeo.com/12979964 [vimeo.com] , but GPU allows me to calculate a lot more particles, so its bigger and It looks more real.

For wet and dry behavior, It calculate forces between neighbour granulate according friction, damping, shear and attraction parametr, which can be different for every particles.


I think that at this moment you can try create tetrahedron from spheres and simulate in ODE solver. Its a good start. Using POP network is suicide for this!

thanks jeff for sharing your knowledge, I agree with you and with cybermax that it is a complicated subject the simulation of sand.

I would like to take it step by step, for example I would to achive first something like the video I posted or cybermax's posted video:http://vimeo.com/12979964 [vimeo.com]
and keep wet sand effect out for now, which makes thing more complicated.

cybermax said:

I think that at this moment you can try create tetrahedron from spheres and simulate in ODE solver. Its a good start. Using POP network is suicide for this

I dont know much about ODE solver to be honest, so I am gonna make a tour through documentations on internet and I will be back with questions. I thing ODE solver is an option in contrast to RBD Houdini's native solver, right? Sow can I use particles in the DOP network? I think the video you posted is particle-based simulation.

I thought I recall hearing something about the Flip fluid Solver being similar to the old sand solver. But don't quote me on this.