Stiky Flip Fluid

[infinity_spiral]

Hello,

I am trying to make flip flid to stick to an object but can't really get good results.I tried to use a popsolver by finding points inside volume and then change particle acceleration by volume gradient in vop .

Stick on colussion option in the solver doesn't really work well it just slows the fluid but doesn't seems really to stick it to the object.

I found this in the Help :

"You can vary amounts of stickiness to different objects, by building your own sticky field. The Fluid Source can be used to take a point attribute like sticky and make it a varying volume where you want stickiness to occur. Then, in the solver, you can fetch this field and either multiply the velocity by its inverse in the volume velocity stage or scale the particle velocity by it directly."

But I am not sure how exactly to fetch the sticky field inside the solver.Can someone explain how exactly to create this setup?And any idea how we can realy stick flip fluid to an object?

Edited May 22, 2013 by T.I.M.

[bloomendale]

why acceleration? you can have 0 accel. and still have velocity. and even if you redirect accel. to inside of the collision objects you still have gravity at least so resulting vector will be some sort of tangential one.

stick on collision option uses field velocities. and then vel field is projected and only then sampled by particles. so you won't have 100% stuck particles.

impementation is simple - sample sdf collision field where sample is inside some threshold values copy the collision object's vel to the vel field (from collisionvel field to vel field) otherwise don't touch vel field.

i'd make the same with point velocities + use collision index field to sort out objects you don't want your particles to stick to.

[infinity_spiral]

Thank you.Here is what I am getting for now.

I think will be nice if the fluid doesn't slide on the surface but also slowly losing the velocity and sticks to the position even the object is moving. Any tips how to achive that?

stiky_test.hipnc

[bloomendale]

1) i think you shouldn't use pop solver if you don't use something unique to pops. houdini copies geometry data to particle primitive, process it and then copies back - it's time.

and you don't need sdf and vel from sops - you already have collision field (sdf representation of the collision objects), collisionvel field - velocity of the collision objects (set velocity type to Point).

add Geometry VOP DOP (something like vop sop + volume vop sop) and sample collision field for every particle - if you closer to 0 than some threshold then sample collision vel and mix (depending on the distance to 0 levelset of the collisionvel field) or just replace v attrib of that particle. BUT. this won't give you 100% glued particles because there are some other forces, project stage of vel field and some cheats (like moving particles outside of collision objects).. this will change calculated velocities.

2) I'm not using flip fluids in Houdini (yet) and don't know exactly what's under the hood, but right off the top of my head:

add sop solver, sample collision sdf for every particle, if particle closer to collision object then threshold find the closest prim uv (do it only once for every time particle approached object too close), store it and then every time find world pos corresponding to that primuv and construct velocity, pointing from particle P to this position. this will force particle to chase some point on object's surface.

not in front of the houdini so only bunch of text) but 1) is working 100%)

[infinity_spiral]

Thank you!I will try your suggestions .The Geometry VOP sound like a good idea to try but I am not sure how exactly to sample collision sdf inside it (will play and see).If you have any simple example will be cool

[bloomendale]

here is for the 1)

3ak_stiky_test.hipnc

[infinity_spiral]

Thanks a lot ! This is looking great.

[MrBag]

Found this while researching FLIP and noticed that H13 has collision stick options, that look strikingly similar to your solution Bloomendale.  Nice one.

 

If anything I appreciate the example of DOP VOPs and getting at the visualisation data.

[johner]

Found this while researching FLIP and noticed that H13 has collision stick options, that look strikingly similar to your solution Bloomendale.  Nice one.

 

 

In H13 the FLIP sticky controls are provided by the new Stick on Collision DOP, which is a small asset that handles volumetric sticking.  It can also be plugged into the Pyro solver, and has an OpenCL-enabled code path.