use sourceVolume node as collsion for flip and pyro sim

[tagosaku]

hi, I tried to use sourceVolume node as dynamics collision. So far, I tried the both of FLIP sim and pyro sim and can see a little bit influence, but it's almost not colliding at all.

For instance with FLIP sim, I set:
1 - source volume > initialize > collision.
2 - check binding is correct.
3 - FLIP object > Guides > visualization > collision. I can confirm collision exists in viewpoint.
The result looks almost zero collision. I knew this method is not accurate compared to staticObject method, but I could ignore a little lower precision. In fact, it's absolutely NOT acceptable result.

On the other hands, when using staticObject, it always works perfect. Is this the exact reason that people combine to use staticObject node for collsionSurface and sourceVolume to add collsionVelocity?

In conclusion,

Have anyone succeed to use volumeSource as dynamicCollsion?

What's differences of colliding mechanism between staticObject and sourceVolume?

[Dam]

Hi there! 

Using volumesource for collision should work well for flip and pyro sims, but this needs a correct collisionVel field and more important a close SDF volume,

something the sdf it's generated from a model with open flat polygon and in that case the sdf has some holes and even that the visualization inside the DOP object seems correct, the volume collision will not work as it should. 

For simple objects It's possible to make a ray from an highRes sphere to the object surface and build after that the the SDF, but for complex models it needs some work around, multiple ray projection inside a foorEach loop for instance, but probably there are best way of that.

hope it helps in somehow   

Cheers,

 

 

Edited August 28, 2017 by Dam

[tagosaku]

hi, Peon

Do you mean VDB surface won't work as collision with sourceVolume node? I tried pyro sim and FLIP sim with sourceVolume node using as static collsion, and I clearly see collision shape by FLIP/pyro object > Guides > visualization > collision, but it's not colliding very well. I guess a visualizer shows the exact collision shape but actual collision sim does not happen.

[jamesr]

If using a VDB surface with the Source Volume DOP for collisions, you need to switch the "Source Volume" Parameter from 1 to -1

[tagosaku]

To jamesr,
Yes, I did. That's why I can see correct collsion shape in visualizer. Hopefully someone provides me a collision sample file of either flip or pyro sim...
 

[fuat]

you can also do it with vdbs only wothout using volume source. 
in SOPs, do vdbfrompoly and use surafec as collision. plus create another attrib (in vdbfrompoly) set to "Tuple/Vel/Acceleration" and use point.v as collisionvel.

done.

bring this into DOPs with fluid source DOP. set it to "Collision". when you now check your SOPtoDOP bindings tab, you will see that it need one "collision" field and one "collisionvel" field.

good luck!

fuat

 

[Halltom92]

As fuat said, except when you use a VDB as a collision volume, you need to make sure you set the source volume scale to -1. Then it should work properly.

Tom

[jamesr]

@tagosaku You say it's like it's not colliding at all right? But you can see collision geo when you enable the guide on the FLIP Object? Is your sourcevolume DOP wired into "sourcing"? If so, try switching it to the third input, "Volume Velocity". I'm unable to send a scene atm. 

[fuat]

1 hour ago, Halltom92 said:

As fuat said, except when you use a VDB as a collision volume, you need to make sure you set the source volume scale to -1. Then it should work properly.

Tom

yes, exactly like Thomas is saying. i forgot that (brain is in sleep mode, too late already)
thanks Thomas.

[Dam]

Hi Saku,
What I meant is, considering a case when the collisionvel it's setted correctly on vdbfrompoly and fluidsource in Dop as @fuat says. An other thing I found useful to check is that the vdb it's closed or better.. the mesh where you generated the vdb from doesn't contains holes or that it's not just a flat surface - if it does your vdb will be not closed and good probability that your sdf collision will be screw up - Cheers

Edited August 29, 2017 by Dam

[jamesr]

Hey I know you were hoping for a scene file earlier, so here is one I made for you. 

I hope this helps/is what you're after. 

sourcevolume_collisions_for_tagosaku.hiplc

[tagosaku]

Jamesr,

I had not been checking this thread, sorry. But yes, I downloaded your hip file and understood input slot well.

Thank you so much!!

Edited December 29, 2017 by tagosaku

[jamesr]

Good to know!