Forums SI Users Flip Solver Collision Inside Closed Volume

Flip Solver Collision Inside Closed Volume

   20508   15   1
User Avatar
pauloduarte
Member
9 posts
Joined: Feb. 2014
Offline
 
Hello everyone, my first post here.

I'm Softimage user since version 4, a couple of months ago I've been studying Houdini.

I have this simulation where there is a emitter inside a closed geometry, after the fluid is emitted there is a animation on the collision object and begins to rotate, but at a certain point begins to appear holes where the particles begin to escape. This purple grid in attachment is the viewing of collision object inside flipfluidobject

Can someone help me?

Thank you.
Paulo Duarte

Attachments:
houdini_flip_collision.jpg (245.9 KB)

 Report
User Avatar
pauloduarte
Member
9 posts
Joined: Feb. 2014
Offline
 
58 views and no answer, I thought it was a simple thing, should be at least, put a liquid inside of an animated object, he just need to collide with the object.
 Report
User Avatar
eetu
Member
606 posts
Joined: May 2007
Offline
 
These things are very hard to answer without seeing the scene - there are so many ways in which things can go wrong.. You will maximize your chances of getting help if you attach a minimal .hip which shows the problem you are having.
Edited by - May 2, 2014 13:29:48
 Report
User Avatar
eetu
Member
606 posts
Joined: May 2007
Offline
 
Maybe your collision object is grazing the edge of your simulation domain, and there is no more collision field to keep the particles from escaping?
 Report
User Avatar
pauloduarte
Member
9 posts
Joined: Feb. 2014
Offline
 
Allright, here is the file, go to the frame 170 and that's begin the problem.

Thank you.

Attachments:
PD_pyro_test_a_008.hip (2.3 MB)

 Report
User Avatar
old_school
Staff
2540 posts
Joined: July 2005
Offline
 
The .abc file is missing that is used to construct the collision geometry.

It looks like the construction of the collision volume is failing. Try to work the settings in the Fluid Source SOP that is used to construct the collision volumes.

See if there is a hole or leak in the incoming geometry that is causing that one corner to fail in constructing the volume.
There's at least one school like the old school!
 Report
User Avatar
pauloduarte
Member
9 posts
Joined: Feb. 2014
Offline
 
Sorry Jeff, here is all files.

The collision volume is clipping some times, I don't understand why it happens, collision mesh is fully closed.

Thank you.

Attachments:
PD_Fluid_Test.rar (225.4 KB)

 Report
User Avatar
goldleaf
Staff
4159 posts
Joined: Sept. 2007
Offline
 
Hey Paulo,

I'm not sure why you were getting those problems; here are a few optimizations for getting some collisions inside of a transforming box (the hip file has notes for these too, in case I missed anything here).

- Fluids only use volume collisions, and to get the fluid to stay inside of a shape, you need to turn on Invert Sign
- Using Object transforms is MUCH faster. Deforming geometry means it recalculates the collision sdf every frame.
- There is an example for extracting obj-level transforms from your sop-level transforms, though in your case, you could just copy that animation to the OBJ level.

Not sure why you were having those problems, but hope this helps!

Attachments:
splash_inside_box.mp4 (3.8 MB)
PD_Fluid_revised.hip (1.7 MB)

I'm o.d.d.
 Report
User Avatar
pauloduarte
Member
9 posts
Joined: Feb. 2014
Offline
 
Goldleaf, thank you very much for take a time to view my file.

I understood what you meant, but I don't think you understood well the behavior that I'm trying to do.

In attachment I'm putting two movies that represent better what's going on, at some time the particles start to escape, the collision mesh collapses.

Well, I think I will try changing my collision mesh.

Thank you.
Paulo Duarte

Attachments:
pd_fluid_010_collision.mov (4.2 MB)
pd_fluid_010.mov (2.9 MB)

 Report
User Avatar
goldleaf
Staff
4159 posts
Joined: Sept. 2007
Offline
 
Ah, sorry about that. What I did with the transform, and inverting the sign should still help.

Is this more of what you're looking for?

Attachments:
PD_Fluid_revised_shape.hip (1.7 MB)
splash_inside_shape.mp4 (4.3 MB)

I'm o.d.d.
 Report
User Avatar
pauloduarte
Member
9 posts
Joined: Feb. 2014
Offline
 
WOW, PERFECT.
So why you construct that collision_static, and use “timeshift” for freeze and store information? Then you pass the transform information to the new collider and inside the new collider is other timeshift.

EDIT: Well, now I understood, was because it has different animations.

“Peak” is to translate points and construct like a cage?

Thank you again.

Cheers.
 Report
User Avatar
goldleaf
Staff
4159 posts
Joined: Sept. 2007
Offline
 
Yeah, I peaked to give a little extra room, to make sure no fluid particles started colliding.

The timeshifts were to make sure I only had the first frame. The one inside of the animating collider tube simmed correctly (since it only looked at the first frame), but in the viewport, it was moving twice - at the object level, and inside from the alembic cache.

Glad it's worked out!
I'm o.d.d.
 Report
User Avatar
pauloduarte
Member
9 posts
Joined: Feb. 2014
Offline
 
I would like to thank the help of everyone, here is my first work done with the help of Houdini, did the animation and fluids in Houdini, particle mesh, light and render in Softimage. I bake the simulation in Houdini and export to Alembic.

http://paulo-duarte.com/project/visual-experiment-fluids/ [paulo-duarte.com]

 Report
User Avatar
cgcris
Member
67 posts
Joined: May 2014
Offline
 
great work Paulo!

I've showcased it in http://softimage.tv [softimage.tv]
 Report
User Avatar
pauloduarte
Member
9 posts
Joined: Feb. 2014
Offline
 
Wow, thank you
 Report
User Avatar
fahad64
Member
80 posts
Joined: May 2014
Offline
 
Thanks for the info….I am using deforming object shelf tool instead of static object from rigid body shelf for alembic files. Working all good now..
 Report
  • Quick Links