FLIP collision problem

[Muhammad Junaid Baig]

Hi,

I have been trying everything I know to get this working. I'm new to Houdini and specially to FLIP fluids. It is very very simple yet I can't figure it out. I'm trying to drop a simple FLIP liquid in a glass, but FLIP liquid doesn't seem to be colliding with the glass geo. I have tried to model geo in houdini from scratch but no luck. I have tried reversing the normals using reverse SOP, yet no luck. I'd appreciate any help.

Thanks,

[Skybar]

Have you set the glass up as a collision volume, or made an RBD object out of it?

[Muhammad Junaid Baig]

I made it an RBD object from the shelf tool. I also attached a file in my last post, but it didn't show up. I'll try and upload again.

[Muhammad Junaid Baig]

Here you go. Attached is the file to review.

Glass.rar

[rbowden]

The problem right now is that your collision geometry in your RBD object is not solid. If you check the box 'Show Collision Guide Geometry', you will see what I am talking about. Go to the collisions tab - RBD solver - volume to find that. Mess with the uniform divisions and the offset surface parameter to get a solid collision geo out of your glass.

I would personally create an SDF out of the glass and use that as a collision. I modified your file and attached to show you that method.

RBGlass_01.hip

Edited July 12, 2013 by rbowden

[Muhammad Junaid Baig]

Thanks Ryan, that worked like a charm. And I also looked in your file and perhaps that is a better way of doing the collisions with FLIP. However, there is one more problem I can see now since my particles are actually colliding with my geo, they do not last and preserve their volume till the end and they all start dying. What I'm saying is that I'm left with 10 particles at frame 240 while I start with 30,338 particles at frame 1. Is this something to do with collision? Or is it something else where I should look at?

Thanks for the help.

[johner]

Thanks Ryan, that worked like a charm. And I also looked in your file and perhaps that is a better way of doing the collisions with FLIP. However, there is one more problem I can see now since my particles are actually colliding with my geo, they do not last and preserve their volume till the end and they all start dying. What I'm saying is that I'm left with 10 particles at frame 240 while I start with 30,338 particles at frame 1. Is this something to do with collision? Or is it something else where I should look at?

Thanks for the help.

The disappearing particles are due to Reseeding deleting overcrowded particle areas. The problem is your very thin collision object; the Particle collision detection can see it, but it's too thin to be resolved on the FLIP grid. So the particles stay on the inside of the glass due to the collision detection, but they bunch together too much since the fluid solve doesn't see the collision well. When overcrowded, Reseeding deletes them.

I took Ryan's file and PolyExtruded the outside of the glass and things worked much better. Also increased the substeps to 4, as this is pretty fast moving fluid, though it may not be necessary.

RBGlass_01_extrude.hip

[Muhammad Junaid Baig]

That's perfect John, Thanks much! It works pretty good. I'm a Naiad user and this is all very fresh to me, though concepts are generally the same. I appreciate the help.

Best,

[ryew]

You might find Peter Quint's Wine Spill tutorial useful - http://vimeo.com/15490394

In it, he discusses some of the problems and solutions from this thread as well as general FLIP workflow tips

[Skybar]

I haven't looked at all the other hips here, but in the first one you posted:

Turn off Reseeding, and turn off "Use Volume Based Collision Detection" on your Static Object. This will grant you very accurate collisions, and it is quite much faster if you would otherwise need to bump up your Uniform Divisions to insane numbers - and that is painfully slow.

[Muhammad Junaid Baig]

Thanks for the tutorial link Ryew. I appreciate. David, I eventually found out what you suggested in your reply. And I do agree with you that it is much faster and accurate. But I'm still stuck in particle sinking-to-death problem. I will look at Peter Quint's tutorial now and see where am I doing it wrong.

Thanks.

[Muhammad Junaid Baig]

I just completed the tutorial but He is using H11 and I'm working on H12.5 which sets up the FLIP in different way. I'm gonna try with what my current setup. Although, I learned a few very interesting techniques from the tutorial which may prove to be helpful.

Thanks for the help.