Flip leaking from geo problem

[dpap]

The  problem is that I cant get the collision geo to have a hole underneath it as the actual geo does. 

 

Check the "Collision Guide Geometry" and you'll see that there in no hole...

 

How can I make a small hole underneath the container?

 

[teratera]

I think you have to add thickness to your geo.

[dpap]

I added thickness but still no luck.

 

No leaking from the hole. Plus the fluid penetrates the Collision Geo (although I increased substeps and collision separation). 

 

Not fun, not fun at all....

 

Any suggestions?

 

 

 

 

 

Btw: I subtructed 2 vdbs and converted them to a polygon mesh to get thickness. Is there more straight forward way to add thickness?

Polyextrude definitely doesn't work it in this case... 

Duplicating the geo,scaling the one down and bridging the two holes by hand is not practical at all.

leaking from geo 2.hip

[dpap]

I did try another approach by pluging in the vdb  to flip solver using Source Volume but no success either.

 

Collision visualization seems to be ok...Or am I wrong?

leaking from geo 3.hip

[H2O]

Here is some solution.

 

Hope it's helpfull.

 

cheers,

 

leaking from geo 3_fixed.hip

[Doum]

Here is some solution.

 

Hope it's helpfull.

 

cheers,

 

Nice I didn't knew that flip could collide with bullet ?

[H2O]

I don't know since when it's possible, but yes, I use it very often and it works.

[dpap]

Thank you Paul for your answer and the example file.

 

I would never thought of trying Bullet..

 

Although it seems to work if you emit only for 1 frame (as I did in the attached file) you'll see the fluid collapses on itself so something  is going wrong...

In other words with your setup if I only change the emission setting so that  the source acts like the fluid's initial state (something like the "FLIP fluid from some" shelf button ) the scene doesn't act as expected.

It should slowly get empty and not collapse on it self in a few frames. Right?

I tried different values in Substeps, Grid Scale , Collision Separation but nothing changed...

 

Apart from that it should be nice to have Volume based collision solution since the documentation and everyone here keeps preaching that flip collides work better with volumes 

 

 

 

 

 

 

 

leaking from geo 4.hip

[eetu]

You have a very fundamental misunderstanding, namely the Grid Scale parameter in the Flip Fluid Object. It defines the resolution of the flip volume grid as a multiplier of the particle separation. In your scene you have a Grid Scale of 250, which means that there is practically a single voxel for the whole simulation, which understandably fails.

 

Always, always check the collision field visualization of the Flip Object!

To get a flip to behave properly, my rule of thumb is to have all colliders at least 2 voxels thick -to get there I extruded your geo a bit more, and increased the sim res a bit

 

Also, setting the Particle Radius Scale in the Flip Object to 0.5 caused most of the fluid collapsing on itself.

 

leaking_from_geo_4e.hip

[dpap]

Hi Eetu and thanks for your answer!

 

Sorry for the misunderstanding but the l upload the leaking from geo 4.hip  just to show Paul that things don't work as they should in his suggested approach.

My settings(leaking from geo 3.hip) where totally different and my collision was looking good (..now that sounds like a little child apologizing, haha!!!)

So I plead not guilty for the Grid Scale= 250    (Please don't be offended Paul!)

 

Anyway to be a little more serious, I need some time to have a thorough look at your file and find the differences from mine and then I 'll post back.

[eetu]

Ahh, sorry, I guess I didn't read the thread properly, just grabbed the latest hip - this v3 makes a lot more sense

 

I think your troubles come from the way you construct your vdb, all this unsigned distance field business etc. Also, usually you do need to multiply an sdf by -1 when you bring it into DOPs, as the inside/outside "sense" is the other way round there (sigh)

 

Also, you have a very low res velocity field with a very high res collision field, that doesn't usually help that much in leakage situations like these (I guess it's mainly for getting more detailed particle collisions?)

As said above, try to keep your collisions two velocity voxels thick.

 

Here I changed to a regular sdf, adjusted the resolutions and multiply the sdf by -1

 

leaking_from_geo_3e.hip

[H2O]

sorry guys for this missunderstanding.

I wanted to do 25 and not 250, my bad (no offense dpap  )

 

But that's not change anything. It was not a good solution for this effect!

 

cheers

[Druidl]

Thanks eetu! You saved my ass! Once more!

I owe you a beer for sure or maybe a beer pack! In Croatia we call it "GAJBA"