Houdini FLIPS collision disappearance issue

[hb4k]

Hi Guys... How are you?

I know there is a lot of information there that try to explain why this issue happens. However, I´ve tried to set up a basic scene to replicate the most common problem out there

The scene attached is not the final or nothing at all... It was at the point of testing when I'm writing this post 

HOUDINI COLLISION FLIP ISSUES

The main idea is this:

Making the inside of the character-filled with liquid (just a little more than half). The liquid MUST be the same at all time.

I had to invert signs of the static object to make it "work"

And here the magic dust came along. THE LIQUID DISAPPEAR

I´ve tried the following attempts with no luck so far:

Reduce the grid size to 1 so the voxels of the flips have the same size of particle separation,

Reduce the Particle scale to 1 to avoid conflicts in collision

Try to multiply the scale by 3 to make it bigger and up the timescale of the solver to match it.

I also try to advect the collision field with a volume source with VDB but it was worst and I discard that option in the minute

 

Thoughts:

I think it has to be with the particle collision mixture between volumes and particles. For some reason (i think particle accumulation) the solver reaps particles that are very close together and in the next frames, it does the same thing over and over until you are left with no particles at all...

I tried to adjust the reseeding and even not reseeding at all and no luck either...

It will be SO COOL to finally leave this point crystal clear to avoid future users to melt their brains out

Cheers to all 

Peace! 

Wolf.rar

Edited August 21, 2019 by hb4k

[AslakKS]

Hi,

I recently struggled with this exact issue.

What I ended up doing was,

1. Extract the transform of the animated character and use the "normalized" character as the collision.

2. Disabled "Reseeding".

3. Setting the "Particle Radius Multiplier" on the flip object quite high (3-4), I needed the character to be completely filled throughout so you might get away with a lower value.

4. Activate "Apply Particle Separation", less particles seemed to work better. So you might want to do a first pass and somehow upres it afterwards.

5. Apply inverted extracted transform.

Even with this I was experiencing volume-loss at random times, would love to get a good solution for this though.