Object picking up sticky FLIP fluid

[tinyhawkus]

Hey Dudes,

I'm building an ice cream set-up that has a bunch of cool stuff happening, but I'm stumped on one thing.

 

Here's the example: A simple FLIP tank with really viscous fluid, a sphere moving down into the fluid and then back up

out of it with the fluid sticking to the sphere. Think dipping your finger into honey.

I've looked at some sticky fluid set-ups on odforce that involve a fluid being emitted onto an object but nothing so far is

working on the simple scene I've described. 

 

Help!

 

Lewie

 

[Guest tar]

Here's a post about mud on a tire - some good examples there, see if it helps

 

http://www.sidefx.com/index.php?option=com_forum&Itemid=172&page=viewtopic&t=31205&highlight=sticky+flip

[tinyhawkus]

ARGH! sidefx.com isn't loading here in Melbourne. 

[edward]

sidefx.com has change it's IP address so it might take a bit for your DNS server to update. Some things to try:

- Quit your web browser and restart it. (Firefox is notorious for caching DNS look ups)

- Temporarily use Google's public DNS server which I do know is updated (their primary DNS server is 8.8.8.8)

[tinyhawkus]

Working now. I checked the hip. It works fine for that demo of fluid falling onto a colliding object, but not if the fluid is stationary 

and the object presses into it. Essentially I have a character that has to land on an ice cream and he pushes a bunch of it around.

 

The system is working as follows.

 

I have a coarse VDB of the character that I pipe into a POPVOP and it reduces the velocity of the FLIP particles to zero, until the

characters proximity is within a certain distance. That way the ice cream stays completely still without needing to have insane viscosity

values.

 

I have a full res version of the character interacting with newly unlocked FLIP particles, pushing the ice cream out of the way, and transferring

viscosity values to create localised melting from the character. This is achieved using a multi solver with the FLIP object plugged in first port 

and then a SOP solver and the FLIP solver plugged into the other port. The SOP solver has attribute transfer to pass the different viscosity values.

 

I also have some metaballs for localised hero drips. All good so far.

But I'm stuck on how to create the sticky aspect of it. Where the character will take some of the FLIP particles with it as it pushes away from the ice cream

after having been pushing into it.

Would maybe an attraction POP setup where the character is the attracter be the way to go? Running in the particle velocity port of the FLIP solver?

I'm fairly new to this, so go easy on me.

I've also setup a gradient volume on the points made from the ice cream geometry that the FLIP object is sourcing from. This gives the ice cream a highly

viscous core, but the outer level of the ice cream is less viscous. Just like a real one bought on the beach!

I know I can cheat (and probably will) and attribute transfer on collision to activate points scattered on the character to sell the idea that he has made

contact with the ice cream, but I'd dearly love to be able to have this last piece of the puzzle.

Cheers.

 

Lewie

[JuriBryan]

you can lay down a sopsolver. in there use a sdf of the geometry you want the flip to stick to.
then get a gradient from the geo to the points, then scatter a few points on the geo.
look them up in a point cloud and check the distance, then add that to your flip particles with in the gradient already establish.
now those flip points will follow the geo, and since the flip is sticky surrounding particles will follow and you get the stretchy sticky look.
the whole thing can be done with one object merge, a scatter and a vop sop.
 

Edited May 22, 2014 by JuriBryan

[tinyhawkus]

Okay, I think I'm following what you're saying. To a point. Could you run me through a bit more in depth? I'm pretty new to this, so a little hand holding

is required, but I promise to buy you dinner and take you to a good show!

 

Lewie

[JuriBryan]

okay no problem.
So lets say you have a sphere that is supposed to pick up some honey.
You create the animation on the sphere and the flip tank.
Then make then collide and give the fluid some viscosity, the normal dop stuff happens and the fluid does not stick to the sphere.
Lay down the sop solver and connect it to the particle velocity input.
inside the sop solver:
object merge in the sphere convert it to a vdb and then scatter some points into that vdb.
then lay down a vop sop, connect your sim into the first input, vdb field into the second and points into the third.
inside the vop.
lay down a volume sample from file, sample the vdb, and right after that a fit range.
set the fit range to something like 0-0.1 to 1-0.
that should now give the points with in the radius of 0.1 a gradual value of 1 towards 0.
then put down a point cloud and trough that get the positions of the points you have scattered on to the sphere before.
then subtract that position from the current particle position, multiply that result by the distance from the volumesampling and add it to the position of the particle sim.
now all points with in the radius of the sphere, which you define in the fit, will be at the position of the points on the sphere with in the radius defined in the point look up.
you can also compare the value you have after subtraction the point and the particle distance and check if it is smaller then 0.001 then use the current position rather then the new position you get from adding the two.
that is good if you don't want them to clump so much.
and pipe that out to the position.
now in the sim the points close to you collision object will follow that object extremely close, so lay down a enable solver and only turn the effect on when you actually want it.

And since the points will follow the sphere when the sphere moves and since you have high viscosity values you get the effect you are after.
 

Hope that helps a bit.
I will also try and create a test file over the weekend to show the whole thing in action.

Cheers,

Juri

[tinyhawkus]

Thanks heaps Juri!

 

I'm back at work and just going through your notes now. 

Will post my finding ASAP.

 

If you get time to make a test scene that would be awesome.

 

Cheers.

 

Lewie

Edited May 25, 2014 by tinyhawkus

[tinyhawkus]

Definitely going to need that test scene.  :-(

[JuriBryan]

hey,
sorry was really busy over the weekend...
I will try and get one together for you tonight... In the mean time you could play around with the gasstickoncollision dop (http://www.sidefx.com/docs/houdini13.0/nodes/dop/gasstickoncollision). 

that might get you there as well.

Cheers,
Juri 

[JuriBryan]

okay....
so i tried to put together a test file over lunch and well... it does not work.
sorry about that.
My method cant work for two reasons.
We need to keep track of the points that collided with the collision object.
That is easy enough with a pop group that gets triggered by an attribute and preserves the groups, so that is in reality a none issue.
The main problem that I assume you were facing to start with is that no matter how high the viscosity is the surrounding points to not move with the particles that stick to the collision object.
So there is a different approach needed.

Here is a file that is not perfect... but I think with some tweaking it should work just fine:D
It uses velocities instead of positional data... so that surrounding points move as well.
There is some delay.. but I think with more particles and maybe the sphere/collision object as an emitter it should work with out a problem.

Hope it helps you out a bit.
Cheers,
Juri

Sticky.01.jubr.hip

[tinyhawkus]

Yep, ran into that issue. Thank you so much for taking the time to test and build me a hip.

I'll look at it straight away. If you're ever i Melbourne i'll buy you a beer!

 

Cheers.

 

Lewie

[Daniel Lefebre]

Having the same issue, As far as i can see this is a major struggle with flip. The example solution is in my option not a viable option as it does not behave like a normal sticky surface or like a normal fluid.

results with the object beginning in the fluid and being pulled out are correct and ideal but you can not get those results if you just dip the object in. If i had to dip an object and wanted stick I would have a dummy object that i would initialize in the fluid. I would then have my hero object marry up with the dummy and then pull them both out at the same time essentially mimicking a surface that fluid will actually stick too.

After a few days of playing around with every solution i can find this is the best for a straight up dip effect. maybe one day this will be fixed.