Dialing In Waterfall FLIP Scale?

[Atom]

Hi All,

I am working on a waterfall scene and I have a basic setup working. I have watched Jeff Wagner's FLIP tutorials parts 1 & 2. Somewhere in there he mentioned that slowing/decreasing gravity gives a larger feel to a FLIP simulation. I have changed gravity to -7.80665. This does have a small effect but the overall look is still too small, especially in the lower basin area.

Does anyone have any tips or VEX velocity recipes for making a FLIP fluid seem larger?

I have added a supply of water in the basin so the falling water does not have to fill the lower area.

Edited January 2, 2017 by Atom

[Guest tar]

Same deal with shooting miniatures - put in a human scale object, add more details, add atmospheric haze.... take an art class  

[bunker]

I would add a transform SOP to make your set bigger before it goes to DOPs
that way you don't need to change gravity  
If your mountain is supposed to be 100m in height, make sure it's 100 units first
then you can always scale it a bit up/down if you need slower/faster look.
 

[Atom]

I have restored my gravity, it was just a suggestion from Jeff Wagner's tutorial. The size is rather large, I realize there is no scale in the posted images but a full size person could stand at the top of the water fall and essentially be only 4-8 pixels high in the image.

My biggest concern is how to control velocity when the fluid collides with itself. That is what separates a large body of water from a small body.

As far as I can determine, the velocity controls have no effect on the FLIP sim whatsoever. For instance, on the source node, I set the scale velocity down to 0.01 from 1.0. This has no effect. Under volume operations I have tried Subtract, Copy, Add, Minimum and I experience no change in the simulation. In the FLIP solver under collision I have lowered the velocity scale and I experience no change in the sim. I have tried lowering the value of the scale on the creation of the velocity volumes. No effect. The box size of my FLIP solver is ~ (100,100,120).

If only I had a place to put a wrangle? All I want is something like this on collision...

v@v *= 0.1;

Is there anyway to directly manipulate the velocity in FLIP?

Why don't the provided velocity controls have any effect?

In this image you can see a hint of the scale. The default grid is slightly visible at the start of the animation near the top.

Edited January 2, 2017 by Atom

[Skybar]

2 minutes ago, Atom said:

If only I had a place to put a wrangle?

You have a bunch of inputs on the flip solver. To do something with the particles you usually use the particle input, and with volumes the volume input (can't remember the specific names). So just lay down a geometry wrangle/gas field wrangle and pipe it into an input.

[Atom]

Thank you, the geometry wrangle is a start and does work. With one line of code I have nearly stopped my simulation from moving.

So the next question is how do I apply the velocity reduction only when it collides with the lower basin portion of fluid. These two fluid parts are part of the same volume. The emitter is "sourced" into the sourcing input on the FLIP solver and the basin is simply geometry that is part of the initial data defined under the SOP Path of the FLIP object.

Here is pseudo code for what I am trying to accomplish.

if (top_part_of_water collides_with lower_part_of_water) {
    v@v *= 0.1;
}

 

[Skybar]

Havent worked with flip for a long time but there is settings on the flip solver to control how splashy it is i think. If that is what you are trying to do. But you could maybe do a ramp along Y and increase the drag the further down it is. 

[Atom]

The ramp is a good idea and I have tried it out but I can't seemed to make it work. I took a similar approach with a simple IF/THEN. But both of these approaches reduce the velocity to zero for the basin water at the bottom.

These approaches are really just work arounds because the main velocity controls do not seem to have any effect.

Does anyone know why the provided velocity controls do not work? I really just want dampening on collision.

Edited January 2, 2017 by Atom

[Atom]

Here is a more clear picture of the problem. When the upper water comes in contact with the lower fluid I see a lot of high velocity white particles. Notice that I have set velocity scale to 0 on the collision TAB. I guess this collision tab is not for colliding with itself? Do I need two FLIP solvers?

 

Edited January 2, 2017 by Atom

[stickman]

Hoping you can figure this out (it's looking great!)
I've yet to get anywhere near this far in houdini, but it seems you want to break this down into modules. Why not just kill the water when it reaching the bottom (no collision with the water there) and then just set up a vector field for the water surface there and a whitewater emitter that you can control independently?

[Atom]

Yeah, I hope to work this out soon, I can't spend another day just wondering why these tools don't work.

I do want ripple interaction a the basin so a sink won't work, it would just drain the basin as well.

This is basically pouring water into a container. I wish I could find the velocity dampening slider. Even the cup examples I have found on this forum suffer from this excessive velocity excitation. Also, my faucet does not react in this manner so I think the FLIP solver may be broken.? Using 15.5.673.

Here is another ridiculous example where velocity runs amok.

[stickman]

Well... that's what I mean. It's fairly easy to create ripples - just take the lake part at the bottom as one problem, the falls as a second... maybe the gischt as a third.

 

[Atom]

Here are reference photos. Notice how when the water enters the basin, in nature, there is not an over zealous amount of velocity and splash.

Edited January 2, 2017 by Atom

[stickman]

If you want to share a scene file I could have a go at it tomorrow.

 

[Guest tar]

3 hours ago, Atom said:

Yeah, I hope to work this out soon, I can't spend another day just wondering why these tools don't work.

I do want ripple interaction a the basin so a sink won't work, it would just drain the basin as well.

This is basically pouring water into a container. I wish I could find the velocity dampening slider. Even the cup examples I have found on this forum suffer from this excessive velocity excitation. Also, my faucet does not react in this manner so I think the FLIP solver may be broken.? Using 15.5.673.

Here is another ridiculous example where velocity runs amok.

 

For the lulz setting the Volume Motion/ Velocity Transfer to Splashy Kernel, Force Scale == 0 and Velocity Smoothing == 0  amazingly works. Damn tools!

[Atom]

Thank you Marty, switching to the Splashy Kernel works wonders. Zeroing out the values does not seem to remove the velocity, you can see all the white particles, but it does prevent the velocity from affecting the surface.

I had been using the Swirly Kernel because I wanted less splashes.

 

@stickman: Thanks for taking a look at my file, I have attached the HIP file shown in this GIF. It is fairly large (51Mb) mainly due to the density of the collision object. But the nodes to model the collision object are included as well.

ap_cmf_waterfall_rebuild_010217.hiplc

Edited January 2, 2017 by Atom

[Skybar]

14 hours ago, Atom said:

I had been using the Swirly Kernel because I wanted less splashes.

I think the name splashy and swirly is quite misleading. Splashy uses FLIP and swirly uses APIC. Even if the names might describe them in some situations, APIC can be a lot more energetic in others because it retains the rotation/angular velocity of the particles. See the glasses here at 5:05 for example https://vimeo.com/159438315

 

 

[Atom]

Thanks for the video. It is nice to see that many examples comparisons in a single video. It looks like Disney deems FLIP unstable in certain situations. But there are these FLIP companion values of 0.95 and 0.99 which seem reasonable. Do you think those two values are Force Scale settings?

[Skybar]

I'm not exactly sure, but I think it's a blend to PIC. So for example 95% flip and 5% pic. I THINK there is a setting on the flip solver to do this but i cant remember now. 

[stickman]

Atom.... haven't forgot you. Just had a job come in. And incompetence on top of that... so anything houdini takes me ages.