Flickering Surface and other liquid-related problems

[MANUEL_MOUSIOL]

I am working on my "wine-tasting" video and just poured some liquid into a glass.
I am having following problems:

• The liquid's surface is developing some sharp upward shapes. There is some motion going on but it doesnt call for these movements. These sharp small sharp shapes flickering out of the liquid are bothering. I already animated the viscosity and surface tension but they are still there.
• The liquid doesnt seem to be able to relax and take a steady form. Maybe the initial force of filling the glass is so heavy that it takes a long time to balance everything out. is there a way to make that faster? again, I tried animating viscosity and tension etc, but it still needs more time than I wish it did.
• I am really having problems with the way PhoenixFD is detecting and calculating the objects it has to interact with. Sometimes I have to make the glass really thick so no liquid is leaking, sometimes my original thin glass model is sufficient. Instead of modifying the model, isnt there an option to tell Phoenix how to react with the object (make the objects LOOK like they are thicker from the simulators point of view)? and to make that calculatable from the users point of view?

Find attached a screenshot with a sample of the sharp upward shapes and the max file.

Thanks for any suggestions!

Manuel

[Ivaylo Katev]

Hi,
the sharp surfaces you have are always present and usually are cut off by the gizmo. Seeing the image your rendering setup is not perfect, here you can see how a glass filled with liquid should look. Note the fact that the glass is not visible, it seems like the liquid starts from the very beginning of the wall. In your rendering the walls are visible, that means the liquid does not penetrate the glass.
after seeing the scene will be able to say more

[MANUEL_MOUSIOL]

yes, that is connected to the other problem - that Phoenix has problems identifying something as a wall.
Since I made the glass thicker so the liquid doesnt go through it but made it to the original state for my rendering, the liquid didnt touch the glass anymore..

[Ivaylo Katev]

about the liquid penetrating the wals - by default the mode is inscribed. i recommend to keep it like this for such simulations, and to increase the resolution until the glass walls are at least two cells wide.
about the residual movement - animating the viscosity and all other stuffs are not helping, actually during the years i accepted the position that animating simulation properties does not help. i changed the scane to more realistic one and increased spf to 8, now i'm waiting the result

[MANUEL_MOUSIOL]

Did you change something in the latest build? I can't find the inscribed mode.. Or am I blind?

[Svetlin.Nikolov]

Yup, it moved to the Interaction panel - sorry for the inconvenience. We decided to unload the specific controls from the Simulation panel and send them around to more appropriate places.

[Ivaylo Katev]

yes, actually the voxelization belongs to the interaction panel by meaning. this is confusing for the old users, but is convenient for the new ones

[Ivaylo Katev]

here is the result
i have changed the scene scale to more realistic , the resolution to 20M, the voxelization to center, spf to 8.
glass scale simulation have very high dynamics and actually 8 is too low, the liquid surface is rough, but the mesher's smooth factor is 60 that smashes the small waves. actually the high smooth factor produces the small teeth at the liquid-glass interface, but they are hidden into the glass wall. surprisingly for me, the liquid-glass interface is rendered well even without gizmo (it's not supported in mesh mode). the shell was switched off during the simulation, and switched on during the rendering
i will try to find better settings, because the realistic scale causes too slow simulation, the liquid must have perfectly flat surface even without smoothing, and some surface tension also must be added, because now it looks like large scale simulation.

[MANUEL_MOUSIOL]

hi, sorry for the late reply, I didnt realize there were more answers since i looked.

thank you very much for looking into it!
would you mind sharing the scene with me?
I just started the simulation with the settings you suggested and - yes - it became extremely slow.
what would be a good setting for previews? just reduce cell size (but not too low or else liquid will be leaking again)?
what is a realistic scale for you? it was set to mm and now I set it to cm. for a small-scale simulation like this both seem quite realistic to me...

btw, I just added the shell-modifier to make the glass thicker so it wont leak. In the final render I would like to keep it switched off.

but what worries me is that at the end, when the liquid begins to settle, it should really look like liquid in a glass. that means small movements and maybe some ripples and not one huge wave going back and forth from the glass walls. I am not finished with the simulation yet (est. is 3h!) but from what you wrote and the screenshot it really looks like a large scale simulation. does one have to animate the simulator to achieve this? that would be kind of weird...

[Ivaylo Katev]

here is the scene
right now we are preparing a sample scene for pouring wine glass from a bottle, this is more complicated case, but many people asked for it.
the main reason for the slow simulation in your case is the thin walls of the glass. this requires big resolution of the grid.

[MANUEL_MOUSIOL]

alright, thank you a lot, Ivaylo!

So setting the glass thickness higher would make the simulation easier...there are also artistic questions in the matter of course, but its good to know.

I am looking forward to the sample scene a lot! Hope it will be ready soon

BTW, speaking of making the interface more logical: wouldnt the isosurface level in the rendering panel fit better to interactions too, as it obviously changes the overall shape a lot, and not even only at render times but in the viewport already. just a thought...

[Ivaylo Katev]

interaction panel controls the interaction of the simulation with the scene objects, it is not related to the rendering.
if someday we separate the simulator and the shader (there is such a probability) the interaction panel goes to the simulator.

[MANUEL_MOUSIOL]

I dont quite understand what the isosurface level is connected to the rendering, seeing that it affects the shape already in the viewport...but then again I guess it doesnt have anything to do with actual simulation result...?

I did the simulation again, just with a grid resolution of 16mil. It took 2 hours on my machine for 200 frames. I dont know if this is a normal time or slow...not enough experience.


Please take a look at this video. Why is the liquid reacting so aggressively to the falling drops of water at the end? it is totally going crazy...

also, it bothers me that the liquid takes a long time to run down the glass, it seems like its stuck in the glass geometry even though viscosity is 0...maybe making the glass thicker so it gets swallowed by the gizmo would help? But the gizmo doesnt swallow the results of the drops coming down from the wall of the glass, so that would look weird...

[Ivaylo Katev]

can't say without the settings. the boiling in the left side makes me to suspect FT advection, in the last wine scene i tried to use it for the glass, but it does not produce flat surface, finally i switched to classic.

[MANUEL_MOUSIOL]

can't say without the settings. the boiling in the left side makes me to suspect FT advection, in the last wine scene i tried to use it for the glass, but it does not produce flat surface, finally i switched to classic.

It is basically the file that you shared with me a few posts ago. there the same thing happens.
please find your scenes rendered out attached.