Ship on the ocean sim issues

[rporter8555]

Greetings,
Sorry for the long story... thinking others are seeing related issues.

I have been working this sim for a while (several months?) with mixed success.
I started with the sample scene, and again tried with the preset. Both give the same results.
Current aur frame 635 cache file size is 630MB. Grid is 82million cells with 47 million water, 28 million foam, 285K splashes, no mist or wetmap..
I tried a long time with grid scale 1.0, without real success, and am currently testing scale 2.0 with some improvement.
I am running the sim on a slave machine, and the sim time has been slowing down repeatedly, same as it did with scale 1.0.
Preview cache is set to never read for preview before sending to the slave..
I am using a birth volume to control where splash and foam are generated.
The hull shell solid object, clear inside, animated vertices, motion velocity effect 2.0.
I tried for a long time with wave action to have the water hug the bottom and sides of the hull better (compensate for the gaps along the hull from sideways pull of the phoenixfdoceantex displacement), but the front of the container began drawing deep low water trough in front of the ship, so I took out the wave action and changed to a lower value of velocity coherence.
But the low water trough reappears even without the wave action.
So now this run is approaching six days, and the time to sim a frame becomes excessive at 15 minutes, and I stop the sim job, then resend the job to restore the sim at the last completed frame minus 1 so I know it is a good frame.
The slave reloads and begins producing aur cache at about 8 minutes each, but over the next 10 hours slows down to about 13 to 15 minutes.
Stop and restore and the slave again works at about 8 minutes (acceptable for an old i7 machine) but slows again, and the slow sim time symptom repeats.

Now the fun part. Pictures attached are frame 635. I turned on preview the cap mesh and this is what I see.
If this is "particles stuck to the front of the container", they are becoming a real drag in my simulation and seem to reduce the bow wake action.
Is that cavitation I see? Are the particles at the surface dragging grooves in the water?
Is there something I can do to compensate or stop the particles from sticking?
Last month I made the grid larger and moved the ship further back from the leading edge to see if it would help the bow wake, giving more room for the water to "settle" before impacting the bow.
Rendered from aft of the ship, the wake and splash/foam action looks great, but the swell around the bow is not well defined as expected.
Ship speed is about 30 knots. Hull is 512 feet long, 53 feet beam, keel is just above the bottom of the container.
BTW, using two spinning emitters for ship screws, outgoing velocity 40 feet, and another on the front of the spinning screws drawing water in at negative 40 feet. This makes the wake foam action my shipmates seem to enjoy.

Phoenix version 3.11.04 nightly, build ID 2018100228561 from 10/02/18.
Max 2019.2 Update, VRay 4.02.04.00005

Here is my scene.
Sim102518D.zip

In the rendering attached you can see the low water level trough in front of the ship, as wide as the container.
Any advise is greatly appreciated.

Thanks,
Ron Porter
Mustang, Oklahoma

[Svetlin.Nikolov]

Hey,

In order not to have the liquid offset by the displacement near the ship, set the ship geometry as fade volume for the displacement and adjust the fade distance. This would suppress the displacement near the ship.

Additionally, Clear Inside combined with a solid body and fillup for ocean will remove all liquid from the ship interior and thus will create very small gaps between the liquid and the geometry - try disabling clear inside and this way the interior of the ship will be filled with liquid up tot he ocean fillup level - which should not be a problem at render time.

The slowdown till restore and the liquid subsiding at the front are issue we gotta reproduce and look at. Thank you very much for the scene!

Cheers!

[rporter8555]

Hello Svetlin!
I realize I have multiple issues, first with the dragging particles at the front of the container, second with the displacement not hugging the shape of the hull.
Wave force would change the surface of the water before the hull interacts with the water during simulation. I will probably use that method again if the dragging particles can be resolved.
It occurs to me that the beauty of the oceantex is that it pulls and tugs the mesh both vertical and horizontal direction, and the foam rides these displacement waves horizontally and vertically as they do in the sim.
Running the sim with the hull full of water (not cleared) allows water inside the ship to be pulled sideways through the skin of the ship by the horizontal action of the oceantex. This results in "grooves" in the water surface where the hull interacts with the water. It also allows the oceantex to pull water and splash/foam out through the stern plating of the ship which looks very strange indeed. Many tests of this show it is better to not have that water inside the ship.
The wisdom in having the velocity coherence adjustable allows me to simulate ship action far out to sea where the waves are more random in all directions (low coherence) and closer in to land (higher coherence). Using a higher coherence and aligning the ship travel straight into the waves reduces the sideways pull of the oceantex and helps the foam line better track the ship hull with less pronounced gaps.
Pictures attached are frame 690 with fully developed wake action.
The problem I see is that the fade volume kills both the horizontal and vertical displacement factors, essentially pulling the edge of the water back to the flat horizontal line of the original simulation.
This kills the realism of seeing the ship slice through the waves and swells of the turbulent ocean. This is brought into focus as the foam line changes from tracking the turbulent waves to drawing a straight horizontal line past the ship.

Question: Could the fade volume be separated into horizontal and vertical with separate fade distance controls?
In this way the foam line could track the realistic vertical movement of the oceantex displacement, and so the foam would ride the waves, yet through the fade volume be allowed to hug the side of the ship and not be pulled away by the oceantex.

Ron

[Svetlin.Nikolov]

Sounds like a very good idea! Meanwhile started running the sim, but it's gonna take a few days it seems...

[rporter8555]

Awesome... Looking forward to your results.
BTW, can I ask about your rig used for testing the sim?

Ron

[Svetlin.Nikolov]

Hey,

It's Xeon E5-2670 v2 @ 2.5GHz with 32 gigs of RAM.

Btw, I figured that it should be possible to put together a displacement map in the material editor that will have the properties you are looking for. Here is a link to the test scene: [test] displacement suppression only vertical, max 2016.zip

The idea is to use a V-Ray distance texture as a mask for the zone near the ship geometry and away from it. Near it there will be an ocean displacement that will have its X and Y axes zeroed, so only the vertical remains, and away from the geometry there will be a regular ocean texture.

The X and Y axes are encoded in the R and G color components, so I used an Output texture to leave only the Z in the B component. A very important note on the setup is that the V-Ray displacement has it's zero value at 0.5, so a texture that does not displace is of grey color and RGB = 0.5; 0.5; 0.5. This is why there are several output maps that move the displacement maps 0.5 down so the center becomes 0, and then back 0.5 up to restore the V-Ray convention. There might be a more intelligent way to put this together, but this is what I came up with and it seems to work with a simple test. Please give it a try.

Cheers!

[Svetlin.Nikolov]

Alright, reproduced the slowdown and now I have a rough idea in what part of the code it happens. It will take a few more iterations until I'm able to pinpoint what exactly goes wrong though.

[rporter8555]

That's good news! I appreciate your taking the time to help with this.
Here is frame 884. Time 19 minutes. Aborting the sim due to the deep trough in front making it unusable. The amount of "sticking particles" seems to be increasing.
I will work with the displacement material over the weekend.
Ron

[rporter8555]

Svetlin,
The modified displacement helped very much to keep things tight to the ship. I will post test frames in a day or two.
Hope you can make some magic with the sticking particles soon. I need to get the sim started again when it is working.
Thanks again,
Ron

[rporter8555]

Svetlin,
Here is frame 282.
The modified map is working great and will be part of my standard setup.
Thanks again!
Ron

[Svetlin.Nikolov]

Excellent, thank you for sharing! I'm closing in on the issues and will ping you again when there is some progress

Cheers!

[rporter8555]

Hi Svetlin!
Any progress?
I would be very pleased to test anything you might want me to try regarding the stuck particle problem. I have two sim licenses sitting idle waiting for this.
Ron

[Svetlin.Nikolov]

Hey, yes, a bit on both issues - I managed to improve the slowdown but still working on eliminating it completely - there are no changes in the nightlies yet.

Also reproduced the subsiding ocean front - indeed some particle layers are missing and were not generated from the front of the moving grid, so this is the issue that must be fixed:

[rporter8555]

Glad to know of the progress, Please let me know if there is anything I can do from my end to help.
Thanks,
Ron

[rporter8555]

Hi Svetlin!
This is frame 680 from a new simulation run using 3.12.01 Nightly, Build ID 2018120228748.
Displacement does not hide the depression forming in front of the ship.
Sim time slowed down considerably, this frame took 16 minutes.
Ron