celd Posted October 10, 2019 Share Posted October 10, 2019 (edited) Hi, I am summoning all of you flip gods! In the images you can see the flip explodes upwards for no apparent reason, this seems to happen along the boundary of the collision object. I am using guided ocean layer setup. The collisions otherwise behave correctly. I have tried high res sim, low res sim, changing grid resolution, changing collision geometry (submarine model, simple scaled sphere), deforming collision geometry, object transform collision geometry, zeroing out collisiovel field, high resolution collision vdb, low resolution vdb, playing around with depth of the guided ocean layer, different amount of substeps(both flip solver and autodop), reseeding settings, changing integrator types in the flip solver and probably more... Quite frankly I am at my wits end and I wonder if anyone has ever seen this kind of a problem and if there is a solution to it. I would also like to avoid making the boundary layer exceedingly deep, as this would most likely kill all the detail. I have attached a simplified hip file where this problem occurs. Thank you exploding_water_problem_v1.hip Edited October 10, 2019 by celd Quote Link to comment Share on other sites More sharing options...
celd Posted October 23, 2019 Author Share Posted October 23, 2019 In case anyone ever stumbles upon this thread and wonders what caused these problems, I think I solved it. Inside the gas guiding volume dop, there are 2 gas linear combination nodes for guidingsurface and guidingcollision fields. These nodes should make sure that collision fields are properly accounted for in the guiding surface field, however if your collision field is set up in static object as a "Volume Sample", it produces very stepped guiding surface in the domain of the collisions. The solution to this is to either use "Ray Intersect" on the static object and so calculate the collision field every frame, or disable the 2 gas linear combination nodes, and do the same operation in sops - using vdb combine->sdf difference, and then bring it over with a sop solver. This produces a nice smooth guiding surface and so no weird artifacts happen. 1 Quote Link to comment Share on other sites More sharing options...
dariel Posted August 27, 2021 Share Posted August 27, 2021 (edited) On 10/23/2019 at 11:23 AM, celd said: In case anyone ever stumbles upon this thread and wonders what caused these problems, I think I solved it. Inside the gas guiding volume dop, there are 2 gas linear combination nodes for guidingsurface and guidingcollision fields. These nodes should make sure that collision fields are properly accounted for in the guiding surface field, however if your collision field is set up in static object as a "Volume Sample", it produces very stepped guiding surface in the domain of the collisions. The solution to this is to either use "Ray Intersect" on the static object and so calculate the collision field every frame, or disable the 2 gas linear combination nodes, and do the same operation in sops - using vdb combine->sdf difference, and then bring it over with a sop solver. This produces a nice smooth guiding surface and so no weird artifacts happen. Hi Mate so did this actually fix your issue?, im having the same problem as you with you in the hip file you share but can't get it work, this you make this one work??? Edited August 27, 2021 by dariel Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.