anicg 2 Posted November 7 I (beginner in FLIP Fluid) have a FLIP simulation that gets crazy when I add a collision (static object) I use the test geometry shaderball, used 2 spheres with mountain sop, boolean with front and back of the shaderball. The two deformed spheres are the source. The rest (boolean difference) of the shaderball is the collision. Flipbook Without collision https://streamable.com/qkh2p Flipbook With collision https://streamable.com/922a6 Ii feel it has something to do with possible intersection between particles and collision geo, so I reduced the size of the sphere to leave a gap with the collision geo but the flip sim is still going crazy. Any idea? Issue with FLIP simulation.hiplc Share this post Link to post Share on other sites
LucaScheller 29 Posted November 7 (edited) Hi, Your dop collider doesn't use the sop transform. Make sure you specify the proper "OBJ Path" and check the "Use Object Transform" checkbox. To verify that it works, enable the "Display Geometry" Checkbox (The collider should then move along properly when viewed inside the dop net). This should solve wrong collisions, it still explodes though since every thing is moving really fast. Edited November 7 by LucaScheller Share this post Link to post Share on other sites
anicg 2 Posted November 8 (edited) Thank you. I've fixed that part you showed. But for the "going crazy" part of the flip simulation, that can't be due to the fact that it's moving very fast, because it's moving very fast in both cases (with and without collision). Without collision, the fluid looks very smooth, yes there is a lot of movement but it looks calm. However with collision, the fluid is shooting up in the air, there are particles shooting all over the place, the same set up just added the collision. Here is a second version of the flip books and a hip file attached (after correcting what you've highlighted): Without collision v2: https://streamable.com/m2mpq With collision v2: https://streamable.com/v4cs5 I have the same problem in another simulation (rbd) as well. Is there something wrong in the collision? I'm dong this for the first time, I must be doing something wrong. Issue with FLIP simulation v2.hiplc Edited November 8 by anicg Share this post Link to post Share on other sites
DonRomano 26 Posted November 8 (edited) Are you using drag in your simulation ? If yes, delete it and use a Geometry wrangle with (connected to the particle motion input) : v@v *= ch("drag_threshold"); And how many substeps do you have ? Try 3-3 it may solve your issue. Cheers, Edited November 8 by DonRomano Share this post Link to post Share on other sites
anicg 2 Posted November 8 I did not add any drag (unless it is there by default and I did not find it/turn it of). The set up is very very simple. It is a failed attempt to mimic the free "Applied Houdini - Liquids 1", his fluid sim looks normal, mine is shooting all over the place. Share this post Link to post Share on other sites
DonRomano 26 Posted November 8 Try to change the division method to "division size" and put 0.005, and link it to the volume division in the collision source sop Share this post Link to post Share on other sites
anicg 2 Posted November 8 (edited) Thanks, I've just tried it (also linked to the collision source voxel size) but it is still shooting up in the air, quite high Edited November 8 by anicg Share this post Link to post Share on other sites
Andrea 73 Posted November 8 The issue here is not the velocity of flip but the velocity of the collision geometry which is very high ( too much). It's spinning really really fast, I can see values between -1100 and 600. You can see immediately this problem also in DOPs, enabling "Collision Velocity" under Guides. If you don't want to change the sphere animation, you can reduce the velocities by using a wrangle before OUT_COLLISION_POLY or in the collision tab reducing the velocity scale, which is basically the same thing. Here is the hipfile. Does this help? Issue with FLIP simulation.hiplc Share this post Link to post Share on other sites