ntkirby Posted September 1, 2017 Share Posted September 1, 2017 (edited) Hello to all... I have a scene where I have animated geometry written out and I bring it back in via a file sop. The animation is just of some rocks flying up in the air flying down inside some water. I'm using a vdb from polygons on the file and I'm bringing that into the dop network as a static object with the mode set to Volume Sample and pointing to the object in the proxy volume field . The problem I run into is when the rocks collide with the fluid, the fluid collapses. I can remove the proxy volume and reset the mode to ray intersect and things will be fine....I have tried to explore on how to fix this but I can't seem to find the solution... does anyone have any ideas? I can upload the scene later if needed. flip_tests.hipnc Edited September 2, 2017 by ntkirby adding hip file Quote Link to comment Share on other sites More sharing options...
Dam Posted September 2, 2017 Share Posted September 2, 2017 (edited) Can you send the hip file please? Cheers. Edited September 2, 2017 by Dam Quote Link to comment Share on other sites More sharing options...
ntkirby Posted September 2, 2017 Author Share Posted September 2, 2017 Ok, I uploaded the file!! You'd figure I would know better since I've posted in these forums in the past Quote Link to comment Share on other sites More sharing options...
Dam Posted September 2, 2017 Share Posted September 2, 2017 (edited) Still can't see it! Edited September 2, 2017 by Dam Quote Link to comment Share on other sites More sharing options...
ntkirby Posted September 2, 2017 Author Share Posted September 2, 2017 I edited my original post...it's there Quote Link to comment Share on other sites More sharing options...
Dam Posted September 3, 2017 Share Posted September 3, 2017 Oh right! Having a look now Quote Link to comment Share on other sites More sharing options...
Dam Posted September 3, 2017 Share Posted September 3, 2017 (edited) This is an interesting case! rocks "static object" having hard time to extrapolate correct sdf values.. I'm starting to belive. however working around creating an sdf vdb called @collision with @collisionvel and adding it in to the third imput of the flip solver as a sourcevolume with preset collision, splash will work fine preserving at the same time the level of water in the pool -- Cheers flip_tests_collVel.hipnc Edited September 3, 2017 by Dam Quote Link to comment Share on other sites More sharing options...
Dam Posted September 3, 2017 Share Posted September 3, 2017 (edited) And yes "static object" still works fine.. equal particle separation on ---> rocks vdb -- terrain vdb -- and "collision separation" helps a lot for the flipSIm and vdb collision detection. The water doesn't collapse enymore expecially with lower "particle sep". Setting the "collision separation" as the particle separation increase the stability too with lowRes flipSims. I've attached the new hip.hope it helps Cheers-Dam flip_tests_debugStatiObjectFIX.hipnc flip_tests_testPrtSepFIX.hipnc Edited September 3, 2017 by Dam Quote Link to comment Share on other sites More sharing options...
ntkirby Posted September 3, 2017 Author Share Posted September 3, 2017 Thanks Dam, I'll take a look here in a few Quote Link to comment Share on other sites More sharing options...
ntkirby Posted September 4, 2017 Author Share Posted September 4, 2017 Ok, so ... after looking at your files, I decided to do a little troubleshooting on my own based off your findings....and It looks like all I needed to do was add a trail sop in..and it wouldn't make the volume collapse... Now, as far as the collision separation, I played around turning it on and off and matching the vdb resolution to that value... what I found was that even though it didn't affect the particles (as long as I had that trail sop in) it still showed slightly weird behavior if I didn't match the resolutions.. if I made collision separation different than the particle separation, I would still get a little weird behavior... So... Add a trail sop after the animated objects before bringing them into flips... vdb resoltuion = collision separation = particle separation Quote Link to comment Share on other sites More sharing options...
Dam Posted September 4, 2017 Share Posted September 4, 2017 (edited) Mmm that's interesting, as far I tested, the trail sop doesn't change the sim in my "flip_tests_testPrtSepFIX", what I changed from the original hip was: the popadvectbyvolumes I've moved it to the second input of the flipSolver, on the VDBs I've switched off fill interior.. that slow down the sim and doesn't make much difference in this case. I've changed also a couple of parameters on collision tab of the flipSolver like the collision extrapolation and at the end matching --> vdb resoltuion = collision separation = particle separation, If a simple trail sop fixed the original issue with only matching the resolutions, would be interesting see this case, can you post the updated hip with your latest debug? Cheers Edited September 4, 2017 by Dam Quote Link to comment Share on other sites More sharing options...
ntkirby Posted September 4, 2017 Author Share Posted September 4, 2017 Here's my file with that... I'm going to do some more tests with the popadvectbyvolume in the particle velocity input and see what that does...that's interesting flip_tests2.hipnc 1 Quote Link to comment Share on other sites More sharing options...
ntkirby Posted September 4, 2017 Author Share Posted September 4, 2017 (edited) Ok, so... I just did a quick test with the popadvectbyvolume in both inputs and removing it altogether with and without the trail sop, and on my end, without that trail sop on the unpacked object... the fluid collapsed....without the trail sop.... it's wierd Edited September 4, 2017 by ntkirby Quote Link to comment Share on other sites More sharing options...
Dam Posted September 5, 2017 Share Posted September 5, 2017 (edited) So I took a look at your latest debug, and yes the trail sop did the magic in this case. Visualizing the velocity on the RBD chunks after the unpack looks like even they don't move at the bottom of the pool the pieces have velocity pointing up and down, I guess that's why the water it's been pushed to the bottom.. I did a quick test deleting attribute @v and it's working as well, so the water collapse at the end it's caused by the incorrect velocity coming from the Dop. Good case that deserved an investigation . Popadvect should be plugged to the second input of flipS, third input accept only volume velocity. Cheers - Dam Edited September 5, 2017 by Dam Quote Link to comment Share on other sites More sharing options...
ntkirby Posted September 5, 2017 Author Share Posted September 5, 2017 That's what that is.. POP Advect by Volumes... that's why it was plugged into the third input... I'm grabbing the velocity volume from the ocean source node and blast anything not @name=vel.* Quote Link to comment Share on other sites More sharing options...
ntkirby Posted September 5, 2017 Author Share Posted September 5, 2017 Thanks for the help Dam... it looks like I just need to get rid of the velocity after I write it out and bring it back in.... I think it's coming from the point vop and it's just not really writing it out from the dop network... or just use that trail sop before bringing it into the flip sim.. either way, we figured it out! 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.