Butachan Posted November 12, 2017 Share Posted November 12, 2017 Hi, Can someone please help me confirm is bullet is compatible or not with flip simulations? I set up a simple packed geo as a static object for a flip simulation. The geometry collision is properly calculated as convex hulls… (image1) But as soon as I run the sim this weird shape appears (image2) It is like if the flip sim automatically converts the collision geo into a weird resampled volume. is this true? Would this mean that the flip solver will automatically convert the packed geo into a volume? Feel free to look at my set up… I simplified the scene to run a test. (the hip file is attached at the bottom) bulletandFlip.hiplc I also added this post in SideFx forums as is driving me crazy trying to figure the solution... I have figured a work around transforming my collision geo into volumes and then just using regular static solver but I would like to have a clear answer about what is going on here... Quote Link to comment Share on other sites More sharing options...
Atom Posted November 12, 2017 Share Posted November 12, 2017 (edited) >> It is like if the flip sim automatically converts the collision geo into a weird resampled volume. Correct, that is the proper functioning of FLIP. It resamples dynamic/static collision objects into it's own sample space (i.e. particle separation). Thus you get a lumpy representation of the original object if you don't have enough FLIP resolution. Turn on the collision guide for "Collision" on the FLIPObject. The light purple geometry will display the actual collision geometry that the FLIP Solver will collide with. The blue RBD collision geometry is not used. In this case, there is no need for RBD at all, you have it set to static initialization to begin with. You are better off just creating a Static Collision object instead. Also, if you do want to start off with a static initialization then add activation somewhere in the middle of the simulation you are still better off approaching this in passes. Just solve your RBD as you see fit and write that out as a .bgeo.sc sequence. Then use a File node to bring that sequence in as deforming collision geometry that you submit to FLIP. Here I have converted your RBD object into a static object and it works much better. ap_bulletandFlip.hiplc Edited November 12, 2017 by Atom 1 Quote Link to comment Share on other sites More sharing options...
Butachan Posted November 12, 2017 Author Share Posted November 12, 2017 Thank you @Atom Oh thank you so much. It makes total sense what you say! Like I say it wasn't so much the final result what I was after but at understanding what was going on there, thanks so much! Nico. 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.