sebkaine Posted January 19, 2016 Share Posted January 19, 2016 (edited) Hi Guys, I would like to know if any H expert have try to use bullet soft bodies inside H ? The bullet API has a bullet soft bodie object / solver class http://bulletphysics.org/Bullet/BulletFull/classbtSoftBody.html http://bulletphysics.org/Bullet/BulletFull/classbtDefaultSoftBodySolver.html Maya has an implementation of this : http://help.autodesk.com/view/MAYAUL/2016/ENU/?guid=GUID-6FA1B885-DF4C-45A8-B7E3-B0991449471C I have not seen this yet in H ? So i would like to know if it is possible and how you would proceed ? My goal is to have interaction beetween RBD (Cube) and SBD (Sphere) directly inside bullet without using an other solver. Thanks for your time Cheers E Edited January 19, 2016 by sebkaine Quote Link to comment Share on other sites More sharing options...
sebkaine Posted January 20, 2016 Author Share Posted January 20, 2016 (edited) Well it looks after some more search that all the soft bodie part of the bullet solver is not implemented in H http://www.sidefx.com/docs/houdini15.0/nodes/dop/bulletrbdsolver I guess that 2 only way to get RBD cube interacting with SBD sphere with bullet is to use either a combo of : - bullet / wire solver - bullet / cloth solver EDIT: well it looks that bullet soft bodie is basically a PBD implementation http://www.bulletphysics.org/Bullet/phpBB3/viewtopic.php?t=7990 http://matthias-mueller-fischer.ch/publications/posBasedDyn.pdf http://image.diku.dk/kenny/download/kelager.niebe.ea10.pdf So i guess the best way to mimic this would be a bullet / POP grain combo Edited January 20, 2016 by sebkaine 1 Quote Link to comment Share on other sites More sharing options...
Shinjipierre Posted January 20, 2016 Share Posted January 20, 2016 Bullet and grain solver should work as well Quote Link to comment Share on other sites More sharing options...
sebkaine Posted January 20, 2016 Author Share Posted January 20, 2016 (edited) Thanks for your answer Remi ! Yes it looks that bullet use a PBD implementation for his soft bodies component. As H does have PBD implemented with the grain solver it doesn't have any sense to repeat this. So i will dig the Bullet / POP grain path ! Cheers E EDIT : This solver can use simplified representation of the objects, such as boxes or spheres, or a composite of these simple shapes to make-up a more complex shape. This solver can use arbitrary convex shapes based on the geometry points of the object, and can also collide objects against affectors that are cloth, solid, or wire objects. POP grain is not mention explicitly in the doc so i will double check that both solver can affect each others ... Edited January 20, 2016 by sebkaine Quote Link to comment Share on other sites More sharing options...
Farmfield Posted January 20, 2016 Share Posted January 20, 2016 I can add there's a point where grain based soft bodies get very hard to control and it starts getting easier with FEM... Here's a grains setup I was doing RnD for that just wouldn't play along trying to balance bouncyness with dampening, ets... And here's the FEM version in a slapcomp... 1 Quote Link to comment Share on other sites More sharing options...
sebkaine Posted January 20, 2016 Author Share Posted January 20, 2016 (edited) Thanks for your feedback Johnny ! I have play quickly with a bullet / pop grain combo. It look ugly and it's slow as hell. I am certainly the one to blame , as i have not use the grain solver very much ... but the fact that the connection are build in a esoteric way is really weird for me. i just want that each point of my object became a particle and that each particle get a bend / shear / tear spring at each connection like nCloth. I will also need to put some inflating pressure in the sphere, i'm mot sure i could get this with POP grain solver , so i will now try the bullet / FEM combo ... i've attach a hip that roughly describe what i'm trying to do ! Cheers E test_v001.hiplc Edited January 20, 2016 by sebkaine Quote Link to comment Share on other sites More sharing options...
Shinjipierre Posted January 20, 2016 Share Posted January 20, 2016 Slow? It's not even a second per frame on my old computer... As for it being ugly, I don't know what you're trying to achieve, you may want to increase the resolution and change the explicit constraints values in the grain update node Quote Link to comment Share on other sites More sharing options...
sebkaine Posted January 20, 2016 Author Share Posted January 20, 2016 yes it's pretty slow for a bunch of small cube and a sphere imo ! i have done this 5 min test with nCloth just to illsutrate the idea. basically : - the cube are RBD - the sphere are SBD with a water balloon behavior - both collide together inside spherical object that rotate Thanks again for your help Remi ! Cheers E nCloth_test.mov Quote Link to comment Share on other sites More sharing options...
Farmfield Posted January 20, 2016 Share Posted January 20, 2016 (edited) Here's a quick tweak to get it working better, but this type of setup is all about tweaking - 5 minutes to set up - 6 months to get it looking good. test_v001.FFv1.hiplc Edited January 20, 2016 by Farmfield 1 Quote Link to comment Share on other sites More sharing options...
sebkaine Posted January 20, 2016 Author Share Posted January 20, 2016 5 minutes to set up - 6 months to get it looking good. Exactly The kind of stuff you try to avoid like the plague ... You're test is far better johnny what are the main stuff you modify to avoid the collision problems with the container ? I will also give cloth a try with this one ! Quote Link to comment Share on other sites More sharing options...
Farmfield Posted January 20, 2016 Share Posted January 20, 2016 (edited) For the penetration issues, just increase the collision padding, then I also decreased the grain separation a tad, I also changed the grains from sheet to solids, decreased the constraint iterations to get them a bit spongier, and in SOPs, something that should be on from the start, I activated the "compute mass" (with a low number) in the grain source as, without it, the solver just has a tendency to do weird stuff (though more important with loose grain setups than using them as sheet/solid). Personally, I would say - to paraphrase Donald Trump - the grain update node is a mess. You really have no idea what stuff does and if you read the help section about it and think you understand what they do, it's still not gonna do what you want, sadly. This is actually a type of setup that is easier to control in PFlow where it's all about the constraints while Houdini's grains are their own constraints, if I understand it correctly - either way, they seem to work more like FLIP than as particles with constraints. It's not the easiest feature to setup and get predictable results from, that's for sure. Edited January 20, 2016 by Farmfield 2 Quote Link to comment Share on other sites More sharing options...
Popular Post rayman Posted January 20, 2016 Popular Post Share Posted January 20, 2016 My setup is 30 mins, not 5 like in maya, but it will give you the idea how to set it up using only Bullet. I'm faking it scattering some rigid objects on the softbody object points and using some 'soft' type of constraint between them. This also gives me the ability to use wrangle code to control the constraints manually. Not the perfect solution but much easier to control. DOP_BulletSoftRigidInteraction_V01.hiplc 16 Quote Link to comment Share on other sites More sharing options...
Juraj Posted January 20, 2016 Share Posted January 20, 2016 Hi Pavel, thanks for file, great idea and it's quite fast. At first I thought that constrained spheres would collapse and loose their volume but they didn't fortunately Juraj Quote Link to comment Share on other sites More sharing options...
sebkaine Posted January 20, 2016 Author Share Posted January 20, 2016 (edited) You are a JEDI Pavel ! This is EXACTLY what i have in mind - 100% bullet solver - VERY fast solve - each point of the object represent a collision point - each point of the object are linked with a constrain - Simple / Concise / Controlable Thanks for your magic Pavel ! @Johnny to paraphrase Donald Trump - the grain update node is a mess. You really have no idea what stuff does and if you read the help section about it and think you understand what they do, it's still not gonna do what you want, sadly This is exactly what i think of this POP grain node ... my personnal word to describe it would be esoteric : 1 - either you have spend times and know every weird controller 2 - either you make some incantation , cross as much finger as you have on both hand and pray to get the good enough presets unfortunatly i'm actually in the second category ... you look to be somewhere in the middle of the two ! ... Lucky guy ! Edited January 20, 2016 by sebkaine 1 Quote Link to comment Share on other sites More sharing options...
Farmfield Posted January 21, 2016 Share Posted January 21, 2016 (edited) @Pavel This is way more like I would have set it up in Max using Pflow, where you use the constraints to build (thus pretty easily control-) the soft body shell - though I wouldn't have known how to do this until I saw how you've set it up here. Very cool. Not often I miss Max but for particles, there's parts in Pflow and TP I just love the simplicity of. @Emmanuel My biggest problem is I'm often on autopilot, so I'm not really conscious about what I'm doing, tweaking stuff, even when I get it to work - and it's not that I don't know, I'm just not overly aware of that knowledge, hehe... There's just something with Houdini that just clicks with my mind, I guess, this last Monday, the 18th, was my 1 year anniversary of officially migrating from Max to Houdini, so it's not that I've using it for that long... Edited January 21, 2016 by Farmfield Quote Link to comment Share on other sites More sharing options...
Ryan Posted May 20, 2016 Share Posted May 20, 2016 Hello, Guys. I wanted to share this with you. Here I am using @Pavel's awesome method. I am building the constraints using the geometry edges. I am really surprised how well this works! For the initial rollup, I am using a VEX method from a fellow from this thread: http://forums.odforce.net/topic/25760-curling-hair-curves-with-vex/#comment-149896 Thanks alot! bullet_cloth.hipnc 9 Quote Link to comment Share on other sites More sharing options...
xs2222 Posted May 24, 2016 Share Posted May 24, 2016 Pavel as usual , i reckon Quote Link to comment Share on other sites More sharing options...
dyei nightmare Posted July 1, 2017 Share Posted July 1, 2017 great examples, is somewhat ironic and sad that constraint relationships works a lot better than cloth in terms of collisions, i was thinking that houdini was the best in terms of cloth Quote Link to comment Share on other sites More sharing options...
anthonymcgrath Posted July 4, 2017 Share Posted July 4, 2017 that is a really cool setup - very smart! 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.