Hi, so it's pretty simple after all to have interesting interaction between RBDs and Lagoa simulations: https://vimeo.com/51754990. But, if you have a bunch of objects in your scene that need to interact with the fluid, you'll need to instance them to a pointcloud - not such a big deal most of the time. There is a small bug with this setup, if you remove members from the ICE topology group, it may crash. Safer just to make a new group.
So as far as I see it the problem is not really interaction with the fluids, that's easy. The problem is more like scalability. If you want a huge, end of the world type of flood then Lagoa won't scale too well at all. Realflow is probably the best commercial solution for massive simulations right now. But, if you're just flooding the bathroom, then Lagoa may well do the job! Ciaran p.s. I miss you Naiad. On Fri, Oct 19, 2012 at 12:22 AM, Pablo Tufaro <[email protected]> wrote: > Thank you very much for taking the time to write all this. I think I have > to find a solution outside softimage. > > Thank you very much really. > > I will keep you posted with the advances if I reach one! > > P. > > El 10/18/2012 6:34 PM, Mathieu Leclaire escribió: > > This is not an easy task to accomplish in Softimage, but it is doable. >> >> I see two possible avenue I would explore: >> >> 1 - What I would try with Lagoa is emitting elastic particles for your >> rigid objects and make the elasticity very stiff. Make sure they have an ID >> based on the emitted mesh and a transformation relative to that emitter. >> Then, at every iteration, I'd figure out a way to average these particles >> transforms by ID to create a transform for the rigid particle group that I >> would apply to the rigid object. So the particle group (by ID) moves the >> mesh. Once you get that average tranform, I would overwrite these elastic >> particle positions by multiplying their saved relative transform by that >> new average transform and basically bring them back into a rest state. >> That'll avoid the rigid mass of particle to deform by their elastic >> properties and will allow proper collision for the liquid particles. You >> are forcing particle positions so it might insert some instabilities in the >> simulation, but I think it should work. >> >> So to recap, you let the elastic group deform and interact with the >> liquid group in one iteration. Then you "undeform" the elastic group by >> their averages to bring them back to a stable state as if it where actually >> rigid. You then simply match the transform of your rigid objects to that of >> it's relative group average transform and you keep iterating. >> >> That's one way. >> >> 2 - A second way would be by mixing Lagoa and Momentum. You would simply >> simulate the liquid by Lagoa with your Rigid Body meshes as collision >> object. So you would need a Deform Bodie with ICE controls, and your ICE >> controls would sample the closest Lagoa particles and create a force based >> on their average velocities. I think that would be doable as well. >> >> I haven't tried any of these techniques myself, but that's where I would >> start experimenting. >> >> Now if you need to do a big flood, you will probably need a lot of >> particles to have a nice looking simulation. I hope you have a good machine >> with a lot of memory and a lot of patience. Lagoa is best suited for >> smaller scale simulation. It's not the best for large scale liquid >> simulations. I would look into Houdini or Naiad (if it's still available >> after Autodesk bought them) for better and quicker results. Maybe even >> Realflow could be a better option. But if staying inside Softimage is a >> must, I would explore these two suggestions. >> >> Good luck! Your going to need it. >> -Mathieu >> >> >> >> Pablo Tufaro wrote: >> >>> Well, that may work...! >>> >>> I will investigate a little bit on that one! >>> >>> Thanks ! >>> >>> pablo. >>> >>> El 10/18/2012 4:03 PM, Oleg Bliznuk escribió: >>> >>>> Here is some work on liquid->rbd interaction http://si-community.com/** >>>> community/viewtopic.php?f=4&t=**281<http://si-community.com/community/viewtopic.php?f=4&t=281> >>>> I think adding backward influence is much more easy task >>>> >>> >>> >>> >> >> >> > -- - Ciaran
