I’m trying to implement your paper,and i’m confused about how to make an MAC grid via GVDB? i didn’t see some argument or function about make a vector channel staggered?
hi, I am doing the same thing! According to the information I have right now, you simple use three velocity channels for macgrid. And when doing some gather velocity operation, you can slightly modify the GatherDensity function to suit for macgrid case.
Btw, which step are you in right now? I am writing the CFL condition at the moment, before going to the core steps of FLIP.
Good idea! i’ll try this.
Currently i have a CPU version of smoke simulation using a sparse-grid implemented by self(like a two-level
version of OpenVDB)…and now i’m still learning GVDB’s design and API.
Next step i’ll try to implement the pressure solver first,the MGPCG method.It’s really a challenge XD.
I have CPU version of FLIP using dense grid and PCG. It came to a stage that it’s running but the water is buggy. It took too much time to simulate and debug each single time and Sparse GPU design is the next stage anyway, so I moved to GVDB. Maybe we should talk more about this. I am working full time on this thing.