ikarus

You have some great shader work there

Why not just sim them separately since you dont want them to interact?

Can we get this in H15? http://www.cs.ubc.ca/~zhxx/IVOCK.html

Those vids should be gifs

The smoke solver as a whole is a navier stokes solving algorithm. The gas project non-divergent dop is only one step of the entire thing.

Having scale transforms on cameras can affect the volume step.

Using the default source method you won't get any kind of pressurized behavior, which I think is what your looking for. Source FLIP basically fills the source area with fluid (surface) and particles, but if that area is already full of fluid you get no additional quantity being added to the simulation. Try adding some divergence to your sourcing.

Gas vorticle forces usually directly changes the velocity field. So if you want to visualize it and change it you'll need to create a separate vector field (which you can attach a vector volume viz to) and apply the vorticle forces to that custom field. Then you can do whatever modifications to that intermediate field and add it to the velocity field. If you have resizing volumes you can just use a gas match field referencing the velocity volume (you can also apply a vector volume viz to this as well).

Is that limited to windows or are we talking all platforms?

laplacian.hipnc So i built the laplacian op and i get the same result from the sop volume analysis laplacian (slice1 to slice2) Comparing techniques it appears that splitting the divergence and gradient operations results in a loss of high frequency values (comparing slice2 to slice3 in the file) Problem with discretizing these pde formulas is the inherent loss of data from sample rate, so simplifying the algorithm to one step probably reduces data loss from sample rate. just my guess

Have you tried making the laplacian calculation in a vop?

I chopped my file from a fluid sim recreation in sops, but if you look at the formula for divergence you divide your sum of differences by 2*stepsize. I believe this comes from a need to have your final output be independent of your fluid resolution, though I don't know enough about descretizing pde equations to give a solid answer.

divergence.hip finite difference divergence, the difference of neighboring voxels, summed over all axes, regularized over the step size of the grid, from a cfd standpoint atleast. good fluids cheat sheet http://http.developer.nvidia.com/GPUGems/gpugems_ch38.html

The paper also includes their approach to simulation using the SPGrid structure.

I saw this was rolled out at siggraph asia 14 but haven't heard anything about this so far http://pages.cs.wisc.edu/~sifakis/project_pages/SPGrid.html There's some definite comp-sci wizardry happening here involving sparse memory paging that's rather difficult to comprehend fully. They compare it to VDB and atleast in the access test results they showed it has quite the speed improvement (What looks to be 10x) in access. Maybe we'll see a VDB-topper for volume data/simulation using techniques like this in the near future? Thoughts?