Here are all the new additions and updates!

UI / UX

• Completely re-worked how the UI is defined on the compiled plug-in. Instead of re-setting all parameter each time step the solver only evaluates animated parameters. Static parameters only get set when the parameter is changed by the user. This improves parameter performance.
• Replaced control ramps with the same multi-parm control ramps from Paradigm. This means you can have a virtually infinite number of control ramps on a force.

Sourcing

• Mask volumes can now be the same name as the source volumes when sourcing VDBs. This means you can mask a source by itself.

Forces

• You can now emit color, aux1, & aux2 from combustion.
• You can toggle the emission into specific fields during combustion.
• Fixed a bug with wind that caused it to not evaluate in some situations.

Core

• Complete swap of the sparse data structure. Moving from an octree to a hash table. The hash table offers better GPU performance of voxel index lookups compared to an octree. It also means that expensive operations such as neighbor sparse tile lookup can be moved from the CPU to GPU. Overall CPU time is much much lower now compared to using an octree.
• Voxels allocations can now be now increased mid-simulation based on sources and other requirements rather than blindly increasing based on a percentage which may not be sufficient. Should eliminate any domain collisions that could have occurred before.
• Added PCG project non divergent. This is slower than Jacobi but is much more accurate. This includes multi grid as well.
• Updating the Jacobi solver for project non divergent to use the same sampler as PCG. This will change the behavior of Axiom 4.0 simulations dramatically. You will need to change the pressure being sourced and combustion to get a similar look. This change does make Jacobi converge much faster and overall the simulations will be more accurate.
• Adding a hour glass filtering option to project non divergent. This should help resolve artifacts that come from using center sampled velocity grids.

Performance

• Added support for another corse level in multi grid project non divergent. This helps it converge faster.
• Adding a new Modified MacCormack advection mode. It is less precise, but faster than normal MacCormack.