It features native CUDA support, Apple silicon support, 16-bit fields, drag force, improvements to source shapes, performance optimizations, and more.
Matt Puchala has released Axiom 3 – a new version of the sparse GPU accelerated volumetric fluid solver for computer graphics and visual effects. The main new features include native CUDA support for NVIDIA GPUs, Apple silicon support, 16-bit fields, drag force, improvements to source shapes, and performance optimizations.
Previously, Axiom would simulate using 32-bit grids, but now you can optionally use 16-bit grids that have less precision but also use half the memory. Puchala promises about a 36% decrease in total memory when using lower-bit grids.
Now, when using CUDA and Metal APIs, Axiom can use your system memory in addition to GPU memory, so if your GPU runs out of memory, the program will allocate additional resources.
Another change is that source shapes now have a viewport handle, so you don't need to drop a transform after the node to manipulate its position anymore. They also inherit the orientation of their input in addition to their size. Overlapping source shapes will now composite their values when using multiple influence forces.
Here are some other novelties of the release:
- Larger neighbor tile cache sizes. Speeds up faster moving and/or higher resolution simulations.
- Faster advection tracing.
- Improved project non-divergent. Produces less artifacting and is more stable.
- Solver is 8-32% faster overall.
- 3rd disturbance and turbulence force.
- Disturbance seed parameter.
- Velocity drag.
- Plane type source shape. Anything below the plane will be affected.
- Updated to the latest NanoVDB version.
You can use the non-commercial version of the software for free. The Indie license costs $99/year, and the Commercial license is $199/year.