Fluid Compress geometry node

Compresses the output of fluid simulations to decrease size on disk

On this page	Inputs Parameters Particles Volumes
Since	14.5

The Fluid Compress SOP compresses fluid simulations, usually before saving the data to disk. The compression is generally lossy, in that FLIP particles will be culled below the specified depth, and volume bandwidth is limited by zeroing out values. However, SOPs such as Whitewater Source SOP and Particle Fluid Surface SOP are designed to detect compressed fluid input and regenerate any required missing data.

The typical input of this SOP are:

Particles from a FLIP simulation
A surface field representing the SDF field of a liquid simulation
Three fields name vel.x, vel.y, and vel.z representing the velocity vector field of a fluid simulation.

The primary methods of compression by this operator are:

Particle culling below specified depth.
Partitioning the particles spatially and converting them into packed primitives.
Volume bandwidth limiting and conversion to sparse VDB format.
Flagging the VDB’s to save as 16-bit on disk, further reducing disk space.

Note

This SOP creates several detail attributes indicating that the fluid has been compressed. Several of these are also stored to the info block when written to file, making them accessible through the File SOP's Info Load mode or the gstat command line tool. These attributes include:

fluidcompress_particleband and fluidcompress_volumeband - the Particle and Volume bandwidth respectively.
fluidcompress_particlesuncompressed and fluidcompress_particlescompressed - the number of fluid particles before and after compression.
fluidcompress_volumesize and fluidcompress_volumet - the original size and center of the fluid volumes.

For more information, see the Fluid Compression help page in the Fluids chapter.

Inputs ¶

Fluid Simulation

Particles and surface and velocity volumes from a fluid simulation.

Parameters ¶

Particles ¶

Particle Separation

The distance between two particles in the fluid simulation. This parameter should generally reference the same parameter on the FLIP Object SOP of the input simulation.

Cull Bandwidth

If turned on, any particles below this depth in the surface field will be deleted. This parameter is specified in units of Particle Separation.

Keep Attributes

If turned on, all particle attributes except those specified here will be deleted.

Compress to 16-bit

If turned on, all particle attributes specified here will have their precision set to 16-bit floats.

Note

Integer attributes such as id should not be added to this list to prevent them from being cast from integers to 16-bit floats.

Storage Type

The geometry type used to store the fluid particles.

Points

Point geometry.

Packed Points

Partitions the particles into packed primitives using the Pack Points SOP SOP. Packing allows on-demand loading of particles from disk if the resulting files are loaded with Delay Load Geometry enabled on a File SOP or File Cache SOP SOP.

VDB Points

Store the particles into a VDB Points primitive using the Convert VDB Points SOP SOP. This storage type allows data quantization (compression) by storing position relative to the center of the parent voxel.

Position Compression

The position can be stored relative to the center of the voxel. This means it does not require the full 32 bit float representation, but can be quantized to a smaller fixed point value.

Volumes ¶

Note

Input volumes named surface and vel will be converted to VDB format.

Limit Bandwidth

Limit the bandwidth on input surface and vel volumes beyond the specified distance from the fluid surface.

Note

This value should ideally be greater than the Foam Depth parameter on any Whitewater Solver SOP nodes used with this data as a source.

Vel Resolution Ratio

Scales the resolution of the velocity field vel. A resolution ratio of 0.5 will reduce the resolution by half on each axis such that the total voxel count will be reduced by a factor of 8.

Advection Time

The vel volume is limited by advecting backwards through time and clearing any values that do not result in advection to within the Limit Bandwidth of the surface field. If operating at a different timescale you can adjust the parameter here.

Advection CFL

The CFL constraint on the advection used for velocity limiting. Lower numbers are more accurate but more expensive.

Min Speed

Velocity values below this speed are assumed to be zero and cleared to save memory.

Write 16-Bit Floats

A flag to specify that when writing to disk that the volumes should be down converted to 16 bit. This does not affect how they are stored in memory.

Keep Pressure

Keep pressure volume and convert it to VDB.

Inputs ¶

Parameters ¶

Particles ¶

Volumes ¶

Geometry nodes