Solvers

Atom Solver

Simulates sand, snow, mud, jelly and other continuum materials on the GPU.

Inputs: Emitters (several): Particle or rigid-piece sources for this solver. Accepts severa; Material: Physical material of the particles or pieces. Connect one mater; Forces (several): Forces that act on the simulation. Accepts several.; Colliders (several): Colliders, floors and sinks of the simulation. Accepts several.; Multiphase: Interaction between Atom materials. Accepts an Atom Multiphase

Outputs: Simulation State: The solver’s result. Pass it through meshers to the Simulation

Setting	What it does	Default
Mode	Integration scheme: Explicit is accurate and crisp; Position-Based is 3x-4x faster at few substeps but softer	EXPLICIT
Resolution	How the background grid resolution is set	GRID
Grid	Background grid cells across the domain. Up to 256 the grid is dense (fastest); above it a sparse grid allocates only the cells the material occupies	64
Particle Separation (m)	Spacing of the particles a volume is filled with. The grid cell is twice this, so the resolution is Domain Scale / (2 x separation), 16 to 1024 cells	0.01
Grid Layout	Memory layout of the background grid	AUTO
Max Particles	Particle capacity. Volume emitters fill their shape at the grid’s particle spacing; past this the fill is sampled coarser. Allocated in tiers of 250k / 1M / 4M	1000000
PB Iterations	Position-based: constraint iterations per substep (more = stiffer, costlier)	3
Melting Point	Temperature (BLUX’s normalized scale) where snow melts and, with Freezing, water freezes	0.35
Latent Heat	Heat (in temperature units) snow absorbs at the melting point before it melts, and water gives back before it freezes: 0 melts at once	0
Freezing	Water colder than the melting point freezes into snow / ice once its latent heat is out	False
Export Deformation	Cache each particle’s deformation gradient and plastic state for rendering and the GPU mesher. Adds 44 bytes per particle per frame	False
Debris Points	Emit a point wherever material breaks (damage crosses the threshold) for secondary debris instancing; shown with the rigid-body debris points	False
Debris Damage	Damage a particle has to reach this frame to emit a debris point	0.5
Sleeping	Particles that stay slower than the sleep speed for the sleep time stop integrating until the material around them moves faster than the wake speed	False
Sleep Speed (m/s)	Particles slower than this for the Sleep Time go to sleep and stop being simulated	0.02
Sleep Time (s)	How long a particle must stay below the Sleep Speed before it falls asleep	0.5
Wake Speed (m/s)	Sleeping particles wake when the material around them moves faster than this. 0 uses twice the Sleep Speed	0.1
Substeps	Fixed substeps per frame when Auto Substeps is off. Each is 0.25 ms of simulated time (x Time Scale), so the result is slow motion	20
Auto Substeps (CFL)	Pick the substeps per frame automatically from particle speed and stiffness, for real-time motion. Off: use the fixed Substeps count	True
CFL	Max grid cells a particle may cross per substep (lower = safer, slower)	0.5
Min	Fewest substeps per frame with Auto Substeps on. Calm frames never go below it	4
Max	Substep ceiling per frame. Stiff materials (snow, rubber) need hundreds; above it the solver runs past its stability limit	400
Time Scale	Simulation speed multiplier (1.0 = real time with Auto Substeps on, 0.15 = bullet-time slow motion, 2.0 = fast-forward)	1
Domain Scale (m)	Edge length of the cubic simulation domain in m. At the same grid resolution a larger domain is coarser	2
Seed	Random seed of the particle sampling. A simulation reruns bit-identically for the same seed	0
Domain Object	World-space position of the simulation domain’s centre. Unset keeps the domain centred on the world origin, floor-aligned at its base
Domain Walls	Boundary wall collision behavior at the edges of the Atom simulation domain	SLIP
Wetting	Fluid moistens nearby Sand / Snow / Elastic; wet Sand and Elastic turn into Mud	True
Wetting Rate (/s)	How fast contact with fluid soaks a particle	20
Mud Threshold	Wetness above which Sand and Elastic convert to Mud	0.6

Blaze Solver

Simulates smoke and fire on a GPU grid.

Outputs: Simulation State: The solver’s result. Pass it through meshers to the Simulation

Setting	What it does	Default
Voxel Size (m)	Edge of one gas cell, in metres: the resolution. Halving it gives 8x the cells where there is gas (the sparse backend only simulates bricks of 8x8x8 cells that hold gas)	0.0625
Size (m)	Dimensions of the domain box, in metres, centred on the world origin with its base at Z = 0 (a Domain object’s bounding box replaces it)	(4, 4, 4)
Domain Object	An object whose world bounding box is the domain box (position and size). Unset: Size, centred on the world origin with its base at Z = 0
Move With Object	The domain follows the Domain object every frame, in whole bricks, keeping the gas where it is in the world (a torch carried by a character, a vehicle’s trail). Sparse backend only	False
Backend	How the lattice is stored on the GPU	SPARSE
Time Scale	Simulation speed multiplier (1.0 = real time at 24 fps)	1
Buoyancy	Upward acceleration per 1000 K of heat, in m/s² at Earth gravity (it scales with the Gravity Force). A Buoyancy Force node adds to it	1.6
Smoke Weight	Downward pull per unit smoke density, in g	0.25
Wind (m/s)	Ambient wind the gas is dragged toward everywhere. A Wind Force node adds a local or global wind of its own	(0, 0, 0)
Wind Drag (1/s)	How fast the gas takes the wind’s velocity	1.5
Rest Reset (frames)	Baked VDBs carry two rest-position grids so render noise moves with the smoke; each restarts this often (the other half a period later) before it stretches too far	24
Ambient (K)	Temperature of the surrounding air, in kelvin (293 K = 20 C). Heat is measured from it	293.15
Advection	How the gas is carried along by the velocity field	AUTO
Multigrid Pressure Acceleration	3-level Geometric Multigrid V-Cycle preconditioner with Red-Black SOR smoothing for 3x-5x faster pressure projection on 128³-256³ grids	True
Open Faces	Domain faces left open to ambient air: gas leaves, and air is drawn in, through an open face. A closed face is a wall. Order: -X, +X, -Y, +Y, -Z, +Z	(True, True, True, True, False, True)
Activity Threshold	Sparse backend: smoke, flame or fuel below this counts as empty air, so its bricks are freed	0.01

Hydro Solver

Simulates liquids with splashes, spray and foam on the GPU.

Inputs: Emitters (several): Particle or rigid-piece sources for this solver. Accepts severa; Material: Physical material of the particles or pieces. Connect one mater; Forces (several): Forces that act on the simulation. Accepts several.; Colliders (several): Colliders, floors and sinks of the simulation. Accepts several.; Two-Phase: A second fluid for a Hydro simulation. Accepts a Liquid Two-Pha; Whitewater: Spray, foam and bubbles for a Hydro simulation. Accepts a White

Outputs: Simulation State: The solver’s result. Pass it through meshers to the Simulation

Setting	What it does	Default
Preset	Turnkey liquid and whitewater settings. Picking one overwrites the values below; Custom keeps yours	CUSTOM
Color Diffusion	Rate of molecular color blending when liquid streams collide	0
Phase Separation	Hydrophobic interfacial repulsion keeping immiscible liquid phases apart (e.g. Oil & Water)	0
Phase Density Ratio	Density of the second liquid (emitter Phase ID 1) relative to the first (Phase ID 0), e.g. 0.85 for olive oil on water	0.85
Domain Scale (m)	Edge length of the cubic liquid domain in m. Its walls contain the liquid	2
Domain Object	World-space position of the liquid domain’s centre. Unset keeps the domain centred on the world origin, floor-aligned at its base
Time Scale	Simulation speed multiplier (1.0 = real time)	1
Substeps	Hydro substeps per frame (automatically adjusted by CFL acoustic sound speed)	4
CFL Safety	Courant-Friedrichs-Lewy stability number (lower = more stable at high sound speeds)	0.4
Particle Radius (m)	Effective radius of individual fluid particles and boundary kernel margin	0.015
Rest Density (kg/m³)	Reference liquid density at rest (1000 = water, 1260 = glycerin, 1800 = mud)	1000
Viscosity (Smoothed)	Artificial kinematic viscosity (higher = thicker, smoother fluid)	0.01
Surface Tension	Cohesion / surface tension strength (causes droplets to bead and round off)	0.05
Restitution	Boundary collision bounce restitution (0.0 = inelastic splash absorption, 0.8 = bouncy drops)	0
Wall Friction	Coulomb tangential friction along domain walls and colliders	0.3
Enable Whitewater	Simulate secondary 3-tier whitewater (airborne spray droplets, surface foam, submerged bubbles)	True
Whitewater Budget	Fraction of whitewater capacity spawned per second at full kinetic/curvature potential	0.5
Spray Velocity (m/s)	Minimum liquid velocity required to spawn airborne spray droplets	2.5
Foam Curvature	Trapped-air surface curvature potential needed to spawn surface foam	3
Foam Lifetime (s)	Decay duration before surface foam dissipates	2
Wave Crest Threshold	Surface curvature threshold for wave crest spray emission (Ihmsen et al. 2012)	1.2
Wave Crest Weight	Weighting of wave crest potential relative to trapped air	1
Trapped Air Weight	Weighting of trapped air potential	1
Air Drag (Spray)	Quadratic aerodynamic air resistance on airborne droplets: a_drag = - C_drag * /v/ * v	0.15
Surface Snapping	Snaps surface foam to float directly on top of the liquid surface	0.8
Foam Repulsion	Inter-particle cellular repulsion preventing foam clumping and generating organic Voronoi patterns	1
Foam Rafting	Lifespan extension factor for dense foam rafts vs solitary bubbles	0.5
Bubble Buoyancy	Upward ascent acceleration scale for submerged bubbles	4
Implicit Viscosity	Unconditionally stable implicit viscosity for thick honey/syrup/slime (Peer et al. 2015)	0
Enable Anisotropy	Yu & Turk 2013 anisotropic covariance tensor to eliminate particle bumpiness in reconstructed surface	False
Max Stretch Ratio	Cap on the ratio between largest and smallest principal axis	4
Radius Scale	Smoothing neighborhood radius multiplier for covariance tensor	2
Min Neighbors	Particles with fewer neighbors than this fall back to an isotropic sphere instead of a noisy covariance fit	4

Impact Solver

Simulates breaking and colliding rigid pieces: demolition, fracture and joints.

Inputs: Emitters (several): Particle or rigid-piece sources for this solver. Accepts severa; Forces (several): Forces that act on the simulation. Accepts several.; Colliders (several): Colliders, floors and sinks of the simulation. Accepts several.; Constraints (several): Glue bonds between pieces. Accepts Impact Constraint nodes.; Rebar (several): Reinforcement that holds pieces together. Accepts Impact Rebar; Debris (several): Chips and dust spawned when pieces break. Accepts Impact Debris; Joints (several): Joints between pieces. Accepts Impact Joint nodes.; Activation (several): When groups of pieces wake up. Accepts Impact Activation nodes.; Triggers (several): Trigger volumes and frame ranges. Accepts Impact Trigger nodes.

Outputs: Simulation State: The solver’s result. Pass it through meshers to the Simulation

Setting	What it does	Default
Auto-Fit Domain	Automatically compute domain scale and position from the fractured asset bounding box with debris clearance margin	True
Debris Clearance Margin	Multiplier around the fractured asset allowing debris and shards to fly freely	2.5
Domain Scale (m)	Edge length of the cubic simulation domain in m. Used when Auto-Fit Domain is off	2
Domain Object	World-space position of the rigid/fracture domain’s centre. Unset keeps the domain centred on the world origin, floor-aligned at its base
Time Scale	Simulation speed multiplier (1.0 = real time)	1
Substeps	Substeps per frame. More is more accurate for stacks and glued chunks, and slower	10
Solver Iterations	Constraint solver passes per substep. More makes glue and contacts stiffer, and is slower	4
Bond Stiffness	Stiffness of the elastic glue bonds that hold neighbouring chunks together	1e+06
Breaking Force	Tensile/shear constraint Lagrange multiplier threshold required to snap glue bonds	45
Chunk Friction	Coulomb surface friction for chunk-ground and chunk-chunk contacts	0.6
Chunk Restitution	Bounciness/elasticity of chunks upon impact	0.25
Restitution Threshold	Impacts slower than this (m/s) do not bounce, so resting stacks settle without jitter	0.2
Static Friction	Give shards a separate static friction that holds a resting contact; off uses Chunk Friction for both	False
Static Friction	Coulomb friction that holds a contact slower than the stick speed (at least Chunk Friction)	0.8
Stick Speed	Contacts sliding slower than this (m/s) can be held by static friction (never less than one substep of gravity)	0.05
Friction Combine	How the friction of two different surface materials combines at a contact	AVERAGE
Bounce Combine	How the restitution of two different surface materials combines at a contact	AVERAGE
CCD Speed	Chunks faster than this (m/s) are swept with continuous collision detection against the ground and box colliders to stop tunnelling. 0 keeps CCD always on	3
Yield Strain	Strain threshold beyond which bonds undergo permanent plastic ductile elongation (car crumpling / bent metal)	0.08
Creep Rate	Ductile metal plastic hardening/deformation rate	0.4
Splinters Per Snap	Secondary GPU micro-splinters / sparks spawned per snapped bond	4
Sound Speed	Speed of the stress wave a blast or heavy impact sends through the glued chunks (m/s): bonds only crack once the wave reaches them. Concrete ~3400, steel ~5100. 0 = instant	0
Rebar Stiffness	Tether stiffness of the rebar along the fractured body’s reinforcement curves (softer than the glue: a bar flexes)	200000
Rebar Yield Strain	Stretch, as a fraction of the bar length between two shards, beyond which the rebar yields and elongates permanently	0.1
Rebar Max Stretch	Most permanent elongation the rebar can take, as a fraction of the bar length between two shards. Past it the bar is stiff again; it never breaks	0.5
Rebar Creep Rate	How fast a yielding rebar takes up its overstretch as permanent elongation	0.3
Rebar Links	Tether anchor pairs at each joint where the bar passes from one shard to the next; more resist bending	2
Linear Sleep Speed	Linear velocity threshold below which chunks begin deactivation (m/s)	0.04
Angular Sleep Speed	Angular velocity threshold below which chunks begin deactivation (rad/s)	0.06
Sleep Delay Frames	Consecutive frames resting below speed threshold required before freezing	12
TGS Contact Damping	Temporal Gauss-Seidel velocity correction factor during contact resolution	0.15
Dynamic Re-Fracture	Enable hierarchical in-flight re-fracturing of shards upon high-energy impacts	True
Shatter Speed	Minimum relative normal approach speed required to trigger impact shattering (m/s)	7.5
Shatter Energy	Minimum impact energy per kilogram of shard, 0.5 v^2 (J/kg; 25 is about 7 m/s). Each doubling above it adds a piece	25
Max Fracture Depth	Maximum hierarchy levels a shard can be re-fractured (prevents infinite recursive subdivision)	2
Max Pieces	Most pieces one impact breaks a shard into (Voronoi cells around the impact)	4
Eject Speed	Speed the new pieces fly apart at, on top of the shard’s motion (m/s)	1
Min Shard Volume	Cutoff volume below which sub-shards will not fracture further (m^3)	0.0005
Plastic Deformation	Let glue bonds yield: stretched past the Yield Strain they lengthen permanently at the Creep Rate (bent metal, crumpled panels). Off: bonds stay elastic until they snap	True