Class PoissonDiskSampler

A static utility class for generating uniformly distributed 2D points using Poisson disk sampling. Produces sets of points in which every pair of points is separated by at least a specified minimum distance, resulting in a natural, clump-free distribution that is visually superior to pure random placement.

Inheritance
object
PoissonDiskSampler
Inherited Members
Namespace: Scylla.Core.Util.Random
Assembly: ScyllaCore.dll
Syntax
public static class PoissonDiskSampler
Remarks

Poisson disk sampling is ideal for procedural placement of game objects such as trees, rocks, enemies, collectibles, spawn points, and decorations, where uniform coverage without clustering or overlapping is desired.

This implementation uses Bridson's algorithm (Robert Bridson, "Fast Poisson Disk Sampling in Arbitrary Dimensions", SIGGRAPH 2007). The algorithm maintains an active list of accepted points and attempts to generate up to k candidate points in the annulus around each active point. Each candidate is accepted if it satisfies the minimum distance constraint against all nearby points, checked efficiently via a spatial hash grid. Time complexity is O(N) for N output points.

Three entry points are available:

All methods accept any IRandomSource, enabling fully deterministic output when a seeded RNG is provided.

Methods

Sample(IRandomSource, Vector2, BoundChecker, float, int)

Generates a list of 2D points distributed within a custom region defined by a PoissonDiskSampler.BoundChecker delegate, using Poisson disk sampling. This is the core implementation used by SampleCircle(IRandomSource, Vector2, float, float, int) and SampleRectangle(IRandomSource, Vector2, Vector2, float, int). Use this overload when the sampling region is irregular, non-convex, or defined procedurally.

Declaration
public static List<Vector2> Sample(IRandomSource rng, Vector2 startPoint, PoissonDiskSampler.BoundChecker isInBounds, float minDistance, int k = 30)
Parameters
Type Name Description
IRandomSource rng

The random source used to drive candidate generation. Providing a seeded IRandomSource produces a deterministic, reproducible set of points.
Vector2 startPoint

The initial accepted point that seeds the active list. This point is always the first element of the returned list. It does not need to satisfy the isInBounds check (it is accepted unconditionally), so callers should ensure it is within the intended region.
PoissonDiskSampler.BoundChecker isInBounds

A PoissonDiskSampler.BoundChecker delegate invoked for each candidate point to determine whether it falls within the desired sampling region. Must not be null.
float minDistance

The minimum Euclidean distance that must be maintained between any two accepted points. Must be greater than 0.
int k

The number of candidate points tried around each active point before that point is retired. The default value of 30 is recommended by Bridson and produces near-maximum density. Must be at least 1.
Returns
Type Description
List<Vector2>

A List<T> of UnityEngine.Vector2 points accepted by the algorithm. All points satisfy the isInBounds check (except startPoint) and are mutually separated by at least minDistance. The first element is always startPoint.
Remarks

The algorithm starts with startPoint (which must be within bounds, or it will be the only point returned). It then iteratively selects an active point, generates up to k candidate points in the annulus between minDistance and 2 * minDistance from that point, and accepts candidates that satisfy both the isInBounds check and the minimum distance constraint against nearby accepted points. Active points that fail to produce a valid candidate after k attempts are retired.

The spatial grid cell size is set to minDistance / sqrt(2) so that a 5x5 cell neighbourhood search is sufficient to find all potentially conflicting accepted points.

Exceptions
Type Condition
ArgumentNullException

Thrown if rng or isInBounds is null.
ArgumentOutOfRangeException

Thrown if minDistance is less than or equal to 0, or k is less than 1.
See Also

SampleCircle(IRandomSource, Vector2, float, float, int)

Generates a list of 2D points distributed within a circular region using Poisson disk sampling, guaranteeing that no two points are closer than minDistance to each other. Generation begins at the center of the circle.

Declaration
public static List<Vector2> SampleCircle(IRandomSource rng, Vector2 center, float radius, float minDistance, int k = 30)
Parameters
Type Name Description
IRandomSource rng

The random source used to drive candidate generation. Providing a seeded IRandomSource produces a deterministic, reproducible set of points.
Vector2 center

The world-space center of the circular sampling region.
float radius

The radius of the circular sampling region in world units. Must be greater than 0. Points are included if their squared distance from center is at most radius * radius.
float minDistance

The minimum Euclidean distance that must be maintained between any two accepted points. Smaller values produce denser point sets; larger values produce sparser sets. Must be greater than 0.
int k

The number of candidate points tried around each active point before that point is retired from the active list. Higher values fill the region more densely but require more iterations. Defaults to 30, which is recommended by Bridson's original paper. Must be at least 1.
Returns
Type Description
List<Vector2>

A List<T> of UnityEngine.Vector2 points, each within the specified circle and separated from all others by at least minDistance. The first element is always center.
Exceptions
Type Condition
ArgumentNullException

Thrown if rng is null.
ArgumentOutOfRangeException

Thrown if radius is less than or equal to 0, minDistance is less than or equal to 0, or k is less than 1.
See Also

SampleRectangle(IRandomSource, Vector2, Vector2, float, int)

Generates a list of 2D points distributed within an axis-aligned rectangular region using Poisson disk sampling, guaranteeing that no two points are closer than minDistance to each other. Generation begins at the center of the rectangle.

Declaration
public static List<Vector2> SampleRectangle(IRandomSource rng, Vector2 center, Vector2 halfExtents, float minDistance, int k = 30)
Parameters
Type Name Description
IRandomSource rng

The random source used to drive candidate generation. Providing a seeded IRandomSource produces a deterministic, reproducible set of points.
Vector2 center

The world-space center of the rectangular sampling region.
Vector2 halfExtents

The half-width (x) and half-height (y) of the rectangle in world units. Both components must be greater than 0. The rectangle spans from center - halfExtents to center + halfExtents.
float minDistance

The minimum Euclidean distance that must be maintained between any two accepted points. Must be greater than 0.
int k

The number of candidate points tried around each active point before that point is retired. Defaults to 30. Must be at least 1.
Returns
Type Description
List<Vector2>

A List<T> of UnityEngine.Vector2 points, each within the specified rectangle and separated from all others by at least minDistance. The first element is always center.
Exceptions
Type Condition
ArgumentNullException

Thrown if rng is null.
ArgumentOutOfRangeException

Thrown if either component of halfExtents is less than or equal to 0, minDistance is less than or equal to 0, or k is less than 1.
See Also