Poisson Points

Poisson Points scatters points at random but never closer together than Min Distance. That constraint is what produces blue noise: the "random yet evenly spread" look, with none of the clumps and holes that a plain uniform random scatter always shows.

The sampler works in 2D: the cloud fills a Width × Height rectangle centred on the origin, and every point sits at Z = 0. Attempts is how hard the sampler tries to place a neighbour around each existing point before giving up on it — higher values pack the plane more densely and cost more time. Seed replays the same scatter exactly.

The point count is not something you set; it falls out of the area and the minimum distance. Count reports what you actually got.

• Natural-looking scatter — Trees, stars, freckles, confetti: anything that should look random but not clumpy.
• Even sampling — A base cloud for Plexus, where clumps would produce a tangle of short connections.
• Stipple and halftone — Combine with an image mask to keep only the points where the picture is dark.
• Particle spawn positions — An evenly spread birth cloud reads far better than a uniform random one.
• Instance layouts — A Replicator stamping a shape per point gets a pleasing, non-gridded arrangement for free.

• To get more points, lower Min Distance — do not raise Attempts. Attempts affects how thoroughly the space is filled at a given spacing.
• Width and Height are in Grafit units, where the shorter edge of the render output spans 2 units.
• Change Seed to re-deal the same distribution; keep it fixed so a project opens the same way every time.
• For a 3D scatter, place the 2D cloud and displace Z with Point Animation or a Noise field.
• Very small Min Distance with a large area is expensive — the sampler is O(points).