Blender Point Distribute vs Distribute Points on Faces

In Blender's Geometry Nodes, the transition from version 2.93 to 3.0 marked a major architectural shift from a text-based attribute system to the modern "Fields" system. As part of this overhaul, the legacy Point Distribute node was deprecated and replaced by the Distribute Points on Faces node. While both nodes generate a scatter of points across the surface of a mesh, the newer Distribute Points on Faces node provides a more direct, flexible, and visual workflow with built-in rotation and normal outputs.

The Legacy "Point Distribute" Node

In Blender 2.92 and 2.93, Geometry Nodes relied on named attributes entered as text strings. The Point Distribute node required users to:

  • Specify distribution methods via a dropdown menu (Random or Poisson Disk).
  • Manually type in vertex group names or attribute strings to control point density.
  • Use separate nodes, such as Attribute Vector Math or Point Rotate, to calculate alignments and normals for instanced objects.

This string-based approach often led to errors from typos and made data flow difficult to read across complex node trees.

The Modern "Distribute Points on Faces" Node

Introduced in Blender 3.0, the Distribute Points on Faces node operates within the node-and-socket Fields system. Instead of relying on string inputs, it directly accepts field connections, allowing for procedural control over scattering.

Key features include:

  • Socket-Based Density: Density and density factors can be controlled directly by connecting procedural textures, math nodes, vertex weights, or image maps into the input sockets.
  • Direct Selection Masking: A dedicated Selection input allows you to restrict point distribution using boolean logic, face sets, or geometry proximity.
  • Integrated Rotation and Normal Outputs: The node automatically calculates and outputs surface normals and aligned rotations, eliminating the need for extra alignment nodes when instancing objects like trees, rocks, or debris.
  • Distribution Modes: It retains both Random and Poisson Disk methods, with Poisson Disk adding minimum distance and density maximum controls to prevent overlapping points.

Summary of Differences

  • Data Handling: Point Distribute used text strings for attribute lookups; Distribute Points on Faces uses visual data-flow sockets.
  • Orientation Data: Distribute Points on Faces automatically outputs point rotations aligned to surface normals, whereas Point Distribute required manual vector calculation.
  • Efficiency: The modern node integrates seamlessly with procedural noise, ramps, and math nodes without the need to create, store, or clean up intermediate named attributes.