Point Cloud Surface

Requires the Max or Studio license

Point Cloud Surface reconstructs a controllable 2.5D mesh from a Blender Point Cloud or mesh vertices. It is intended for floors, roads, walls, facades, and other surfaces that can be represented as one depth over a fitted plane.

How to use it

  1. 1 Select a Point Cloud or vertex mesh and use it as the source.
  2. 2 Choose Best Fit, world-axis, or object-axis projection.
  3. 3 Set percentage or cell-size resolution and a hard vertex safety budget.
  4. 4 Choose Triangles, Quad Dominant, or Grid Quads.
  5. 5 Optionally enable Density-Adaptive Detail and set Density Weight plus Refinement Levels.
  6. 6 Configure representative depth, gap preservation, smoothing, thickness, UVs, and shading.
  7. 7 Generate a new surface or update the previously managed output.

Point Cloud Surface Panel

New in 1.0.7 · Max

This panel converts a native Blender Point Cloud object or the vertices of a mesh (a Heavy sparse or dense cloud, or the MM3D Triangulated Points mesh) into an editable 2.5D surface using Delaunay triangulation. It is intended for floors, roads, walls, facades, terrain, and set-extension patches: anything with one depth for each position on a fitted plane. Folds, overhangs, and multiple depth layers must be built as separate surfaces. The core controls sit in a compact two-column grid with the large Generate New Surface button near the top; advanced options are in the collapsible Extra Settings section.

Main Controls

  • •
    Source: the point cloud or vertex mesh. Use Points for Surface in the 3D Point Generator panel fills this automatically; Use Active Output adopts an existing surface.
  • •
    Reconstruction plane: Best Fit, World XY / XZ / YZ, or the source object's local axes.
  • •
    Resolution: by percentage or physical cell size, with a hard Vertex Safety Budget as the upper bound.
  • •
    Topology: Triangles (keeps every unique point at 100 %), Quad Dominant (merges compatible triangle pairs), or Grid Quads (regular cells, can fill small holes).
  • •
    Depth per cell: Mean, Median, Minimum, Maximum, or Percentile.
  • •
    Update Existing: rebuilds the managed output instead of creating duplicates.

Extra Settings

Gap preservation, bounded hole filling, smoothing, normal flip, surface offset, Solidify, generated UVs, smooth shading, and optional wire display. Per-vertex sample_count and surface_depth attributes preserve the reconstruction evidence.

Density-Adaptive Detail

Off by default. When enabled, a budgeted cell hierarchy repeatedly subdivides the region with the highest blended area/density priority, so densely sampled regions receive smaller cells and more vertices while sparse regions keep larger cells. Density Weight blends between uniform spatial coverage (0) and fully density-driven detail (1); Refinement Levels caps how small a cell can get. The vertex budget remains a hard cap, adaptive outputs carry an adaptive_cell_size attribute, and Grid Quads automatically disables adaptive detail because uniform grid faces cannot represent cell transitions without T-junctions.

Clean the point cloud first (Point Cloud Cleanup) and align it (Aligner) so the Best Fit plane is a real floor or wall. Full guide: Point Cloud Surface.

Technical notes

Density-Adaptive Detail leaves larger cells in sparse areas and subdivides densely sampled areas. A weight of zero is spatially uniform; a weight of one is fully density-driven. The vertex budget remains a hard cap, and Grid Quads stays uniform to avoid T-junctions.

Triangles preserve exact unique points at 100 percent when the safety budget permits. Quad Dominant merges compatible triangle pairs. Grid Quads averages regular cells and can fill small holes or add boundary triangles.

This is a height-field reconstruction. Build floors, walls, overhangs, and overlapping depth layers as separate surfaces when they cannot share one projection plane.

Docs 3D Point Generator

The 3D Point Generator triangulates a real world-space point from two observations of the same feature on different frames of a tracked camera. You work directly in Camera View over the footage: Ctrl + Middle Mouse a feature, move in time, and Ctrl + Middle Mouse the same feature again.

Docs Camera Footage Projection

Camera Footage Projection creates a live, non-destructive front projection from a tracked camera background image, sequence, or movie clip onto a mesh.

Docs Multi-Frame Texture Bake

Multi-Frame Texture Bake fuses visible observations from one or more tracked camera plates into a permanent packed texture for static reflection, lighting, and shadow-catching geometry.

 Tutorial How to Add 3D Objects to Tracked Footage in Blender (HDRI, LiDAR and Shadow Catcher) Tutorial Camera Tracking vs 3D Reconstruction: What You Get and When You Need It