BUILD_POTREE

Generates a Potree-compatible point cloud tileset from a LAS/LAZ dataset. The step first normalizes the input dataset using PDAL to ensure a valid bounding box and consistent LAS structure. The corrected dataset is then processed using PotreeConverter, producing a hierarchical octree tileset optimized for interactive visualization in Potree-based viewers.

Typical use: publishing LiDAR datasets for browser-based visualization using Potree or Potree-derived web viewers.

---

Contract

Type	BUILD_POTREE
Accepts	input_las: las
Produces	output_potree: potree
Params	none

---

Inputs

Slot	Type	Description
input_las	las	Source point cloud dataset

Outputs

Slot	Type	Description
output_potree	potree	Potree tileset directory

The Potree output is a directory containing metadata.json, octree.bin, hierarchy.bin, and related files.

---

What it does internally

1. Normalize the input dataset

Before running PotreeConverter, the workflow runs:

pdal translate input.las fixed.las

This normalizes the LAS structure, repairs bounding box issues, and prevents PotreeConverter crashes on malformed files. This preprocessing step is important — PotreeConverter can fail on certain raw LAS files that PDAL handles gracefully.

2. Run PotreeConverter

PotreeConverter fixed.las -o tileset

Produces a hierarchical octree tileset with all necessary metadata and binary octree structures required by Potree clients.

3. Upload and return

The entire tileset directory is uploaded to MinIO and returned as the output_potree artifact.

---

Recipe usage

{
  "id": "build_potree",
  "type": "BUILD_POTREE",
  "inputs":  { "input_las": "job:input_las" },
  "outputs": { "output_potree": "step:build_potree.output_potree" }
}

---

Artifact storage path

artifacts/job_{id}/build_potree/

---

Related executors

• BUILD_COPC — COPC format, better for streaming/analytics
• BUILD_EPT — EPT format, better for Cesium / large datasets