Unity Package for PDAL

The Pointcloud Data Abstraction Layer (PDAL) is a C++ BSD library for translating and manipulating point cloud data. It is very much like the GDAL library which handles raster and vector data.

This repo is a Unity Package for using PDAL in a project.

This Package is part of the ViRGiS project - bringing GiS to VR.

Installation

The Package can be installed from Open UPM. If you use this method, the dependencies will be automatically loaded provided the relevant scoped registry is included in your project's manifest.json :

scopedRegistries": [
    {
      "name": "package.openupm.com",
      "url": "https://package.openupm.com",
      "scopes": [
        "com.openupm",
        "com.virgis.geometry3sharp",
        "com.virgis.conda-extensions"
      ]
    }
  ],

The Package can also be installed using the Unity Package Manager directly from the GitHub Repo.

Version numbers

This package is a wrapper around a C++ library. We want to keep the link to the library version. However, we also need to be able to have multiple builds of the package for the same underlying library version. Unfortunately, UPM does not have the concept of a build number.

Therefore, this package uses the version numbering proposed by Favo Yang to solve this. This adds two digits for build number to the SemVer patch value i.e. 3.1.1 => 3.1.100, 3.1.101, 3.1..102 etc.

This has the unfortunate side effect that 3.1.001 will revert to 3.1.1 and this means :

Developement and Use in the player

NOTE - For the avoidance of doubt, conda is NOT required on machines running the distributed application. The required libraries are automatically included in the distribution package created by Unity

The scripts for accessing PDAL data are included in the pdal namespace and follow the PDAL C Api.

For C# API is shown in the API Documentation.

The PDAL library is loaded as an unmanaged native plugin. This plugin will load correctly in the player when built. See below for a note about use in the Editor.

This Library currently works on Windows, Linux and Mac platforms.

Running in the Editor

This package uses Conda to download the latest version of GDAL.

NOTE - When installing Miniconda on Windows, you should select the option to add to the Windows Path. This is not the preferred option but IS required for this package to work.

For this package to work, the development machine MUST have a working copy of Conda (either full Conda or Miniconda) installed and in the path. The following CLI command should work without change or embellishment:

conda info

If the development machine is running Windows, it must also have a reasonably up-to-date version of Powershell installed.

NOTE - recent versions of Miniconda for Windows by default create a "Conda Shell" for running conda commands and have not included the conda executables in the general Windows path. For this package to work, the command listed above MUST work in the general Command Prompt and thus conda must be in the path. If the path entry is not created during the Miniconda install, this usually (if Miniconda was installed for one user only) means that the following must be added to the path environment variable in Control Panel: %USERPROFILE%\miniconda3\condabin. If Miniconda was installed for All Users, the actual path will be different and this entry should be updated accordingly

The package will keep the installation of GDAL in Assets\Conda. You may want to exclude this folder from source control.

Documentation

See the API Documentation.

A typical sample program :

using System;
using System.Collections.Generic;
using Pdal;
using Newtonsoft.Json;
using g3;
using UnityEngine;

namespace pdal_mesh
{
    class Program
    {
        static void Main(string[] args)
        {
            Console.WriteLine("Hello World!");
            Config pdal = new Config();
            Console.WriteLine(pdal.Version);

            List<object> pipe = new List<object>();
            pipe.Add(".../CAPI/tests/data/las/1.2-with-color.las");
            pipe.Add(new
            {
                type = "filters.splitter",
                length = 1000
            });
            pipe.Add(new
            {
                type = "filters.delaunay"
            });

            string json = JsonConvert.SerializeObject(pipe.ToArray());

            Pipeline pl = new Pipeline(json);

            long count = pl.Execute();

            Console.WriteLine($"Point Count is {count}");

            using (PointViewIterator views = pl.Views) {
                views.Reset();

                while (views.HasNext())
                {
                    PointView view = views.Next;
                    if (view != null)
                    {
                        Console.WriteLine($"Point Count is {view.Size}");
                        Console.WriteLine($"Triangle Count is {view.MeshSize}");

                        BakedPointCloud bpc = BakedPointCloud.Initialize(view);

                        BakedMesh bm = BakedMesh.Initialize(view);
                        Mesh mesh = bm.ToMesh();

                    }
                }
            }
        }
    }
}

Minimum Project / Test Project

You can see a minimum working project in the test project used to test build this package (and two others):

https://github.com/ViRGIS-Team/test-project.

Use with Unity Cloud Build

As of release 2.4.1, this package will work with Unity Cloud Build.

You will need to add a pre-build script to your build configuration in Unity Cloud Build to load Miniconda.

For Windows configurations - the following script works well:

https://gist.github.com/nimaid/a7d6d793f2eba4020135208a57f5c532

For Macos and Linux, something like this:

echo starting conda install

curl https://repo.anaconda.com/miniconda/Miniconda3-latest-MacOSX-x86_64.sh -o conda.sh
bash conda.sh -b -p ~/local/miniconda3
echo completed conda install