PointCloud-Visualization-Tool

This is a repository for point cloud visualization built on Unity3D.
The point cloud data can be loaded in format .bin (xyz coordinates in binary format) and .ply.
Any pull requests and issues are welcome. If you find it useful, could you please leave a star here? Thanks in advance.

Projects built based on this repo

MeTACAST

Features

• pointcloud data reader in .bin(xyz coordinates in binary format) and .ply
• Kernal Density Estimation(KDE) of the point cloud density
• Iso-surface construction with Marching-Cube algrithm based on point cloud density
• Color-coded based on point cloud density
• Halo Visualization for point cloud

Install the project and Play the demos

Install the project

• Download Unity3D and Create a new project. Here is a tutorial for Unity3D Setup.
• Clone the repo with git lfs installed or download the archive https://github.com/LixiangZhao98/PointCloud-Visualization-Tool/archive/refs/heads/master.zip and open the project using Unity (versions equal to/higher than 2019 have been tested).

Demo1: Read data, visualize data, calculate density field

• Demo in Assets/PointCloud-Visualization-Tool/Scenes/PointCloudVisualization.unity
• To switch datasets, click the gameobject script/RunTime in Hierarchy and change datasets in the inspector window.
• To calculate the density field of point cloud by the modified Breiman kernel density estimation with a finite-support adaptive Epanechnikov kernel, click the gameobject script/RunTime in Hierarchy and set CalculateDensity as true in the inspector window before running the game.
• To generate iso-surface and change threshold, click the gameobject script/RunTime in Hierarchy and adjust MCGPUThreshold in the inspector window. Then you can see the iso-surface enclosing the region with density higher than MCGPUThreshold just as follows.

Demo2: Color mapping

• Color mapping based on density from low (blue) to high (red)
• Demo in Assets/PointCloud-Visualization-Tool/Scenes/ColorMapping.unity

Demo3: Halo visualization

• Halo visualization published in 10.1109/TVCG.2009.138
• Demo in Assets/PointCloud-Visualization-Tool/Scenes/Halo.unity
• The .ply files can be downloaded from https://graphics.stanford.edu/data/3Dscanrep/

Scripting

Demo in Assets/PointCloud-Visualization-Tool/Scenes/MyPointCloud.unity

Load data from binary files

• Add RenderDataRunTime to an empty GameObject (you can name it whatever you like, here we call it "Runtime").
• Create a new script (you can name it whatever you like, here we call it MyPointCloud.cs) and add it to GameObject "Runtime".
• In MyPointCloud.cs, We first initialize two varaibles particleMat and visCenter. Remember to assign these two variables in the inspector.

public Material particleMat;  // the material of the points
public GameObject visCenter; //The visualization will always follow this GameObject when starting the game. 

• To load data from binary files, we can simply call DataMemory.LoadDataByByte(fileName). An example code is as follows:

public class MyPointCloud : MonoBehaviour
{
    public Material particleMat;  // the material of the points
    public GameObject visCenter; //The visualization will always follow the `Vis center` when starting the game. 
    void Start()
    {
        DataMemory.StacksInitialize();  //Initialize
        DataMemory.LoadDataByByte("Flocculentcube2");  //load the data from the the binary file; the input is the name of the binary file
        RenderDataRunTime.visSize = 1f;  //Set the size of the visualization as 1 meter
        RenderDataRunTime.Init(visCenter,particleMat);  // Assign materials and center to the RenderDataRunTime.cs`
    }
}

• The binary files stores x,y,z coordinates (32bits for each coordinate) for each point in binary located in Asset/PointCloud-Visualization-Tool/data/data folder. Here is a full review of the Point Cloud Dataset included in this repo.

Load data by point positions

• To load data by point positions, we can build a Vector3[] vector3Array and call DataMemory.LoadDataByVec3s(vector3Array,name). An example to generate a group of points in a cubic range is as follows:

public class MyPointCloud : MonoBehaviour
{
    public Material particleMat;  // the material of the points
    public GameObject visCenter; //The visualization will always follow the `Vis center` when starting the game. 
    void Start()
    {
        Vector3[] v = Generate_Cube();  // Generate random points in Cubic shape
        DataMemory.StacksInitialize();//Initialize
        DataMemory.LoadDataByVec3s(v, "cube");  // the first input is Vector[], the second is the name of the data (you can name it as you like)
        RenderDataRunTime.visSize = 1f;  //the size of the visualization
        RenderDataRunTime.Init(visCenter,particleMat);  // Assign materials and center to the RenderDataRunTime.cs`
    }
    public Vector3[] Generate_Cube()  // Generate random points in Cubic shape
    {
        Random.InitState(2);
        int num = 100000;
        int i = 0;
        Vector3[] v = new Vector3[num];
        while (i < num)
        {
            v[i] = new Vector3(Random.Range(-1.0f, 1.0f), Random.Range(-1.0f, 1.0f), Random.Range(-1.0f, 1.0f));
            i++;
        }
        return v;
    }
}

Load data from ply files

public class MyPointCloud : MonoBehaviour
{
    public Material particleMat;  // the material of the points
    public GameObject visCenter; //The visualization will always follow the `Vis center` when starting the game. 
    void Start()
    {
        DataMemory.StacksInitialize();  //Initialize
        DataMemory.LoadDataByPly("dragon_vrip");  //load the data from the the ply file; the input is the name of the binary file
        RenderDataRunTime.visSize = 1f;  //Set the size of the visualization as 1 meter
        RenderDataRunTime.Init(visCenter,particleMat);  // Assign materials and center to the RenderDataRunTime.cs`
    }
}

Traversal of points

To get information of each point, such as the position. We can simply do as following:

void Start()
{
for(int i=0;i<DataMemory.allParticle.GetParticlenum();i++)
{
    Debug.Log(DataMemory.allParticle.GetParticlePosition(i));
}
}

Thanks

Many thanks to the authors of open-source repository: unity-marching-cubes-gpu