geoff-harper / potree-core

Potree point cloud viewer library core components for easier integration in a three.js project.

Potree Core 2.0

• This project was originally based on Potree Viewer 1.6 and is now since version 2.0 based on the shiukaheng fork of the Potree-Loader.
• Potree is a web based pouint cloud visualizer project created by Markus Schütz.
• This project contains only the main parts of the potree project adapted to be more easily used as a independent library, the code was adapted from the original repositorys.
• Support for pointclouds from LAS, LAZ, Binary files.
• Some features require support for the following GL extensions
  • EXT_frag_depth, WEBGL_depth_texture, OES_vertex_array_object

Demo

• Live demo at https://tentone.github.io/potree-core/
• Double click the models to raycast the scene and create marker points.

Example

• The project can be build running the commands npm install and npm run build.
• Download the potree build from the build folder or add it to your project using NPM.
• Include it alonside the worker folder in your project (can be found on the source folder).
• The build is a ES module, that can be imported to other projects, threejs should be available as a peer dependency.
• Bellow its a fully functional example of how to use this wrapper to load potree point clouds to a three.js project

const scene = new Scene();
const camera = new PerspectiveCamera(60, 1, 0.1, 10000);

const canvas = document.getElementById("canvas");

const renderer = new WebGLRenderer({canvas:canvas});

const geometry = new BoxGeometry(1, 1, 1);
const material = new MeshBasicMaterial({color: 0x00ff00});
const cube = new Mesh(geometry, material);
scene.add(cube);

const controls = new OrbitControls(camera, canvas);
camera.position.z = 10;

const pointClouds = [];

const baseUrl = "data/test/";
const potree = new Potree();
potree.loadPointCloud("cloud.js", url => `${baseUrl}${url}`,).then(function(pco) {
   scene.add(pco);
	pointClouds.push(pco);
});

function loop()
{
   potree.updatePointClouds(pointClouds, camera, renderer);

	controls.update();
	renderer.render(scene, camera);

	requestAnimationFrame(loop);
};
loop();

Notes

• Since potree-core is meant to be used as library and not as a full software as potree some features are not available.
• EDL shading is not supported by potree core.
• Removed classification and clipping functionality.
• Removed Arena 4D point cloud support.
• Removed Entwine Point Tile file support.
• GUI elements were removed from the library
  • PotreeViewer
  • Controls, Input, GUI, Tools
  • Anotations, Actions, ProfileRequest
  • Potree.startQuery, Potree.endQuery and Potree.resolveQueries
  • Potree.timerQueries
  • Potree.MOUSE, Potree.CameraMode
  • PotreeRenderer, RepRenderer, Potree.Renderer
    • JQuery, TWEEN and Proj4 dependencies

Potree Converter

• Use the (Potree Converter)[https://github.com/potree/PotreeConverter/releases] tool to create point cloud data from LAS, ZLAS or BIN point cloud files
• Potree Converter 1.8 creates a multi file structure with each node as an individual file.
• Potree Converter 2.1 creates a single file for all points and separates files for hierarchy index, its faster to create files. Requires a HTTP server configured for file streaming.
• Tool to create hierarquical structure used for point-cloud rendering using potree-core.
• There are two main versions 2.1 witch generates 4 contained files with point data, hierarchy,
• To generate a folder output from a input file run the command .\PotreeConverter '..\input.laz' -o ../output

TXT2LAS

• The potree converter tool only supports las and laz files, so textural file formats such as .pts, .xyz, have to be first converted into a supported format.
• The TXT2LAS tool from the (LASTools)[https://github.com/LAStools/LAStools] repository can be used for this effect.
• To run the tool use the command .\txt2las64 -i input.pts -ipts -parse xyziRGB -set_scale 0.001 0.001 0.001 -set_version 1.4 -o output.laz

To Do

• Supports logarithmic depth buffer (just by enabling it on the threejs renderer), useful for large scale visualization.
• Point clouds are automatically updated, frustum culling is used to avoid unnecessary updates (better update performance for multiple point clouds).