 Hello, welcome back on my YouTube channel. Since QGIS 3.32 Lima, QGIS offers native tools for processing point clouds. I'm producing a series of videos on these tools. Today I'll start with explaining virtual point clouds or VPCs. Virtual point clouds are useful because they make it easier to handle large point cloud data sets that are split into square tiles. That's how we often download it from the internet. By referencing other point cloud files, virtual point clouds allow you to view and process point cloud data as a single map layer in QGIS, as an alternative to physically merging them into one file. Additionally, virtual point clouds can reference data hosted on remote servers, allowing point cloud data to be streamed to QGIS for viewing and processing. This can save time and resources when working with large data sets. This blog by Lutra Consulting gives a nice overview of virtual point clouds. You'll find the link to the blog in the description of this video. In this video, we'll make use of AHN4 point cloud data for the Netherlands, which has been made available through geotiles.nl. You'll find the link in the description of this video. We'll use tiles for Rotterdam. If we zoom in on this map, we can see how Rotterdam is subdivided in many tiles. We'll download these tiles and stitch them together as a virtual point cloud. After downloading the tiles, we can add the LAZ files to the map canvas of QGIS. QGIS will then automatically convert the LAZ files to cloud-optimized point cloud files. Let's first check if these point cloud tiles have been well georeferenced. We can easily do this by adding OpenStreetMap as a reference layer to the background. Here we can see that point cloud tiles nicely match the backdrop. Let's open the Processing Toolbox and check which tools are available for point clouds. Here we find point cloud conversion tools, point cloud data management tools, and point cloud extraction tools. Before we continue, I'm going to use the information tool to get more information about the point cloud data that we're looking at. If I run it for one of the tiles, I'll get information such as the attributes and the projection information. Here we see we use the Dutch projection, but with a specific NAP height attribute, which is in EPSG7415 instead of 28992. The tool also reports in the results viewer, and this is also available in the HTML format in the HTML file. I'll close the results viewer, and now I'm going to create a virtual point cloud or VPC. However, if the input files for the virtual point cloud are not COPC files, but last files, QDS will currently only show their boundaries in 2D and 3D views, but processing algorithms will still work fine. But QDS has also created these COPC files, so I'm going to add those to the map canvas instead, and we'll use that then to create the virtual point cloud layer. Here you see that they work. If I zoom in, then those areas show a great detail. Now we can stitch together these COPC layers by running the build virtual point cloud VPC tool. Click the three dots to select all the tiles in our map canvas. Then check the box to calculate boundary polygons, which will enable QDS to show the exact boundaries of the data rather than just rectangular extent. Check the box to calculate the statistics, which will calculate the ranges of values of the attributes. And then finally check build overview point cloud, which will thin the resulting point cloud using every thousandth point from the original data. The overview point cloud will be created next to the VPC file. Use a file name. Here I use AHN for RotterDem and choose the VPC output format. After calculation, click close to close the dialog. Let's remove the COPC layers and inspect the result of the virtual point cloud layer. First we don't see the points, but when we zoom in, the points will appear. And by default it will use the classification attribute. In the layer styling panel, you can switch to the RGB attribute. Here you see all the details when we zoom in, when we're zoomed out, we just see the boundary. This also works well in the 3D view. I'll create a new 3D view here, undock the window and maximize the window, and then here we can see the result in the 3D view, which is very efficient. So the tiles are virtually stitched together and we don't need to bone about the tiles and can simply navigate the tiles that we have.