Create 3D data from 2D imagery
Download the imagery
Your drone's mission is complete, and you've received a series of images of the Wörthsee development. The images that the drone captured are stored online in a zipped file folder. You'll download the folder, unzip its contents, and look at some of the images before you add them to a Drone2Map project.
- Download the Drone_Images folder.
- Locate the downloaded file on your computer.
Depending on your web browser, you may be prompted to choose where the file will be downloaded. Most browsers download to your computer's Downloads folder by default.
- Right-click the file and extract it to a location where you can easily find it, such as your Documents folder.
- Open the folder.
The folder contains 36 images. You'll take a look at a few of the images before proceeding.
- Double-click the first image, DSC00153, to open it with your computer's default image viewer.
The image shows the office building that is being developed. The building's construction appears to have been completed, which means you'll be able to show the building's stakeholders a fairly close representation of the finished structure.
- Close the image. Open some of the other images to see what they show.
The images show the same building, but from different angles. Drones typically can take two types of images: straight down (nadir) or at an angle (oblique). Both types of images can be used for GIS mapping, but they produce different types of datasets. Nadir images can be used to create 2D orthoimages, which are aerial photographs that show an area with consistent scale and minimal distortion. Orthoimages can be stitched together into larger orthomosaics, from which imagery basemaps are created.
Oblique images, like those you downloaded, can be used to create a 3D point cloud, which uses thousands of points to depict surfaces in a 3D space. A point cloud can then be used to create a 3D mesh, which connects the points with triangles to form a realistic-looking 3D diagram of the area. The end result is a navigable 3D scene that you can share with the stakeholders.
- Close any open images.
Create a new project
Next, you'll create a new project in Drone2Map and add the drone images to a map.
- Open Drone2Map for ArcGIS.
If you don't have Drone2Map, you can sign up for a free trial.
- Sign in using your licensed ArcGIS account.
If you don't have an organizational account, you can sign up for an ArcGIS free trial.
The application opens to the Create new project window. You can choose from five project templates for your project: Rapid, 2D Mapping, 3D Mapping, Inspection, and Batch Processing. Each template creates different types of outputs. The Rapid and 2D Mapping templates create orthomosaics and digital surface models (raster datasets that show the elevation of an area). The 3D Mapping template creates point clouds and textured meshes. The Inspection template orients the images in an inspection viewer, which allows you to cycle through and annotate overlapping images of the same area. The Batch processing template allows you to process multiple images in sequence, prioritizing which projects to complete first.
While the Inspection template may be useful for quickly displaying your images in a spatial context, you'll choose the 3D Mapping template to create a navigable 3D scene of the building.
- For Select Project Template, click 3D Mapping.
The text and images to the right of the list of templates change to explain more about the selected template. You can learn about all of the templates in more detail by clicking them in the list.
- Under the list of templates, click Next.
Next, you'll specify some of the project's parameters and choose the images to use in the project.
- For Give Your Project A Name, type Building Development Project.
You can also choose the location on your computer where the project data will be stored. You can change the default location by clicking the Browse button. Next, you'll add images to the project. You can add image individually, or by uploading a folder.
- Click Add Folder.
The Select Drone Imagery Folder window opens.
- Browse to and click the Drone_Images folder (the folder that contains your images). Click OK to add the folder's images to the list of images.
Each image has a Lat [Y] (degrees latitude), Long [X] (degrees longitude), and Altitude [Z] (height) attribute. These geographic attributes were stored by the drone when it captured each image, allowing the images to be located on a map. If your drone data was taken with degrees from a specific coordinate system, you can specify it here. For these images, the default coordinate system is acceptable.
- Click Create.
Note:Depending on your organization's settings, your basemap may be different. To change the basemap to the Imagery one in the lesson, in the Layer group, click the Basemap button and choose Imagery with Labels.
The project is created and the images are added to the map, with each image represented by a blue point. The points are connected by orange lines; these lines represent the flight path that the drone took to capture the images. The flight path and images indicate that the drone traveled in an orbit around a building (the imagery that displays the building is not from the drone imagery, but is part of the Imagery with Labels basemap that was compiled from satellite imagery). The building looks similar to the one in the drone images, but to make sure the images were properly located on the map, you'll zoom out and view the surrounding area.
- Zoom out until you see Etterschlag and Waldbrunn labeled on the map.
Etterschlag and Waldbrunn are communities within the municipality of Wörthsee (if you zoom out even farther or pan to the south, you can find Wörthsee labeled on the map). The images appear to have been located correctly.
- In the Contents pane, under Project Data, right-click Images and choose Zoom To Layer.
The map zooms back to the circle of blue points.
Process the imagery
Now that you've created a new project and added the drone images to the map, you'll process the images to create a 3D textured mesh to show to your stakeholders. Before you begin the processing, however, you'll specify the processing parameters to ensure that you produce the output that you want.
- On the ribbon above the map, in the Processing group, click Processing Options.
The Processing Options window opens. This window specifies all of the parameters for image processing in Drone2Map. You can specify five kinds of processing options: Initial, Dense, 2D Products, 3D Products, and Resources. Initial and Dense options determine how Drone2Map analyzes each image to match them with the other images, while Resources specifies basic parameters about the project (you already specified these parameters when you created the project). The remaining processing options relate to the type of output data that you're creating. Because you're using the 3D Mapping template, 3D Products is checked and 2D Products is unchecked by default.
When working with your own data, you would generally want to use the highest-quality processing when creating a data product to show stakeholders. For the purposes of this tutorial, you'll instead specify the parameters to reduce processing time.
Depending on the parameters that you specify and your computer's resources, processing time can take between 10 minutes and 1 hour. If you want to create a more accurate output and you're willing to wait the extra time, choose the parameters that prioritize accuracy instead.
- In the Processing window, click the Initial tab. For Keypoints Image Scale, choose Rapid.
Keypoints are points that are generated during image processing and represent a unique location in an image. Generally, thousands of keypoints are generated for each image. Keypoints that represent the same area in different images, known as tiepoints, are matched to determine where images overlap. The Keypoints Image Scale parameter determines how many keypoints are generated depending on the size of the image. The Full option adjusts the image scale for the most accurate results, while Rapid adjusts the scale for faster processing. You can also specify a custom image scale on a scale from 2 to 1/8, with 2 being the most accurate and 1/8 being the fastest.
The next parameter is Matching Image Pairs, which determines which kinds of image pairs are matched with the keypoints. Aerial Grid or Corridor is best used for images that are organized along a grid or corridor (straight line) flight path, while Free Flight or Terrestrial is best used for images that are organized around a specific building or geographic feature (such as the images in your project). Additionally, custom pairs can be set for advanced projects with unique characteristics that need to be taken into account, such as the time images were captured or the distance between images.
- If necessary, under Matching Image Pairs, choose Free Flight or Terrestrial.
The last Initial Processing parameter that you'll set is Matching Strategy. If Use Geometry Verified Matching is checked, another process will check the geometry of the images to ensure that the matches are accurate. This process takes more time to run, so you'll make sure it's turned off for the purposes of this tutorial.
- If necessary, for Matching Strategy, uncheck Use Geometry Verified Matching.
- Click the Dense tab.
The first series of parameters determines how points are created in the point cloud. The Image Scale parameter determines the scale of the composite images that are created from the original drone imagery. Smaller image scales process points faster but are less accurate; you'll accept the default scale. The Point Density parameter determines the number of points per unit of area in the point cloud. Lower point density computes significantly faster, so you'll change this parameter.
- Change Point Density from Optimal to Low (Fast).
The Minimum Number of Matches parameter determines the number of images that each keypoint must appear in to be added as a 3D point to the point cloud. A lower minimum will increase the number of 3D points but will create more noise and artifacts in the resulting point cloud. The default value of 3 is the recommended value, so you'll leave it unchanged.
The Point Cloud Densification parameter determines, in pixels, the size of the grid that is used to match keypoints. It has two options: 7x7 Pixels, which is better for straight-down imagery, and 9x9 Pixels, which is better for oblique imagery. Your images are oblique.
- If necessary, change Point Cloud Densification to 9x9 Pixels.
- Click the 3D Products tab.
The first section of the tab, Create Point Clouds, allows you to choose the output file types for the point cloud. LAS, which is the default, is the standard file type for point clouds in ArcGIS. PLY and XYZ are other available formats for point clouds. You don't really need a point cloud file at all, because you intend to only show the 3D textured mesh to your stakeholders, but it may be useful to keep a point cloud for your own use in the future. You'll leave the default parameters unchanged.
- For Create Point Clouds, confirm that LAS is checked and the other options are unchecked.
Next, you'll choose the output file types for the 3D textured mesh. The first format, Scene Layer Package, is a file type that can be opened in ArcGIS and shared to ArcGIS Online, making it the appropriate choice for your goals. The next format, OBJ, is a widely-accepted file format for 3D graphics and can be opened in a variety of applications. It's also a key component of the Scene Layer Package, so it must remain checked for you to create the Scene Layer Package. FBX, AutoCAD DXF, and PLY are other file types that you can create but are not necessary for the Scene Layer Package. Lastly, you can choose to create a 3D PDF. A 3D PDF is similar to a normal PDF but is capable of displaying 3D images. Because PDFs do not require a GIS application to run, they can be easily shared with your stakeholders.
- For Create Textured Meshes, confirm that only Scene Layer Package, OBJ, and 3D PDF are checked.
The 3D PDF can include your company logo. The default image is the Esri logo, but you can choose the name and location for a different image by clicking Browse.
You'll accept the default values for the remaining settings, which determine the resolution and color balancing. You can learn more about these parameters in the Drone2Map for ArcGIS Help.
- Click OK.
The parameters are applied. The imagery is now ready for processing.
- On the ribbon, in the Processing group, click Start.
The process runs. It may take over 20 minutes to complete, even with settings to minimize processing time. The status bar at the bottom of the application tracks the progress of the processing while it runs and informs you when the process is complete. After the process ends, the Contents pane changes to include a 3D Products group, which contains the Textured Mesh layer (the Scene Layer Package). The LAS point cloud dataset and the 3D PDF, which you also created, are not on the map but are saved in your project folder. For now, you'll explore the 3D mesh.
- Use the mouse scroll wheel to zoom until you can see the entire building in the 3D mesh. Click and drag to pan the map; right-click and drag to tilt or rotate it.
The 3D mesh contains an accurate representation of the building as shown in the original drone imagery. When zoomed in closer, the textures become more noticeable (using the processing parameters for higher accuracy would likely smooth these textures out), but otherwise the mesh will give your stakeholders a good indication of how the building looks in a 3D setting, as opposed to a series of images. You'll share this mesh with your stakeholders, but first you'll open the 3D PDF that you created.
- Open your machine's file browser and open the Drone2Map Projects folder.
You created this folder when you created the project. By default, it was placed in your Documents folder or a similar folder.
- In the Drone2Map Projects folder, open the Building Development Project folder. Open the products folder, and open the 3D folder.
The 3D folder contains all the 3D products that you created in the Building Development Project. It contains the Scene Layer Package (the file with the extension .spk), the point cloud dataset (with the extension .las), the OBJ file, and the 3D PDF.
- Open the Mesh folder, and open the PDF folder. Double-click the project_simplified_3d_mesh PDF to open it.
You need a program capable of reading PDFs, such as Adobe Reader, to read the file. Most computers have an appropriate program installed by default.
- Navigate the 3D PDF. Click to tilt and rotate; rotate the mouse wheel to zoom.
The PDF includes the project name and the date on which the PDF was created. It looks similar to the 3D mesh that you explored in Drone2Map, although it does not contain a basemap. While not as useful as viewing the mesh in a map, the PDF is a much quicker way to share the output with stakeholders. The PDF is also useful for showing the images to stakeholders who may not have access to ArcGIS Online.
- Close the PDF.
Publish the data to ArcGIS Online
Lastly, you'll publish the Scene Layer Package from Drone2Map to ArcGIS Online. Doing so will allow you to share the data among your stakeholders as a scene.
- On the ribbon, click the Share tab. In the Share As group, click Scene Layer.
In the Share As Scene Layer pane, you can choose which data product to share, and change the metadata and sharing properties.
- For Products, check SLPK Mesh.
- For Title, type Building Development Project and add your name or initials to the end of the default name to make sure the name is unique within your organization.
You can also choose which folder in your ArcGIS Online account you'll save the Scene Layer to (the default is My Content). Additionally, you can add tags to describe your layer and make it easier to search for on ArcGIS Online. The default tags are acceptable but are not particularly descriptive of the data.
- For Tags, add Bavaria, Germany, and office building to the list of tags, separating each with a comma.
- For Share With, choose to share the scene layer either with everyone, the members of your organization, or the members of a specific group.
- Click Share.
The Scene Layer Package and a corresponding Scene Layer are published to your ArcGIS Online account. A message window opens, confirming the publication.
- Click OK to close the message window.
You now have a variety of data products you can share with your stakeholders. You have the 3D PDF, which you can open and access without an ArcGIS application. You have the Scene Layer, which you can add to a scene in ArcGIS Online. You also have raw data, such as the point cloud dataset and the OBJ file, which can be opened in other ArcGIS applications. With these datasets, you can present the results of your drone mission to interested parties so they can inspect the results in a 3D environment. Drone2Map can transform your raw drone data into a variety of shareable datasets, allowing you to get the most use out of your drone.
You can find more lessons in the Learn ArcGIS Lesson Gallery.