> For the complete documentation index, see [llms.txt](https://stage-precision.gitbook.io/grid/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://stage-precision.gitbook.io/grid/spatial-interactive-experiences/building-an-interactive-tracking-project/project-guide-overview.md). # Project Guide Overview This guide walks through the complete process of building a spatial tracking project in Grid Studio using real sensor data. The same fundamental workflow applies whether the project uses a single 2D LiDAR, multiple 3D LiDARs, depth-based sensors, or a combination of different spatial sensing technologies. Rather than building the project around a particular sensor type, the guide focuses on creating a reliable processing pipeline that turns spatial measurements into useful interaction data. *** ## What You Will Build During this guide, you will create the complete processing structure required for a typical interactive tracking project. Starting with one or more physical sensors, you will: * Connect the sensors and verify incoming live data * Create reference snapshots for setup and alignment * Align all sensors inside the shared project coordinate system * Combine their data inside a Point Cloud Volume * Define the area that should participate in tracking * Remove unwanted or unstable point data * Configure clustering for the intended application * Refine and stabilize the resulting tracked clusters * Create spatial Zones and trigger Workflow logic * Output real-time tracking data to external systems * Create smaller local tracking areas from a larger global tracking environment * Generate depth images for real-time visual and content-reactive systems * Test and tune the completed tracking pipeline The final project can use any combination of these outputs. A project does not need to use every feature covered by the guide. *** ## The Project Workflow The complete setup follows a consistent sequence: **Connect → Align → Merge → Define Area → Clean → Cluster → Refine → Interact → Output → Test** Each stage builds on the previous one. ### Connect Create the appropriate Sensor Object for each physical sensor, configure its connection, and verify that live point data is visible in the Viewport. ### Align Position every sensor inside the shared project coordinate system so that measurements from different sensors describe the same physical environment. ### Merge Add the sensors as children of a Point Cloud Volume. Their aligned measurements can then be processed as one shared point cloud. ### Define Area Use Include and Exclude geometry to define which parts of the physical environment should participate in tracking. ### Clean Configure the point-processing filters required to remove background geometry, noise, outliers, or unnecessary point density. ### Cluster Choose a clustering method appropriate for what the project needs to detect, such as complete people, objects, hands, or touch interactions. ### Refine Apply additional processing to the resulting clusters, including functionality such as smoothing, prediction, merging, and ID handling. ### Interact Use point-based or cluster-based Zones to turn spatial activity into parameters, counts, and Workflow Events inside Grid Studio. ### Output Send structured tracking data to external systems, create application-specific local tracking areas, or generate depth images. ### Test Validate the complete system under realistic conditions and progressively tune the tracking area, point processing, clustering, and cluster processing. *** ## One Pipeline, Different Applications The project structure is not determined by whether a sensor is described as 2D or 3D. Once aligned, all measurements exist inside the same spatial project environment. For example, a people-tracking setup may combine points from several 3D LiDARs with additional measurements from a 2D LiDAR. The clustering stage can process those measurements together as one spatial data set. A different application, such as a touch wall, may instead use dedicated sensors and a clustering method designed for much smaller interactions. The important design question is therefore: **What should this Point Cloud Volume detect?** The answer determines how the tracking area, filters, clustering, and outputs should be configured. *** ## Working with Multiple Sensors Each physical sensor is represented by its own Sensor Object. After connection and alignment, all Sensor Objects that belong to the same processing goal are placed as children of a Point Cloud Volume. A Point Cloud Volume can therefore contain: * One sensor * Multiple sensors of the same type * Different sensor models * A combination of 2D and 3D spatial sensors The resulting measurements are processed together inside the shared volume. {% hint style="info" %} A Sensor Object can currently belong to only one Point Cloud Volume. If a project requires a completely different clustering pipeline, dedicated sensors must currently be assigned to a separate Point Cloud Volume. {% endhint %} *** ## Three Main Ways to Use the Result A completed tracking pipeline can provide several different types of output. ### Interaction Inside Grid Studio Spatial Zones can react either directly to point data or to the final tracked clusters. They can expose information such as: * Active state * Object or point count * Cluster IDs * Positions * Other available tracking information Zones can also generate Workflow Events and become part of the wider project logic. ### Structured Tracking Data Final cluster data can be sent through Grid Studio's mapping system to protocol objects such as OSC. Typical output data includes: * IDs * Positions * Bounding Boxes * Object counts Additional Volume Objects can also select smaller areas from the global tracking environment and output positions in global, relative, or UV coordinates. ### Depth Images The processed point cloud can be projected into depth images using: * Perspective cameras * Orthographic cameras * Mesh UV coordinates These images are exposed as Image Providers and can be sent through media outputs such as SDI or NDI for use in external real-time visual, interaction, and content-reactive systems. *** ## Recommended Setup Approach Spatial tracking systems are normally built and tuned progressively. Do not begin by trying to perfect the clustering. A more reliable approach is to work through the pipeline in order: 1. Make sure the sensors provide stable data. 2. Align the sensors correctly. 3. Define the useful tracking area. 4. Remove static environment geometry and unwanted measurements. 5. Tune the remaining point data. 6. Configure clustering. 7. Refine the tracked clusters. 8. Add interaction and output logic. 9. Test the system under realistic conditions. Changes made early in the processing chain affect everything that follows. For example, improving the tracking area or removing unwanted point data can often produce a better clustering result than trying to compensate with increasingly aggressive cluster settings. *** ## Using This Guide The following pages follow the recommended setup order: 1. **Connect and Prepare the Sensors** 2. **Align the Sensors** 3. **Create the Point Cloud Volume** 4. **Define the Tracking Area** 5. **Clean the Point Cloud** 6. **Configure Clustering** 7. **Refine the Tracked Clusters** 8. **Add Zones and Interaction Logic** 9. **Output Tracking Data** 10. **Create Local Tracking Areas** 11. **Generate Depth Images** 12. **Test and Tune the Project** These pages focus on the practical project workflow. When a specific filter, clustering method, object, or parameter requires a deeper explanation, the guide links to the corresponding **Reference** documentation instead of interrupting the setup process with detailed feature descriptions.