Unreal Engine World Partition guide: Manage large scenes with cells and Data Layers

Unreal Engine’s official documentation explains how to use World Partition to manage large maps and Data Layers to group scene elements. This guide is for architectural visualization, real-time presentation, and large environment-scene teams looking to set up these systems in the editor, understand cell streaming, and use layers in a controlled way. The documentation does not provide general compatibility information for a specific Unreal Engine version; its Data Layer conversion example uses the UE_5.1 folder path. This example does not mean the feature is exclusive to that version.

1. Enable World Partition and understand how it works

World Partition reduces the need to manually split a large level into sublevels. The world is stored in a single persistent Level file and divided into cells within a configurable runtime grid. Cells are loaded or unloaded based on the location of a streaming source, such as the player, so only the areas currently needed are brought into memory.

When creating a new project, you can use templates in the Games category; the documentation notes that World Partition is enabled in most of them. The Open World level template, designed for large open areas, comes with World Partition, One File Per Actor, Data Layers, and Hierarchical Levels of Detail enabled. Its example map includes a 2 km x 2 km Landscape and an outdoor lighting setup.

To convert an existing level, use Tools > Convert Level or the World Partition Convert Commandlet. If you need to preserve and test the source level before conversion, the documentation’s -ConversionSuffix option adds the _WP suffix to the converted map. -ReportOnly reports the operations that would be performed without converting the level. Checking these options in a working copy can help you handle operations that may affect source files more carefully.

2. Configure cell streaming and loading distance

In World Partition, an actor’s relationship to cells is determined by the Is Spatially Loaded setting in the World Partition section of the Details panel. Enabled actors are loaded when they enter the range of a streaming source, unless they are assigned to a disabled Data Layer. If this setting is disabled, the actor is loaded unless a Data Layer disables it. Keep this behavior in mind when separating elements that need to remain accessible from environment assets tied to a specific area.

Runtime grid settings are in the World Partition Setup section of the World Settings panel. The documentation states that a 2D Runtime Hash grid is present by default and that using multiple grids can hurt performance. Cell Size determines the size of the generated streaming cells, while Loading Range determines the distance from a streaming source at which cells are loaded. The measurements below are example values from the official documentation; they should not be treated as fixed recommended settings for every project.

Use Preview Grids to inspect grid lines in the viewport. The Player Controller is enabled as a streaming source by default; you can also add a World Partition Streaming Source component. Settings such as Target Grid, Priority, and Target State determine which grid cells are affected and whether the target state is Loaded or Activated.

3. Organize scene elements with Data Layers

Data Layers are used to organize actors in the editor and at runtime. It is important to distinguish between the two parts: a Data Layer Asset contains shared, project-level information, while a Data Layer Instance is a world-specific instance. This lets you use the same Asset in different worlds while giving each Instance its own world settings.

To create a new Asset, navigate to the target folder in the Content Browser and select Miscellaneous > Data Layer. Then open Window > World Partition > Data Layer Outliner, right-click, and create an Instance using Create New Data Layer. Assign the appropriate Asset to the Instance. To add selected actors to a layer, use Add Selected Actors to Selected Data Layers in the Outliner context menu, or assign them from the Data Layers section of the actor’s Details panel.

Editor Data Layers are used to organize assets within a project and load them or change their visibility in the editor. Runtime Data Layers can be managed at runtime through Blueprint or C++. For example, keeping different scene configurations or event elements in separate layers and loading them as needed can help manage parts of a large presentation scene. This use adapts the layer loading and unloading functionality described in the documentation to architectural presentations; it is not a guarantee of compatibility with any specific visualization software or ready-made project setup.

4. Considerations for architectural visualization workflows

Teams working with large environment models, alternative regional layouts, or interactive presentations can use World Partition to divide a scene into cells and load relevant areas based on streaming sources. Data Layers can be used to separate editor organization from game logic or to control which elements of the same environment are activated. With One File Per Actor, actors are stored in separate files; the official documentation notes that this can reduce the need for team members to check out the same Level file.

However, creating layers without limit is not a good approach. According to Epic’s documentation, loading too many assets at once can cause performance issues. Common assets used across many Runtime Data Layer Instances can also reduce World Partition streaming performance. Test cell loading and layer usage in your actual scene; don’t distribute assets across an excessive number of runtime layers just to create categories.

Common mistakes

• Not checking whether World Partition is enabled in the map when setting up a Data Layers workflow that requires it. Existing levels can be prepared using Tools > Convert Level or the conversion tool.

• Treating a Data Layer Asset and a Data Layer Instance as the same thing. An Asset is used at the project level; an Instance is used at the world level.

• Applying the example values for Cell Size and Loading Range as fixed settings suitable for every scene. The documentation provides them as examples; project-specific testing is necessary.

• Loading too many assets at once or using too many Runtime Data Layers. The official documentation warns that these situations can cause performance issues.

After setup, test actor assignments, cell visualization, and loading behavior in a small work area first. If conversion is needed, review the options and use a test workflow that preserves the source level; this can help you identify the effects on scene organization and team collaboration earlier.