Descrição
Modern AECO and heritage documentation projects generate datasets spanning gigabytes to terabytes, exceeding traditional import limits. While proprietary cloud conversion services exist, they impose cost barriers and raise data sovereignty concerns for confidential projects.
This session recorded at Unreal Fest Chicago 2026 presents three open source pipelines for processing and streaming massive 3D datasets into Unreal Engine at runtime.
Firstly, we cover photogrammetry mesh processing using a Blender-based workflow that converts high-resolution scans into optimized 3D Tiles.This approach leverages adaptive octree tiling and UV-based texture baking to achieve up to 62% size reduction on larger models.
Second, we demonstrate a point cloud pipeline using Entwine, PDAL, and Blender to convert E57 and LAZ files into streamable 3D Tiles with proper LOD hierarchies.
Third, we introduce BIM runtime streaming using the Fragment 2.0 format, enabling dynamic loading of architectural models without pre-packaging assets.
The first two workflows leverage Cesium for Unreal's open source 3DTile reader for georeferenced streaming. The third uses a custom open source plugin with Google's FlatBuffers and ThatOpenCompany's Fragment schema for runtime import into UE. All three can be fully processed on-premises. You will gain practical knowledge of each pipeline's architecture and performance—plus access to open source tools for immediate implementation.
Find out more about using Unreal Engine for AECO projects here: unrealengine.com/uses/architecture
#BIMRuntimeStreaming, #PointCloudPipeline, #PhotogrammetryMeshProcessing, #AECO, #Architecture, #CesiumForUnreal, #UnrealEngine, #UnrealFestChicago2026