Business context
The client works with large point cloud datasets from reality capture and survey. Each dataset runs to several GB, far too large for a web browser to load whole. To share and use this data online, the client needed a way to deliver it at web scale.
The challenge
The client needed to render point cloud data several GB in size in the web browser, efficiently and with interactivity. A browser cannot download or hold a full dataset in memory. The data had to be restructured so the viewer loads only what each view needs, and the processing had to run without manual work for every new dataset.
Our solution
We built a backend and frontend that process and streamline the point clouds. An automated pipeline turns LAS and LAZ data into 3D Tiles, an open standard, using octree preprocessing at adaptive resolution. A streaming service then sends only the detail each view needs, from AWS S3 to a progressive WebGL renderer.
- Automated generation of 3D Tiles from large point cloud data
- Octree preprocessing at adaptive resolution
- Support for the LAS and LAZ point cloud formats
- Storage and delivery from AWS S3
- Service to stream 3D Tiles to the browser
- Progressive WebGL rendering with control over level of detail
- Real-time interactivity in the browser
Business impact
Smooth rendering of large point cloud datasets in the browser.
- Teams explore scans in a browser link, with no desktop point cloud software
- New datasets go online through the pipeline, with no hand preparation per scan
- 3D Tiles output works with other viewers, so the data is not locked to one tool
- Object storage and streaming let the data library grow without changes to the viewer
