In the GIS industry, when we talk about LiDAR or 3D point clouds, we tend to think of professional data acquisition equipment, dedicated software, and high costs. Recently, I came across an interesting case: Metaroom, an Austrian company, used an iPhone Pro with LiDAR to scan a hospital in Germany and delivered an editable parametric BIM model. How did they achieve this? What about the accuracy? Let me walk you through it.
Business Background
The team scanned offices, machine rooms, data centers, restaurants, a rehabilitation pool, and operating rooms that were still in use. What impressed me was that the entire process did not disrupt the hospital's normal operations. The scanning was completed simply by walking through the building with a phone. Compared with professional equipment, using an iPhone Pro is light enough. Three people scanned on site for six days, and post-processing took only four days. The whole project took about two working weeks.
According to the company, non-professionals can be trained in about one day, and professional equipment operators are not even required. In Europe, a single terrestrial laser scanner typically costs between EUR 20,000 and EUR 100,000. For a large hospital, a full package from point cloud to BIM often reaches around EUR 500,000. This time, however, acquisition and modeling cost only about EUR 40,000.
Data Accuracy
Since the cost has been reduced so much, is the data accuracy poor?
According to official data, local geometric accuracy has a deviation of about 1%. If two walls are 5 meters apart, the model distance usually differs by about 5 centimeters, roughly corresponding to USIBD LOA 10. Global accuracy mainly depends on SLAM drift, with a magnitude of about 1% of the path length. Scanning 20 meters may accumulate an error of about 20 centimeters. The overall model is approximately LOD 200, and local areas can approach LOD 300.
Point Cloud to BIM
In the past, mobile LiDAR was generally used to generate point clouds or meshes. A key innovation this time is that a BIM model was generated directly. Walls, doors, windows, and openings are all independent objects with position, height, and thickness. Rooms and floors are automatically recognized, and multiple scans are stitched into a whole building. Point cloud geometry can be very accurate, but BIM, MEP, and facility management cannot use it directly. The traditional approach is usually to produce a point cloud first and then manually create the BIM model. Metaroom can generate a structured model at the end of scanning and export IFC, 2D DXF, floor-plan PDF, and other formats.
Technical Implementation and Public Beta
So how is this system currently implemented? According to official disclosures, it is built on Apple RoomPlan, while the company has also developed its own capabilities in LiDAR-camera visual fusion, indoor semantic segmentation, and automatic classification of building components.
The feature is now in public beta. The Scan App requires an iPhone Pro / iPad Pro with LiDAR.
Public beta address:
Enterprise Building Capture Public Beta: https://amrax.ai/3d-model-interaction/metaroom-enterprise-building-capture/
Summary
Large public buildings have traditionally relied on professional terrestrial laser scanning, followed by substantial manual work to convert point clouds into BIM models, with long schedules and high costs. If Metaroom's approach can be widely adopted, it has the potential to dramatically reduce costs and make previously inaccessible modeling and updating tasks feasible. However, there are still many limitations, such as data accuracy and modeling completeness. After all, consumer-grade LiDAR has inherent hardware shortcomings. At present, it is more suitable for documentation and design phases, and should not replace millimeter-level surveying or high-LOD deliverables.
References
- Scanning a 60,000 m² Working Hospital on an iPhone Pro (Lidar News): https://lidarnews.com/iphone-lidar-app-hospital-metaroom/
- Metaroom / Amrax Official Website: https://amrax.ai/