Comparative accuracy analysis of iPhone LiDAR applications for high-resolution 3D reconstruction

Accurate three-dimensional modeling of buildings and cultural heritage objects is crucial for applications such as modeling, engineering, and documentation. Traditional methods, such as Terrestrial Laser Scanning (TLS), provide high precision. The recent incorporation of LiDAR sensors into consumer smartphones, like iPhones, presents a cost-effective and accessible alternative. However, the accuracy and limitations of freely available mobile LiDAR applications have not been sufficiently quantified.

This research aims to quantitatively assess the geometric accuracy of iPhone LiDAR as a low-cost alternative for 3D modeling. Three free iPhone LiDAR applications, Modelar, 3D Scanner, and SiteScape, were evaluated across various study cases, focusing on small-scale heritage statues, indoor corridors, and exterior façades of a building. A geodetic reference network was established using a total station, leveler, and RTK GNSS to achieve high absolute accuracy for detailed comparison of the 3D point clouds. Tracking drift was minimized via standardized scanning procedures, maintaining a distance of less than five meters and moving slowly and methodically. The acquired point clouds were processed and compared using CloudCompare, incorporating noise filtering, control point alignment, Iterative Closest Point (ICP) refinement, and multiscale model-to-model (M3C2) analysis.

The results indicate RMS errors ranging from 1.34 cm for small heritage objects to 4.6 cm for building façades, with the Modelar application achieving the highest overall accuracy. Significant errors were concentrated around reflective surfaces such as glass windows, and the removal of these points improved geometric consistency by approximately 50%. All three applications produced point clouds suitable for small to medium-scale indoor and outdoor mapping; however, the 3D Scanner and SiteScape applications exhibited greater deviations, particularly in large or complex environments.

Freely available iPhone LiDAR applications, particularly Modelar, constitute a practicable low-cost option for expeditious centimeter-level 3D modeling in Building Information Modeling (BIM) and heritage documentation. Limitations persist for large-scale architectural elements and reflective materials, wherein TLS remains the benchmark for maximal precision. These results delineate the capabilities and constraints of iPhone LiDAR relative to geodetic references.

Issa Loghmanieh is funded by the Stipendium Hungarian scholarship under the joint executive program between Hungary and Iran.