3D Scanner App vs 3DUNDERWORLD-SLS

3D Scanner App uses Apple devices’ capture methods to create full-color 3D scans of spaces and objects, which users can refine and crop on device. Its modes include LiDAR scanning, RoomPlan for floor plans, TrueDepth for smaller detailed objects, and Photo Mode using Apple’s Object Capture photogrammetry API. Users can measure with a virtual ruler and bounding boxes, and share scans through Share to Web. Scans can be shared as USDZ or exported in formats including OBJ, GLTF, GLB, DAE, STL, PTS, PCD, PLY, XYZ, LAS, and DXF. Geo-referenced LAS point clouds can go to ArcGIS; raw scan data can be transferred to Pix4D or Reality Capture. Scans can also be imported into Blender, 3ds Max, SketchUp, Maya, and other 3D modeling software. The app has a free version and a yearly premium plan at $69.99 a year. It supports iPhone, iPad, and Mac, with minimum requirements of iOS 14.1 or later for iPhone and iPad, and macOS 14.0 or later for Mac. Its privacy policy says captured scans and imagery stay on the user’s device.

3DUNDERWORLD-SLS is a free Linux project that reconstructs point clouds from camera image sequences illuminated by patterned light. Its structured-light process has CPU and GPU implementations, and the command-line program can save reconstructed models as PLY or OBJ files. Version 4.x requires two or more cameras and provides a generic camera interface that programmers can extend. Camera acquisition is not implemented, so users must supply a camera class through the provided interfaces. The project accepts camera image folders, calibration files, and a patterned-light projector as inputs. It does not provide real-time capture or mesh reconstruction. Libraries are included for integration into other projects, with CPU and GPU examples, and an API was included in OpenCV 3.1 for a stereo approach using time-multiplexing Gray encoding. The repository supplies demo binaries, example data, and CMake build instructions. Dependencies are OpenCV 2, CUDA 6.0 or later, and glm. CUDA detection enables CPU and GPU constructors; without CUDA, only the CPU constructor is generated.