| SentiSculpt object model generation technology uses computer vision-based algorithms to process sets of real object photos, which were captured from different angles using off-the-shelf cameras or smartphones. The technology does not require expensive 3D scanners and is suitable for wide range of applications, including virtual or augmented reality, video games creation, 3D printing etc.
With SentiSculpt, a user provides photos of an object which were shot to capture the different surfaces or views (the object may be anything from a small toy to a large building, a honeybee to an aircraft carrier). SentiSculpt begins by creating a point cloud based on the input images (2-5 megapixels is the optimal resolution for photos). SentiSculpt can then be used to generate meshes, or wire-frame models, from the point clouds. It can also perform texture mapping and reconstruct the relative camera position for each photo. The generated 3D models are compatible with modern 3D graphics and CAD software.
The SentSculpt software development kit enables the creation of stand-alone 3D object model reconstruction applications, as well as integration with existing 3D graphics and CAD software. The SDK supports C, C++ and C# development environments under Microsoft Windows platform and includes a number of programming samples. One of the samples demonstrates how to create a plugin for MeshLab, a well-known open source system for the processing and editing of unstructured 3D triangular meshes.
SentiSculpt SDK is the latest offering in Neurotechnology's "Senti" line of AI and robotics products, including the SentiBotics mobile robotics development kit and SentiSight SDK for development of computer and robotic vision applications.