Medium and long-term development of products in the field of photogrammetric 3D modeling – taking into account that the everyday use of a 3D modeling based on photogrammetry usually fails due to two major weaknesses:
- a rudimentary handling of the camera poses;
- a necessary filling of holes in the 3D models.
As a solution to this, two self-developed methods are currently available:
- Own method for a monocular stereoscopy – in particular including smartphone specific boundary conditions: Convergent Monocular Stereoscopy.
- New method for interactive hole filling in 3D surface meshes: Interactive Filling of Holes in 3D Meshes using localized "Screened Poisson Surface Reconstruction".
A third method of estimating camera poses is in progress:
- Stereo Depths and Occlusion Boundaries based Estimation of Camera Poses.
Since the process mentioned under 2. above allows hole filling after surface reconstruction, an alternative process is under development which serves to detect and fill holes in 3D point clouds – namely before any surface reconstruction has taken place.
From around 2019 onwards, TOF cameras (infrared 3D cameras) are increasingly being installed in smartphones, so that 3D modeling based on smartphones has to be considered in a different light – especially with regard to SfM (Structure from Motion) and MVS (Multi View Stereo).