It is the goal of this thesis to address some of the challenges in 3D scanning. This has been done with focus on direct in-ear scanning and on Multiple View Stereopsis. Seven papers have been produced over the course of the Ph.D., out of which, six have been included. Two papers concern volumetric segmentation based on Markov Random Fields. These have been formulated to address problems relating to noise ltering in direct in-ear scanning and Intracranial Volume estimation. Another two papers have been produced on the topic of recovering surface data based on a strong statistical prior. This was done in particular on scans of ear canals, but the methods are general. Finally, an experimental setup has been constructed, which has produced a large versatile data set. The data set has been used as the foundation for two papers on the evaluation of Multiple View Stereopsis. The data have a great potential to be used for advances in Multiple View Stereopsis, robust surface reconstruction and photorealistic modelling.