Depth-Image Based Rendering (DIBR) techniques enable the creation of virtual views from color and corresponding depth images. In stereoscopic 3D film making, the ability of DIBR to render views at arbitrary viewing positions allows adaption of a 3D scene’s depth budget to address physical depth limitations of the display and to optimize for visual viewing comfort. This rendering of stereoscopic videos requires the determination of optimal depth range adaptions, which typically depends on the scene content, the display system and the viewers’ experience. We show that this configuration problem can be modelled by a linear optimization problem that aims at maximizing the overall quality of experience (QoE) based on depth range adaption. Rules from literature are refined by data analysis and feature extraction based on datasets from film industry and a human visual attention model. We discuss our approach in terms of practical feasibility, generalizability w.r.t different content, subjective image quality, visual discomfort and depth quantity, and demonstrate its performance in a user study on publicly available and self-recorded datasets.