Image synthesis is strongly linked to handling computational complexity (large scenes, complex lighting simulations,…). Sophisticated solutions led to almost photo-realistic results. Yet, there is a strong focus on a pipeline considering a standard screen, a compute device, and a single user. This classic setting is about to change, with the advent of many-view rendering scenarios. Stereoscopic, virtual-reality, multi-view, and soon light-field and holographic displays (to provide correct depth perception) enter the mass market. Simultaneously, remote-rendering with large crowds operating low-performance devices, lead to structurally similar challenges. Both areas are potential billion-dollar markets and efficiency of the rendering systems will become a decisive cost factor, because these problems can no longer be handled by increasing compute power. This project proposes to rethink the established graphics pipeline by (i) restructuring computations (e.g., left-right views in stereo are similar, light fields even more so), (ii) reallocating resources to perceptually meaningful aspects, and (iii) building innovative tools for content developers to target many-view platforms. The project addresses three aspects with large potential for scientific and technological impact. First, visibility and level-of-detail are crucial to identify recurring computations and consequently reorder parallel computations in a distributed rendering system. A key is the analysis for visibility simplification with controlled precision. Secondly, the user will interpret the image, which makes it possible to reduce costs by understanding and exploiting perceptual limitations. Finally, the findings are integrated in novel design tools to build efficient content for multi-view platforms. My experience in large-scale rendering, perception, visibility and visualization will be key in this endeavour. From a scientific perspective, the project introduces new theoretical ideas with potential impact beyond the motivation application domain. Further, the results will have direct impact on commercially relevant application domains, such as entertainment, medicine, or geoscience, where many-view approaches are rapidly becoming common place.