The study of the mammalian visual cortex has been immensely influential on theories of perception and cognition, as well as on the field of machine learning. The mouse visual cortex in particular provides us with rich data on the anatomical, chemical, and genetic makeup of the visual cortex. Connecting the aforementioned theories to the biological data remains a challenge, which bottom-up detailed simulation methods may be able to address. We first reviewed the anatomical and physiological underpinnings of orientation selectivity and direction selectivity, key response properties. We then developed a mouse visual cortex model with an unprecedented level of detail in hopes of determining the neural underpinnings of phenomena of visual perception. Reproducing orientation selectivity in the detailed model proved to be more difficult than in simpler models, suggesting that previously hypothesized mechanisms are insufficient to explain this phenomenon in-vivo. In particular, our evidence points to homeostatic plasticity as the key missing mechanism. New approaches are needed to simulate this in large, detailed networks. To facilitate the development and comparison of differing approaches, we present a framework for the validation of computational models and a library of mouse visual cortex validations, which can be used to foster collaboration between computational approaches.