In this talk, we will present how we can design cameras for mirror replacement by performing vision system simulation. To optimize the final shape and maximize the field of view of such a camera, the most crucial aspect is to be able to characterize and predict the impact of each component of the vision system on the overall performance of learning-based applications (computer/machine vision), as well as human vision. By simulating the image obtained through the entire imaging pipeline (protective layer, optics, sensor and image signal processor) we are able to study the impact of each isolated component before investing in the prototyping phase. Thanks to this capability, it is also possible to broaden the possibilities for the camera positioning in the car. Looking at the example of virtual e-mirrors we will highlight how we can simulate and validate the system’s component’s parameters to ensure accurate human vision and machine vision perception performance