Sometimes it is important to forget all the information overload of camera interfaces or computing power. Why? A successful vision system has additional requirements. It would be paradise if every machine vision needed no more than a camera, computing power and an application. The reality is much more complex. How do you precisely trigger camera and illumination? What is real-time related to your application? How do you communicate inside of the machine, the machine network and the IT network? How many years do you expect the delivery of the same hardware, firmware, driver, OS without any changes? Is your vision libary or vision tool perfectly integrated in the hardware infrastructure? These are a few questions needing an answer to design a professional vision system. Therefore forget all the camera interface discussions and CPU benchmarks for a moment.