CMOS sensors
In fact CMOS now exceeds the performance of CCD in many aspects with the latest sensors now delivering sensitivity and dynamic range exceeding that of the traditional interline CCD. This trend is ratified by Sony's announcement of significant investment in new manufacturing plants for CMOS sensors and the closure of the CCD plant mentioned earlier.
The fundamental difference compared to CCDs is the conversion from charge to voltage within the pixel. Each pixel is typically addressable on a row and column basis ‒ the voltages being read out in parallel rather than the sequential method used in CCDs, with a subsequent increase in frame rate and other benefits such as user-definable regions of interest (ROIs). Many basic CMOS sensors have three transistors (3T) within each pixel. More transistors can be added (4T, 5T and 8T) for increased functionality such as noise reduction and global shuttering, but this comes at the expense of the well capacity and dynamic range as the additional transistors reduce the space in the pixel that is light sensitive. It is very difficult to make a number of amplifiers that behave uniformly, and as each pixel has its own, this leads to much greater non-uniformity than is common in CCDs. This non-uniformity is constant however, often being referred to as fixed pattern noise, which can be compensated by using flatline (field) correction.
Two further subcategories of CMOS sensors exist: on-chip and offchip A/D conversion. Whilst CMOS sensors can be made with less expensive processes and with greater integration, they are not always less expensive than comparable CCDs. One exception to this is any sensor that is used mainly in consumer electronics such as cameras and mobile phones, which can also be used in industrial cameras, thus benefiting from economies of scale.