At the IEEE International Electron Devices Meeting in San Francisco, camera and sensor manufactures are beginning to solve the ever growing problem: As resolution increases, how do you store all of that data? Especially at high speeds?
A majority of CMOS technology uses a rolling shutter (a issue we all come across when panning or shooting an object at high speed) due to the fact the sensor is read out line by line. Global shutters read the entire sensor image out in one cycle.
To get around this, CMOS chips can have a memory unit assigned to each photosite on the sensor, resulting in a CMOS chip with a global shutter. Many camera systems out now provide the user the ability to switch between rolling and global shutter. The global shutter usually results in a compromise of frame rate options, or recording length. Example being a Phantom Flex 4K, which while shooting extremely high frame rate video, you are limited to recording time based on your storage, selected frame rate, resolution, and bit depth.
Researchers have presented two ways to improve the memory usage in laboratory situations that result in longer high speed record times. What is more interesting to video production professionals is a development from Canon, who have goals to improve the image quality of “normal” frame rate readouts (30-120 fps).
Canon presented what they call an “engineering trick” that is allowing them to read the data from each photosite 4x per frame, improving the dynamic range of the image in Global Shutter mode, illustrated below.
All of the sensor developments don’t mean anything right now, but they will trickle down from lab cameras into production soon enough. And the frame rates you enjoy now will only get faster, with better dynamic range.