Image sensors explained

Photography has the magical capacity to preserve a moment in time. Key to this is the image sensor at the heart of every digital camera. Just as the retina in the human eye captures light and translates it into nerve impulses that the brain can interpret, the sensor captures light and converts it into an electrical signal that is then processed to form a digital image.

Here, we take a look at how image sensors work, and explore the different types of image sensors used in Canon cameras.

• Digital imaging basics
• CCD sensors
• CMOS sensors
• Developments in CMOS sensors
• DGO sensors
• The SPAD sensor
• Sensor sizes explained

With all types of sensors, the imaging process begins when light passes through the camera's lens and strikes the sensor. The sensor contains millions of light receptors or photosites, which convert the light energy into an electrical charge. The magnitude of the charge is proportional to the intensity of the light – the more light that hits a particular photosite, the stronger the electrical charge it produces. (SPAD sensors work a little differently – more on this later.)

In order to capture colours as well as brightness information, photosites are fitted with red, green and blue colour filters. This means some photosites record the intensity of red light, some the intensity of green, and some the intensity of blue.

The electrical signals from all the photosites in the sensor are passed to the camera's image processor, which interprets all this information and determines the colour and brightness values of all the individual pixels (picture elements) that make up a digital image.

If you're shooting RAW, this data is saved, along with information about the camera settings, in a RAW file. If the camera is set to save images in any other file format – JPEG, HEIF or RAW+JPEG – then further processing takes place in-camera, which typically includes white balance adjustment, sharpening and noise reduction, among other processes, depending on the camera settings. It will also include demosaicing or debayering, which cleverly calculates the correct RGB colour value for each pixel (each individual photosite, remember, records only one colour – red, green or blue). The end result is a complete colour digital image – although, in truth, if the image is a JPEG, more of the original information captured by the sensor has been discarded than has been kept.

You conventionally hear about the number of megapixels (millions of pixels) in a sensor, but strictly speaking the sensor does not have pixels at all, but sensels (distinct photosites). What's more, there is not a one-to-one correspondence between sensels in the sensor and pixels in the resulting digital image, for a whole range of technical reasons. It is more accurate to describe a sensor as having a certain number of "effective pixels", which simply means that the camera produces images or videos of that number of megapixels. The Canon PowerShot V10, for example, has a sensor described as approximately 20.9MP in "total pixels" but some of the sensor data is used for technical processes such as distortion correction and digital image stabilisation, with the result that the PowerShot V10 delivers video (with Movie Digital IS) at approximately 13.1MP and still images (which undergo different processes) at approximately 15.2MP.

There are several different types of image sensor. Digital photography arrived in the mid-1980s with the introduction of CCD (Charge-Coupled Device) sensors. These sensors were the first to make it possible to capture images without the use of film, revolutionising photography.

CCD sensors are composed of an integrated grid of semiconductor capacitors capable of holding an electrical charge. When light reaches the sensor, these capacitors, acting as individual photosites, absorb the light and convert it into an electrical charge. The amount of charge at each photosite is directly proportional to the intensity of the light that strikes it.

In a CCD sensor, the charge from each photosite is transferred through the sensor's grid (hence the term charge-coupled) and read at one corner of the array, in the same way that water might be passed along a bucket brigade or human chain. This method ensures a high degree of image quality and uniformity because each pixel uses the same pathway to output its signal. For this reason, Canon's first professional digital camera, the EOS-1D, launched in 2001, had a 4.15MP CCD sensor. However, this process is also more power-intensive than the process in CMOS sensors.

In 2000, Canon introduced its first CMOS (Complementary Metal Oxide Semiconductor) sensor, in the 3.1MP EOS D30. Unlike the CCD sensor, which transfers charges across the sensor to a single output node, a CMOS sensor contains multiple transistors at each photosite, enabling the charge to be processed directly at the site. This has several implications.

For a start, CMOS sensors require less power, making them more energy efficient. They can also read off electrical charges at a much faster rate, which is crucial for shooting high-speed sequences. What's more, CMOS sensors share the same basic structure as computer microprocessors, which allows for mass production at a lower cost while incorporating additional functions such as noise reduction and image processing right on the sensor.

All of Canon's current PowerShot, EOS and Cinema EOS camera ranges feature CMOS sensors, including the mirrorless EOS R System line.

CMOS sensor technology has continued to evolve. An innovation developed by Canon is Dual Pixel CMOS AF technology, which enables each pixel on the sensor to be used for both imaging and autofocus, resulting in faster and more accurate AF performance.

Another development in Canon's CMOS technology is the stacked, back-illuminated sensor used in the EOS R3. This design places the photodiodes above the transistor layer to improve light collection efficiency, resulting in less image noise and better image quality. Additionally, the stacked structure allows faster data readout, contributing to the camera's high-speed performance. This technology enables the EOS R3 to meet the demands of both high-end video production and high-resolution photography.

Canon's CMOS sensor research and development is ongoing. One recent result of this is an ultra high sensitivity 35mm full-frame CMOS sensor, with much larger photo receptors (approximately 7.5 times the size of those in previous sensors). Larger photo receptors are able to capture more light, in this case achieving a sensitivity equivalent to ISO 4 million, enabling a camera to capture vivid colour images of very dark environments. This technology is used in the Canon ME20F-SH ultra low light video camera.

Canon has also developed an ultra high pixel count sensor, using advanced miniaturisation techniques to reduce the photosite size. This facilitates very high resolution image capture, with a pixel count up to 250MP. In an image captured using this technology, it is possible to distinguish the lettering on an aircraft in flight 18km away and achieve a resolution approximately 30 times higher than that of 4K video. This has great potential for applications in surveillance, astronomical observation and medical imaging.

One shortcoming of current CMOS sensors is that, for technical reasons including data bandwidth, their data is read out sequentially rather than all at once. This results in issues such as "rolling shutter" distortion of fast-moving subjects that have changed their position during the time the frame is being read out. The advanced CMOS sensor in the EOS R3 enables much faster readout speeds, greatly alleviating this issue, and Canon is actively investigating other solutions such as "global shutter" technology, which enables readout of the entire sensor in one go, but this technology is very complex, adds both image noise and cost, and can't yet produce very high-quality outputs.

The DGO (Dual Gain Output) sensor is an advanced image sensor used in the Canon EOS C300 Mark III and EOS C70 professional video cameras.

Canon’s DGO sensor works by reading each pixel at two different amplification levels, one high and one low, and then combining these two readouts into a single image. The high amplification readout is optimised to capture fine details in shadow regions while reducing noise. The low amplification readout is designed to maintain and accurately reproduce information in the highlights. Combining these produces an image that has a broader dynamic range, retains more detail and exhibits less noise compared to images from conventional sensor technologies.

The DGO technology does not consume any more power than a conventional sensor, and is also compatible with Canon's Dual Pixel CMOS AF system and electronic image stabilisation, delivering fast, reliable autofocus and a super-steady image.

CCD and CMOS sensors measure the intensity of light – in other words, how many photons reach the sensor within a specified time. SPAD (Single Photon Avalanche Diode) sensors work differently, using the "avalanche" effect in semiconductors. When a photon strikes the sensor, it generates an electron, which then triggers a chain reaction or "avalanche" of electron production. This cascading effect causes a large current to flow instantaneously, which is read out as a voltage signal in the form of a train of pulses corresponding to individual photons.

This unique light-sensing technology means SPAD sensors can achieve incredible low-light performance. Using the outstanding SPAD sensor, Canon has developed the MS-500, a breakthrough interchangeable-lens camera capable of capturing high-definition colour footage in extremely low-light conditions, even the near-total darkness of a night-time environment.

In addition, the MS-500's bayonet mount for a 2/3-inch broadcast lens enables the camera to utilise Canon's extensive range of broadcast lenses, with their excellent super-telephoto optical performance. This means the camera is able to resolve subjects several kilometres away, even if they are unlit, making it an invaluable asset for security, surveillance and a broad range of scientific applications.

It's clear that a sensor's megapixel count (whether it's total or effective pixels) isn't the whole story. The physical size of the sensor is an important factor. APS-C sensors are physically smaller than full-frame sensors, which means that even if the pixel counts are identical, a camera with a full-frame sensor should deliver a wider dynamic range and better low-light performance – if it has the same megapixel count but over a larger area, then it has larger photosites, which will be capable of capturing more light. This makes full-frame cameras such as the EOS R3 and EOS R5 a favourite choice for professionals, particularly those shooting landscapes, architecture or portraits.

Conversely, because APS-C sensors are smaller, your subject will fill more of the frame than it would if you used the same lens with the same settings on a full-frame camera – so in effect, an APS-C sensor increases the reach of your lens. In Canon cameras, the "crop factor" is approximately 1.6x, giving you an effective focal length 1.6x greater than the same lens on a full-frame camera. This gives a 50mm lens, for example, the field of view of an 80mm lens (50 x 1.6 = 80). This means APS-C cameras are well suited for a broad range of uses including wildlife and street photography. In addition, thanks to the smaller sensor, APS-C cameras such as the EOS R50 and EOS R10 are smaller and lighter than their full-frame counterparts, making them a great option for travel or nature shoots.

Some video cameras use Super 35mm sensors (active area approximately 24.6 x 13.8mm, depending on the resolution setting), which are slightly larger than APS-C (22.2 x 14.8mm) but still less than half the area of full-frame (36 x 24mm). They are widely used in the film industry thanks to their balance between cost, image quality and cinematic look (with a shallow depth of field). Camcorders and other camera types use a range of other sensor sizes, such as the 20.1MP 1.0-type stacked CMOS sensor in the compact PowerShot G7 X Mark III and the 11.7MP 1/2.3 CMOS sensor in the PowerShot PX.

The choice of sensor size depends largely on your shooting requirements and budget. Each sensor size offers distinct advantages, and understanding these can help you select the right camera for your specific needs. However, you can see why standardising on "effective pixels" provides a simpler measure for comparing different cameras and different technologies!