A study in colour - Part 4: Capturing colour

Houseboats in Victoria, BC.  Taken in 2020 with my Panasonic FZ2500 bridge camera.

How, exactly, do digital cameras capture colour?  We know that the digital sensor is the physical device that receives light in the form of an image from the lens, but how does that translate from photons to pixels?  The answer lies in photo cells arranged in something called a Bayer array.  See the image below.

A 20 mp sensor has 20 million pixels, each pixel is made of 4 wells that provide colour data.

The Bayer array is the backbone of the digital imaging industry.  Although there are other ways of getting coloured pixel information, this method is by far the most commonly used.  The sensor does not produce an image directly.  Rather, that job belongs to the camera's CPU, where the information is processed into whatever file type is selected.  Jpegs, as mentioned in a previous blog, use 8 bit per channel encoding, where a value from 0-255 is assigned for each of the RGB slots making up the colour for that individual pixel.  Raw files use considerably more memory because the information carried can be converted into 16 bit per channel information and saved as a tiff or a jpeg-2000 file.  

The Bayer array does suffer from a number of issues.  Moiré (pronounced more-ray) happens when repetitive patterns cause interference lines within the image; this is reduced by either using a low-pass filter in front of the sensor or by relying on large pixel numbers to reduce the likelihood of a problem.  Moiré doesn't tend to be seen in nature.  It appears when patterned fabrics or repetitive patterns appear.  The array will reproduce the information, but the array pattern and the pattern of the image may conflict.  You can see an example of this interference pattern below.

Moiré pattern on a shirt with a fine weave pattern.  You can see the interference issues.

 Another problem with digital sensors is the risk of bloom.  This happens when bright areas in an image cause electrons to overflow wells to neighbouring pixels.  This causes them to produce brighter values than what realistically exists.  I have noticed over the years that bloom is less of an issue than it used to be.  The size of the pixels and their proximity to each other are important factors in how much bloom occurs.  Small and densely packed pixels are at a greater risk of developing bloom.  You can see an example of bloom below (yes, some of the blooms have bloom).

Bloom may occur when areas of brightness are greater than the sensor's capacity to record.

There are other types of digital capturing systems that reduce the likelihood of Moiré.  I found a good site here.  It is worth a look if you are interested.

Thanks for reading.

Eric Svendsen     www.ericspix.com