The inherent flaw of Smartphone Cameras

Quality of the builtin camera has become a key selling point for smartphones these days. Samsung Galaxy K Zoom has gone to such extremes that they put a zoom lens on the phone blurring the lines of a standalone cam and camera phone. There are many who doesn’t want to buy the HTC one only because it has a 4mp cam. There are even people who bought the Nokia pureview lineup just for the camera. Today we are going to discuss an old camera tech that could sort out the low light photography problem  current phones face without adding any expensive image stabilization or even compromises on resolution.

Before explaining the tech itself, first let’s learn how an image sensor captures a picture. Image sensor is a matrix of photodiodes which can just measure the intensity of light falling on it. Which means it can only generate a black and white picture. To get a colour image, we die the individual photo diodes with a primary colour. There are several layouts in which the dies are arranged. The most common is Bayer pattern.

Here a red coloured photo diode measures the intensity of red light and the green measures green and blue does blue. The output from all 4 pixels is used to fix the colour of one pixel in the final picture. This process is called demosaicing. Two green colour pixels  are used because human eyes are more sensitive to the colour green. The problem with this pattern is, a primary coloured die allows the passage of the light of its own colour and absorbs the other two primary colours .

Let us assume that the incident colour has equal amounts of red, blue and green. Red die allows just 1/3^rd of the light to pass through it. Other portion is just wasted. 2/3^rd loss is a big number because the size of the individual photo diode is very small and every bit of light is important. (HTC chose 4mp resolution to increase the size of individual pixels without changing the overall size of the sensor.)

In the initial stages of digital photography, there was another pattern, it used cyan, yellow, green and magenta giving a CYGM pattern.The benefit of the layout is that the dies are secondary colours, they block only one primary colour, means they allow 2/3^rd of the incident light energy to pass through to the sensor. Only 1/3^rd of the light is absorbed. The increased sensitivity would allow even better low light photos without that much of physical change. If price is not barrier, we can further improve the pattern  by adding an image stabilisation and adding another white pixel like in the Motorola X ClearPixel.

             The concept is pretty simple, yet the reason the technology was forgotten is because of the way things work in an industry. Almost everyone used the bayer pattern and hence big money and research was spent on optimising demosaicing and colour correction algorithms for that particullar pattern. No companies wanted to use a technology that was not actively developed. This is very similar to the app dilemma new mobile OS’ face. (people wont buy a phone without apps, app developers wont make apps on a platform that doesn’t have enough users.)

Now with the recent advancements in image processing chips and more advanced algorithms, I think we can give a second chance to this long forgotten design. I wish I could see a pureview phone soon with CYGM pattern. Hope you enjoyed reading. Share your thoughts at the comment section below.

You can share this article with the buttons placed at the bottom left