“Statistics are like a bikini. What they reveal is suggestive, but what they conceal is vital.”
The same can be said about specifications (specs) as well. In the technology industry, specs have always been the most widely used approach for product comparisons and as the above quote says, they are only suggestive. One of the most popular specs is the “Megapixel” to compare resolutions of digital cameras (DSLRs, point & shoot cameras, smartphone cameras etc).
What is “Megapixel”? Mega stands for million & pixel is the smallest unit of an image (like a dot). The following set of images demonstrates how individual pixels combine to form the final image.
A 50 pixel image consists of 50 individual tiny units. A 1 megapixel image consists of 1 million individual tiny units. When you pack more pixels into the same area, the image contains more details and becomes much sharper. Hence, one can assume that more the pixels, better is the image clarity and thats the reason megapixel has become the most advertised specification when it comes to photography or imaging.
With the advent of smartphones, the megapixel battle has become more intense than ever and almost every other phone company is trying to outdo each other by offering phones with cameras having higher megapixels. It might come as a surprise to find that an Android phone costing Rs 10K has a 13 megapixel camera while a DSLR from Canon which has 10 megapixel costs more than Rs 20K. Even among smartphones, the camera in iPhone4 is just 5 megapixel whereas a Chinese branded phone which costs 1/4th the price of iPhone has 12 megapixel camera!! In the following example, 2 images from different devices are compared:
Although both the images have the same resolution (5 Megapixel), the difference in quality of images is very much evident. It further confirms the age old saying “You get what you pay for” and suggests that megapixel might just be a myth afterall and the image quality depends on other factors.
Lets analyze this scientifically to find out if this megapixel phenomenon is just a propaganda used by marketing teams and in the process lets find out the factors which play a role in image quality.
The concept of a camera in its simplistic form consists of a lens and a sensor. The lens is used to converge light and the sensor is used to capture this converged light. Thats it!! This is the only core functionality of a camera i.e Converge light & Capture light. With this concept in mind, lets find out how these components inter work.
If you recollect from your high school science classes, light is nothing but actually photons. When light reflected from objects of different shapes and colors pass as rays through the lenses and fall on the “Sensor”, they are captured, converted into electrical pulses and formed into an image.
The sensor is well within the camera cavity and not exposed. When you click the shutter button, the shutter opens for a short time, allows photons to fall on the sensor and immediately shuts back again. So when someone says that the photo was shot with a 1/200 shutter speed, it means, the sensor was exposed to the incoming photons for 1/200th of a second.
The sensor on the camera consists of millions of tiny units which actually are mapped into pixels of the image. The role of each of this tiny unit is to capture photons and convert to electrical signal forming 1 pixel. If there are 1 million of such tiny units each producing its pixel, we call it a 1 megapixel camera producing 1 megapixel image and so on. The size of an image depends on how many pixels (and correspondingly how many of those tiny units) we have. But most importantly, the quality and strength of the image depends on how much of photons can be captured by each of this tiny unit. (For those who know some electronics concepts, by strength of the image, I mean the intensity & signal to noise ratio).
Before going further, here is an interesting illustration to help us understand better. Lets consider an island inhabited by 100 people. Assume that the govt drops food from helicopters at a constant rate and at a constant density everyday for 1 hour which totals upto 100 kg of food per day. This way, each person in the island receives 1 kg of food which is sufficient to remain healthy & strong and keep the island prosperous.
Now lets assume that the same island is inhabited by 500 people and everything else remains same. i.e The govt drops food from helicopters at a constant rate and at a constant density everyday for 1 hour which totals upto 100 kg of food per day. This way, each person in the island receives only 200 grams of food which is insufficient and thus people in the island are weak & unhealthy.
With this illustration in mind, lets now come back to the actual topic and try to correlate elements of the above example. The island in the above example can be correlated to a sensor, each person on the island to a pixel on the sensor and the food packets to light photons. Irrespective of the size of the sensor or the number of pixels it has, the rate at which photons fall on the sensor is constant. Just like the above island example, if a sensor is densely packed with pixels, it leads to starvation.
Lets go a little more in depth now. Assuming that a 1 sq inch sensor has 1 megapixel (1 million pixels) and it receives 10 million light photons, it means each pixel receives 10 photons i.e 10 signal units. This is the signal component because it carries meaningful information. In digital systems, there is always noise introduced at every level which is independent of the incoming signal and depends on factors like interference, power supply, quality of components etc. Assuming that every pixel of the sensor introduces 2 noise units. Thus, we have 10 signal units & 2 noise units and the Signal to Noise ratio is 10/2 = 5.
Next, lets assume that the same 1 sq inch sensor is now densely packed with 5 megapixels (5 million pixels). Since the incoming light photons is always constant (10 million photons), it leads to each pixel receiving 2 photons. As discussed above, assuming that every pixel of the sensor introduces 2 noise units, we have Signal to Noise ratio of 2/2 = 1, which is much lesser than the previous example, causing a distorted, noisy & blur image. In fact, when we pack more pixels in the same sensor, the increase in pixel density actually creates more interference, leading to more noise.
Coming back to the original topic of smartphone camera comparisons, the thumb-rule is that an increase in megapixel must be accompanied with an increase in sensor size, failing which the sensor pixel size decreases and image quality suffers. But an increase in sensor size is highly challenging because it increases the size of the camera unit (difficult to fit it into a phone) and drastically increases the cost.
The following pic gives an idea of the sophistication & sizes of different smartphone camera sensors:
Apart from sensor size, the quality of lenses (optics) and the number of intermediate lenses also plays a very important role in image quality. The following pic dissects the lens unit of a popular smartphone:
Also, factors like pixel size, pixel density, aperture, type of sensor, sensor illumination, noise filters etc play equally important roles. But marketing departments conveniently ignore these factors and choose to highlight only the megapixel as the specification which is highly misleading and in a way has become a propaganda.
Having said that, the industry is gradually changing and prioritizing on most of the above mentioned factors and highlighting them as well. For example, Apple has begun to specify most of these factors during phone launches and other companies are also following suit. The following screenshot from an Apple Keynote highlights pixel size & the type of sensor (Backside illuminated sensor, which is considered superior).
Here is a screenshot from another Apple keynote where the aperture is highlighted (along with increase in sensor size)
Although pixel size, sensor size & aperture are better indicators (and quantifiable factors) of camera quality, they still do not describe the camera’s capabilities perfectly because factors like sensor quality, lens (aperture) quality, image stabilization cannot be quantified and one has to personally check the end results (or rely on professional reviews or just trust the brand) to come up with any conclusions.
To receive notification of insightful articles in future, consider subscribing to this portal by entering your email id (and confirming it from your mailbox)