If you tell your friend that you bought a digital camera, his first question will be “How many megapixels does it have”? Or, if someone is buying a digital camera and he is seeking your advice the most probable question he will ask is “How many megapixels should my digital camera have?” This illustrates the importance given to this very important specification of any digital camera – the number of megapixels it has. However, is the importance given to this figure justified? What is the impact of number of pixels on the image quality, real or perceived, and if real, to what extent? How does this affect the ultimate form of any image which is the print? This article will look into this aspect plus others such as the effect of pixel size, sensor size, resolution, etc. Before we can discuss in detail about all these issues, you need to get to know the terms pixel, pixel count, resolution and more importantly the difference between the last two.
Pixel Count: The sensor is a device placed in the camera body, and it is here that the light rays from the lens focus. It is made up of light gathering (photosensitive) elements. These elements produce an electric charge proportional to the intensity of the light falling on each element. This means more the light, the greater will be the electrical energy. The circuitry on board the camera measures this energy and hence knows the intensity of light at each photo element. By putting together measurements taken for each element, the on board computer is able to construct the entire image. Each dot of the image thus formed corresponds to one element. Each of these elements is called a Pixel which is the short form for “picture element”. The number of pixels that are present in a sensor is called the pixel count and is normally specified as Mega (or millions of) pixels (abbreviated as MP). Pixels are arranged as a rectangular grid in columns and rows. Hence, it is customary to specify them that way. That is, a 12 MP sensor will have 4300 columns and 2800 rows for example.
Resolution: So, if this is what pixel count is, what is resolution? Resolution is the resolving power and is measured as number of lines per millimeter or inch. That is, it basically tells the detail that a camera can capture. Resolution depends on the number of pixels, lighting conditions, the type of test target, the resolving power of the lens and so on. While this is an important measure there are so many variables involved as mentioned that no camera manufacturer mentions resolution. It is more often used to measure the resolving power of a lens. Another place where the term ‘resolution’ is used is when you print.
Unfortunately these two terms – pixel count and resolution are used interchangeably by many, even though international standards specify that it should not be so used in digital imaging. Many websites say the same thing and even some books say so. But this is not correct. The “pixels” that a digital camera has (for example 12 MP) is called the pixel count and this is not the resolution. If you have any doubts refer to a digital camera specification sheet from a standard camera manufacturer like Canon or Nikon and see what is printed in them.
On the other hand you need to understand the significance of the word resolution when the captured image is printed (or shown on a monitor).
A print is composed of a number of dots. Each dot corresponds to a pixel that has been captured by your camera. As you can expect for a given print size, higher number of dots and smaller sized dots will allow you to see finer detail in a print. So how many dots should be there in a print and what should be size of each dot? To have a norm regardless of the print size this requirement is specified as the number of dots needed per inch. This is abbreviated as dpi and is known as the print resolution. The generally accepted print resolution is 300 dpi. It is not necessary to have resolution beyond this as the human eye cannot resolve beyond this number.
The number of pixels in your camera has a direct bearing on the size of the print you can make. The following example will make this clear. If you take a 12 MP image of 4300 X 2800 pixels and print at 300 dpi, it will yield a print of approximately 14.5 X 9.5 inches. This is called printing at the “native resolution”.
If you want to print beyond this size, say 30 X 20 in with the same number of pixels what one needs to do is to distribute the same number of pixels over a larger area. This will automatically make the dots proportionally larger and will bring down the number of dots per inch. In other words the print resolution reduces. As an example a 12 MP image printed at 30 X 20 in will have a resolution of only 140 dpi. There will be a corresponding loss in detail (sharpness) of the print.
So what is that you need to do to make very large prints? There are four options available.
Option 1: This is the simplest option – do nothing! Just make a print at lower resolution (for a large print a resolution of 250 dpi is acceptable). This is because as the print size increases you will start looking at the print from a longer distance. As the distance grows the eye will not be able to resolve that much. This is the reason bill boards that look fine at large distances will look really fuzzy if you go near them!
Option 2: This is the most popular option to print beyond the native resolution. This is called up-sizing (or sometimes “ressing up”) through a process called interpolation. In this process the number of pixels is increased beyond what has been captured by the sensor. Thus you can print beyond the native resolution. Interpolation works as follows. Remember that the pixels in a sensor are organized as a rectangular grid, perfectly in rows and columns. Due to this very regular structure it is mathematically possible to find out and insert pixels between the existing pixels, thus increasing the total number of pixels. Most standard image processing software packages (like Photoshop) have facilities to support interpolation.
Interpolation sounds like a magic bullet and that is true to a certain extent. However there is a limit to what extent you can increase the file size. You can safely (based on my personal experience) increase the total number of pixels by a factor of little more than four. So if you start with a 12 MP camera which gives a print of only 14.5 X 9.5 in at native resolution, you can go up to 54 MP which will give a print of 30 X 20 inches at 300 dpi.
Note: Sometimes you may want to make the image smaller too. This is done when you want to display the images on a monitor, or to upload sites like Facebook, Flickr, etc. This operation is called down-sizing. The common name for up-sizing and down-sizing as you would expect is re-sizing.
Option 3: This is the best and most expensive solution! Simply use a camera that has a higher pixel count. If you use a 24 MP camera then you can make a 300 dpi print of 20 X 13.5 in at the native resolution itself. This is a clear case where higher resolution helps. In fact the main advantage of cameras with higher resolutions is their ability to make large size prints yet retain detail.
Option 4: There is a way you can get more megapixels from your existing camera without spending anything. That may sound incredible but it is not, provided your subject is suitable. The technique is similar to that of panorama making. Switch to a longer focal length (so only a smaller area is covered) and take overlapped images and stitch them together to cover the full area. If you do not change your position then perspective will not change. So, the stitched image will look the same as that taken with a shorter focal length lens but it will have lot more pixels since you have now used multiple images. There are now devices like Gigapan that perform this task automatically. The subject needs to be static so this technique is not suitable in all situations. However, you can use it for landscapes and still life where high resolutions are needed.
Table showing different print sizes at various dpi and corresponding pixel count needed:
|Print Size in inches (approx.)||dpi||MP||Horizontal Pixels||Vertical Pixels|
|10 X 7||300||6||3000||2000|
|14 X 9||300||12||4300||2800|
|20 X 13||300||24||6000||4000|
|24 X 16||250||24||6000||4000|
|25 X 16||300||36||7350||4900|
|30 X 20||250||36||7350||4900|
|30 X 20||300||54||9000||6000|
Note: The computer monitors as of now are manufactured to a resolution of around 100 dpi (or less). Thus the demands from an image are higher if you want to print it rather than just display it. Also a HD TV has about 2MP.
Pros and cons of high pixel counts:
Apart from getting large prints, there is also another advantage with high pixel counts. This is in cropping. That is, many times you take a captured image and discard a part of it for compositional reasons. Cropping can also be useful for wildlife and bird photographers as they frequently cannot go close to the subject. The result would be an image where the subject is small. By cropping you will be able to make the subject larger. For example, if you start with a 200 mm lens, you can get the effect of a 280 mm lens by cropping with a factor of 1.4. In this process however you will be discarding about 50% of the pixels – not a very good solution but if you are starting with very high number of pixels and if the pixels are of very high quality then you will still be able to get a good print. As an example, if you start with 24 MP you can still get 12 MP after 1.4 X cropping – good enough for a nice big print.
Before you go for a very high pixel camera (they cost serious money), ask yourself how large you want to print and how often. Most of us do not print larger than 18 X 12 in. You can easily print this size with an 8 MP camera (with interpolation). The money instead can be better spent for more useful accessories like lenses, etc. Also, to exploit such very high pixel counts your lenses must be top notch. Very high pixels counts can be counterproductive too. They produce very large computer files which are difficult to edit and store. More importantly the pixel size reduces for a given sensor size as the count increases. As you will see shortly, smaller pixel sizes produce more digital noise and have less dynamic range, reducing the overall quality of the image.
Pixel Count and Resolving power: How are these two linked? I leave out the mathematics and give you the gist. For resolution to double, pixel count has to quadruple (become 4 times). In practice it has to more than quadruple since several factors like the resolving power of lens, etc., also have to be taken into account. So, if the next D-SLR model from your favorite manufacturer has just a couple of mega pixels more don’t spring for it just for those extra pixels as you will get only minimum increase in resolution.
Pros and Cons of a large sensor:
Noise characteristics of the pixels and dynamic range are the other two features of the sensor that are important for picture quality. To understand these you need to look at the impact of the pixel size on these two features.
Larger pixels have better light gathering capability simply because they have more area to collect light. More light means a stronger signal. Hence, the unwanted signal (which is the noise) is less compared to the signal that is generated by light which is what we want. This in turn means that you can get much better high ISO performance. This feature is invaluable for photographers who do a lot of photography in low light. It is also useful for sports and other action photographers since high ISO allows use of higher shutter speeds for the same light. Larger pixels give a higher dynamic range too – that is they capture a greater brightness range – from darkest shadows to brightest highlights.
If large pixels give better noise and dynamic range, why not go for larger pixels always? As the size of each pixel increase, it does not require a mathematician to guess that for the same number of pixels the size of the sensor also increases. Unfortunately, as the size increases, the cost of the sensor very rapidly increases. Not only that, the camera will now need a larger mirror, prism and shutter to complete the picture. To add to this, lenses have to be bigger to cover the larger sensor and hence will be more expensive too! All this put together makes two systems, both of which are otherwise very similar, to differ quite a lot in price if the sensor sizes are different.
Sensor size and printing: Once an image is captured, the size of the sensor has no bearing on the printing. In fact the whole printing process is totally agnostic to the sensor size. This is quite unlike a print made from a film where the size of the negative plays a significant role on the size of the print made (for a given quality). You have seen that if you take a 12 MP and print at 300 dpi, it will yield a print of approximately (at native resolution) 14.5 X 9.5 in. This is regardless of the size of the sensor used. The prints too will be theoretically indistinguishable if pixels are of equal quality. However, this will not be the case in practice as a print made from a camera with larger pixels, will be of superior quality. This qualitative difference will be even more apparent in situations where:
High ISOs have been used
Light is low
Scene had high contrast (that is scenes with high dynamic range)
How many pixels does film have? This question is likely to open a can of worms and the Editors are sure to receive letters from both film and digital aficionados contesting what I am going to say. To make the issue as non-controversial as possible, I will resort to some mathematics with the hope that numbers cannot be disputed. To make this easier to understand let us assume that our film frame is a standard 35 mm frame of 1.5 X 1 inches (or 36 X 24 mm). It is generally accepted that if you scan at about 4000 dpi you can capture all the information from even the finest grain film. If this is the case, the scan of 1.5 X 1 inch frame will yield (approximately) 6000 X 4000 pixels or 24 MP. This number circulates a lot everywhere and is correct.
The wrong notion is the one that comes out as a deduction from it – that is, you need a 24 MP camera to produce the same quality of image of a 35 mm. Sorry, film aficionados but you don’t need that many megapixels to produce an image of equal quality. This is because of the interpolation that was explained earlier. A 24 MP scan (6000 X 4000 pixels) will yield a print of 20 X 13 in (at 300 dpi), something that an 8 MP D-SLR can produce with just moderate interpolation! Today a top of the line 22 MP+ D-SLR can produce prints with the quality of a medium format film camera as per some experts and tests.
So, the next question that comes to you is that, if interpolation is doing the trick for images captured by digital cameras, why not start with a scanned image and interpolate beyond the 24 MP to get an even bigger print?. Unfortunately, this does not work since the scan is of a film which is composed of random grains (no regular structure) and hence is not amenable to interpolation. Scanning beyond 4000 dpi is not productive either since there is not much to capture. The only solution for getting a bigger scanned file is – you have guessed it – to start with a bigger film format like medium or large format.
To conclude, pixel count, beyond any doubt, is one of the most important parameters but not the only one that determines the image quality as many would believe. Its greatest impact is one on the size of the print you make. In fact the importance of pixel count has been over-hyped leading some educated critics calling the race to add more pixels as “megapixel madness”!
I hope you have now got a better understanding what pixel count, resolution, etc. mean and their importance.
All text and images © Ashok Kandimalla
The author can be reached at email@example.com