| Home Alphabetical |
About Pixels, Definition and Resolution |
This page explains basic concepts about dimensions of digital images.
Before you start, forget what you thought the word "resolution" meant because with 99% probability you have the wrong concept in mind.
A digital image is a rectangle of pixels. Look at this small yellow arrow on a green background:
if we enlarge it, we will see something like this:
Here we can see how the image is built from little squares, called "pixels" or "picture elements". Each pixel has a colour.
In our example, most pixels are green, some have shades of yellow, a few on the yellow/green boundaries have colours in between.
Our image is 26 pixels wide and 16 pixels high:
The definition of the arrow image is 26x16. There are no units: definition is purely about the number of elements that compose the image.
The word definition may be familiar from such expressions as high-definition TV (HDTV). There too it expresses the number of elements: HDTV has more lines to build up the image than normal TV does.
The higher the definition the more pixels are used and therefore the bigger the disk space needed to store the file of the image.
Look at these two images:
 A |
B |
The image contains information: flowers, trees, letterboxes, other objects, sometimes even text. The left one, A, has one pixel for each pixel on your screen. The right one, B, has a hundred times fewer pixels to represent the same scene.
The definition of A is 300x220; that of B is 30x22.
B looks worse because there are fewer pixels to hold the information representing the scene.
But both images cover the same view. B just has 100 times less information: 660 pixels against 66'000 pixels for the left one. Its pixels have been made bigger than the pixels of A, so B can cover the same area.
For an identical subject, more pixels is better. One also says that the left picture has a higher definition than the right one.
Let's look again at the two pictures:
A |
C |
In this figure, the same two images are shown using one pixel of your screen for one pixel in the image. We know that C, which is just a scaled-down version of B, has less detail, but it is also smaller than A.
They are exactly the same images as in the previous figure: A is 300x220 and C is 30x22. Remember that C has less information about the scene (a hundred times less: you would need a hundred small pictures C to cover A, 10x10). C is smaller in area because in this image the pixels of C are the same size as those of A.
Resolution is the number of pixels per mm. Look at these three images:
A |
B |
C |
At the left, we see our familiar 300x220 image A, using one pixel of your screen for each of its pixels. It is shown in the same resolution as your screen. The middle picture B is the 30x22 one, it is actually identical to C but its pixels were enlarged to give it the same apparent size as A: it is represented in ten times lower resolution (since there are only 30 pixels to span the same number of mm as for A). The picture C on the right is the same picture as B, but represented in the same resolution as A: one image pixel per screen pixel.
The resolution of C is the same as that of A, the definition of C is the same as that of B.
The number of pixels per mm of C is the same as the number of pixels per mm of A; the number of pixels of C is the same as the number of pixels of B.
There is a great confusion about definition and resolution: screen vendors use the word resolution to give the definition of their screens.
To compute the actual resolution of your screen:
1. Measure the width W of the area that is lit up:
Mine is 366mm wide.
2. Go to the control panel ("Monitors" on Macs, "Displays" on Windows) and find the "resolution", which is actually the definition: it will be something like 640x480 (very old!), 800x600 (old!), 1024x768 (common, getting old), 1252x888, 1600x1024, or 1920x1080 (full HD, you are lucky).
This is the number of pixels your screen presents to you in the viewing area that you just measured in point 1.
3. Compute. The resolution is the number of pixels over the width. My screen is 1680x1050; so it has 1680 pixels over its width of 366mm. Its resolution is therefore 1680/366 = 4.59 pixels per mm.
Today common screen resolutions are around 4 pixels per mm or slightly higher. In the beginning of the 1980s, resolutions were at 2.8 or 3 pixels per mm. So while computers have become hundreds of times faster with disks that are thousands of times bigger, the quality of our screens has not even doubled! The resolution of the first laser printers was 12 pixels per mm, today printers and scanners go up to 60 pixels per mm, but our screens are still at 4.
If you have a large screen but you are using few pixels as a setting, then your resolution will be lower: you have fewer pixels per mm. The same image on your screen will be bigger, but it will not contain more information, the pixels are just larger. Therefore, the size of the screen is not the only measure of its quality.
CRT screens (the bulky kind that gets warm) are TV tubes. They can easily be set to higher or lower resolutions with more or less good effect. But CRT screens have now all but disappeared and have been replaced by flat-panel LCD screens. LCD screens give much sharper images when used at their native resolution and hence they are not good when used at other resolution.
Because LCD screens must be used at the number of pixels they actually have, the most important quality of an LCD screen is its number of pixels, not its actual size.
Shops advertise LCD screens by size, which is wrong. Often the sales people do not even know the number of pixels.
The second most important quality of an LCD screen is its display technology: does it have a digital input (DVI or Digital Video Interface) or can it be driven only by old-style VGA scanning? VGA (and XVGA etc.) does not deliver the perfectly stable pictures that DVI does. For DVI you also need a DVI output from your computer, but most now have that.
The next important quality is size: go for the largest you can afford. Very small pixels mean all text will be very small unless enlarged by some tricks.
Suppose I take a picture of a scene using a normal (argentic) film in a 35mm camera. I will get a negative or a slide of 36mmx24mm. Scanning the film to convert it to a digital image will give me a grid of pixels. That grid has a certain definition. If I scan a 36mmx24mm film frame at a resolution of 1200 pixels per inch (47 per mm) I will get an image that has 1700x1133 pixels. If I scan at a higher resolution, I will capture more detail in my digital image, but I will also have more pixels to hold that information and the file on disk will be larger.
If I keep scanning at higher resolutions, there will be a point at which I will start to notice the grain of the film. This will happen earlier with high-speed film (400 ASA and above) than if I use low-speed film (64 ASA and below). There is generally no use in obtaining more than 4000x3000 pixels from 35mm film. There is either film grain, or the limits of the optical quality of the camera will have been reached: the photo is not that sharp.
The same is true of scanning paper prints with flatbed scanners.
But whatever I do, I end up with a computer disk file whereby the only important measure I have is the definition: the number of pixels. Digital images do not have dimensions in mm!
I have a digital photo at reasonable definition, say 1200x900 pixels. The graphic designer at a newspaper now asks me to send a copy at 300dpi (300 dots-per-inch or 300 pixels per inch or 12 pixels per mm, a common requirement to obtain quality prints).
What does this mean? The request is without meaning unless they tell me also how wide the printed picture should be (in mm).
If they print my image at the requested 12 pixels per mm, they are going to get a printout that is 100mm wide, only 10 cm. Is that what they want?
What if they want the photo to fill the entire width of an A3 page in a tabloid? That is almost 30cm wide, I have only one third of the pixels to cover that if they want 12 pixels per mm!
What if they want to use it for a small illustration of only 5cm wide? Then I have twice the number of pixels they need.
If the graphic designer does not give you the width at which he/she wants to print, ask for it.
Once you know what is wanted, you can then perhaps transform the image.
Suppose your graphic designer wants to print the image 15cm wide at 12 pixels/mm and your digital image is 1200x900.
You need to supply 150mmx12pixels/mm =1800 pixels and you have only 1200.
First, it is always good to have the "original" at hand: if that is a slide or negative, you can always try to scan it at higher resolution if your digital copy has not enough pixels to satisfy the designer.
But if the image was made with a digital camera, then it is the original. There is no way to obtain a better picture at a higher definition.
Fortunately there are imaging tools to help a little bit: they can add the extra pixels by filling them in. They spread the information in the original over more pixels. This will satisfy the graphic designer, but it will of course not give a more detailed image: you will in fact have used 1200 pixels worth of information to spread it over 1800 pixels.
You can see what happens in the exaggerated case of the 30x22 image with the letter boxes:
B |
 |
The left picture is 30x22 the familiar B, the right one is 300x220 whereby the information was taken from B and was cleverly spread out by the imaging application.
Although not good, it's more acceptable to the eye. The following two are a more realistic example where the 300x220 image was transformed into a larger 400x293 image:
A |
 |
At the left the original 300x220 image A, and at the right the same scene transformed to a 400x293 image. The "spreading" is almost not noticeable.
The graphic designer should of course do this work, not you:
The case where you have more pixels than needed is of course easier: just send the file, the graphic designer will throw the extra info away.
