Mike Pill ARPS - I-magery - photography
   I-magery- creative photography
Mike Pill banner name
    
Home button
My Images button
Buy my Images button
Articles button
Links button
About me button
About my site button
Contact Me button

Many people when moving to digital imaging are often confused by the various terms that are often glibly bandied around and what they mean. These include terms such as pixels, dpi, image resolution, file resolution, screen resolution and so on. We all read various magazines, talk to friends and colleagues who ‘know’ about these things but often when it comes to explaining what’s going on can often lead to further confusion rather than helpful clarification. Its not helped by the PC, ink jet printer, camera manufacturers and so on who assume that everyone who can read their literature is completely familiar with all the terminology and has a degree in photography, optics and in these days electronics.

This article attempts to explain some of these terms. So lets try and explain some of the more common terms and how they apply to digital imaging.

But don't forget having more pixels in your digital camera is not a guarantee of better results.

Pixels, colour depth and all that

An image comprises a number of coloured dots (that show up as squares in the images below left).Image showing pixel structures

Let's now explain the value or number part of the pixel. Each pixel has a value or number. This is sometimes referred to as the colour depth, for example, some files, scanners etc. are described as having a colour depth of 24bit, 36bit etc.. Simply, the larger the number the better in terms of being able to convey a wider range of shades.Whilst we have talked about images that we can look at we need to understand a little how digital cameras work. When the digital camera takes an image it uses light sensitive detectors arranged in a rectangular array that capture the light from the lens and turns into electrical signals. However, these detectors are not sensitive to colour – only brightness.

To see colours, a filter, called a colour filter array (CFA), is placed over each pixel of the sensor. The most Pixel structuecommon CFA uses what is called the the Bayer pattern, illustrated by the colour checker board pattern below. In such a pattern, there are twice as many green filters as red or blue. When processing the raw information provided by each photo diode, the camera analyses each pixel’s colour information, assigns a value and the software, the scanner, camera etc. combines it with that of adjacent pixels and recreates the full-colour image. Unfortunately just to add to the confusion there are other sensors that use different sensor shapes and patterns but the principle is the same e.g. Fuji.

Bayer pixel patternMost sensors use CCD or CMOS technology with pixel filtering patterns as described above, apart from the Foveon sensors developed by Sigma used on their SDx cameras. These use the different colour sensors which are in effect stacked on top of each other on the sensor chip.

Digital Image File size

When you have created a digital image, whether it’s from a digital camera or a scanner etc. you will need to store it on your computer.After you have stored it as a file you will see that it has a certain size. The size of a file can be expressed in terms of numbers of bytes or words. A byte is 8 bits and a word can be 16, 24, 36 etc. bits in size.Let's consider the make up of a digital file that contains such an image, whether it’s from a scanner, digital camera etc. Every file contains much the same sort of information (the data). What usually differs is how this data is arranged. We will not go into all the variations but merely outline the broad types of information.

A file will typically contain several sets of information:

The most commonly used file formats used in amateur circles are JPEG and TIFF.JPEG is a file format that uses what is referred to as 'lossy compression' to store the data. It does this by means some clever calculations on the image data that removes redundant information – but at a cost. As it’s a 'lossy' format then more information is lost about the image the more it is compressed. As the compression is increased to reduce the file size then definition decreases and image effects technically referred to as 'visual artefacts' start to become visible. These are often seen as halos around the image.

TIFF is a non-compressed format. (Although just to add confusion some 'non-lossy' compression can also be used in this format).RAW files are becoming popular especially amongst professionals, as these files are literally the basic files created from the camera image sensors without any compression and only basic sensor compensation. This allows the photographer to manipulate the real image without camera imposed image problems such as compression artefacts and oversharpening.

There are many other formats each of which has its pro's and con's, supporters etc.

Image Resolution

When describing an image there are several attributes that describe its ability to show detail.

Dots per Inch (dpi)

As discussed an image comprises many pixels. Sometimes the resolution of an image is described in terms of dots per inch (dpi) – as a result of the UK's excessively slow transition to the metric format some items still sometimes remain – confusingly – as imperial units when describing the sizes of things. (it can also be described in metric units as dots per cm). If you take an image and count the number of pixels over an inch you will get the dpi value.

Absolute Image Size

Pixel structure illustrationAn image can be described in absolute terms as being so many pixels in size horizontally and vertically. For example 3200x1700 pixels

Optical resolution

Whilst we are discussing dpi, image resolution we also need to consider resolution. This is completely separate from the image pixel size. This is why an 8M-pixel camera might produce less sharp images than a 5M-pixel camera if the optics are inferior. I do not propose to go into physics of optics, lines per inch resolution etc. except to say that when choosing a digital camera you need to take into account the optical as well as the digital resolution. In addition avoid cameras that offer a small optical zoom range but a huge digital zoom capability. Generally the best option is the other way round.

Monitor (and video projector) resolution

Irrespective of:

you can only view your image on your monitor at whatever resolution it has been set to by the computer graphics card.When using a 17" monitors, for example, graphics cards are often recommended to be set at 1024x768 pixels. This means that the screen you view, whether CRT or LCD, is a maximum of 1024 pixels wide and 768 pixels tall whatever the image size you are going to display.For example, if your image is 500x300 pixels it will have certain size when displayed on a 17" monitor but when you display the same image on your 21" monitor set to the same screen resolution it will have a larger size physical size. But if you set your 21" monitor to say, 1280x1024 pixels resolution then the image will look smaller than when displayed on the 17" screen set to 1024x768pixels.

However, when you use Photoshop or any other image manipulation programme it adjusts the image to fit the screen size you are using irrespective of the screen resolution set by the graphics card but only for viewing purposes, the image remains at whatever resolution you have set it to.

Scanning resolution

Scanners are used for slides or negatives, whether they are dedicated units or multi-purpose flatbeds. For those of us who still use film as the image capture system these devices are usually described in terms of scanning and output resolution ranges. Again the definitions are similar, in so much that a scanner will typically be described as having a resolution in dpi.In practice this is a function of the image scanning head and the physical size of the scanning mechanism.

A flatbed scanner will typically have a larger mechanical scanning range than a dedicated film scanner purely because it has to cope with sheets of paper as well as well as much smaller film. Its output resolution will be what you set it as. In practice it depend what you want the final scan for.If the images are only ever going to be used for a web site then it’s usually quicker to scan at lower resolution as this also produces a smaller file size. If, however, you may wish to print at A3 then the rule of thumb is use the maximum resolution possible with the maximum colour depth. This rule of thumb also applies for scanning B&W but this does depend on the way in which individual scanners derive the B&W outputs, as some scan with one colour channel and other scan with the colour channels then apply an algorithm to create the B&W image – try scanning the image in either colour or B&W and see which gives the best result - then use that).

If you scan at the maximum resolution and colour depth you then get an image that can be used for any purpose. You should retain the maximum information at all times and only convert the image for its final use when ready to do so. The scanned or final manipulated image should be at the maximum resolution at all times. It’s easy to take detail out but - once it’s gone its gone!

Interpolation

One downfall of the above statement is that one should scan with the maximum resolution. In practice this should read the maximum optical resolution. Many scanner software packages are capable of generating image files with a pixel resolution considerably higher than that of the basic scanner optical resolution. No extra information is found or scanned merely that the scanner software guesses what the extra detail might look like from the pixels that have been optically scanned and makes up a set of values based on these to fill in the gaps. For example, a flatbed scanner may have an optical resolution of, say, 1200dpi but capable of generating files that give an interpolated resolution of 9600dpi.

Printing and Printer Resolution

This is often another point of confusion. Again irrespective of:

the printing and printer resolution are separate from the above.

Printer resolution

This is the resolution of the printer as described on the box! This is not unlike monitor resolution in that it is a way of defining the number of pixels that can be laid down on a piece of paper. If we consider ink-jet printers which most amateurs use these days then these are usually specified in terms of dpi. This is because the ink jet heads are of a finite size and can distribute the ink over whatever paper size the printer has been designed to handle. So this sets a maximum capability. For example an Epson 1290 A3 colour printer will print with a dot frequency of 2880 dpi. In practice the quality is perfectly adequate at 1440 dpi and it is quicker as well.

Printing resolution

This is the resolution that we can use a printer to print our image. We have now created our masterpiece image, for example, a 24bit colour and 3513x1000 pixels in size. Suppose we wish to print on an Epson printer at 1440dpi, what is the best printing resolution? To answer this question we now have to introduce the concept of resizing or re-sampling. As the image is unlikely to ever match the printer resolution we have to create an image that will result in the correct size when printed on paper otherwise it will end up to big or to small. The generally held view is that to get the best quality print from a printer the image dpi that you are going to print should be a multiple (or sub-multiple) of the printer resolution. In other words you resize the image.

In practice the method I adopt is, once I have finished editing the image, I do the following actions in this order:

An example of a workflow for printing a digital image on a 1440dpi printer.

Note that the starting point is a correctly calibrated monitor. First create your image on screen then to print:

It should be noted that like many things in digital imaging, there are many ways of doing the same thing. The above workflow works for many people who have adopted this process and gives a sound basis for a high quality print.

Summary

To summarise, this article has covered:

The key points have been outlined in a small number of pages. Hopefully this clarifies many of the points that have been asked in various workshops.There are a lot of articles that have been written on the subject and for more details go to the links page and browse amongst those.