Image Sizing for Large Format Printing

One question pops up time and time again when producing artwork for use on a popup exhibition stand, bannerstand or other large format printing application – “What d.p.i. should I save the images?” This is impossible to answer without knowing more information about the size of the original, what size it is to be enlarged to, the subject matter, how it is to be used within the design and how it has been produced and a number of other factors. It can be considered that perhaps the question is not the right one to be asked but we will return to that later. The first thing to be considered is the origination. Most images available from image libraries as well as those sourced from professional photographers specifically for the project will have originated on a digital camera. At the time of writing the highest resolutions available are around 15 megapixels. So it can be deduced that for a photographic image the maximum pixel size that can be achieved for a single image is: 4492 pixels x 3328 pixels (which gives a 42.8M image). This, therefore gives us a maximum achievable image file size of 42.8M. Just to get a handle on why this is significant let us now consider that image in terms of d.p.i. At screen resolution – 72 dpi this will translate to a physical dimension of approximately 158cm x 117cm At a print resolution of 600 dpi which most large format printers are capable of this equates to a physical size of 38cm x 28cm Both images are displaying the same amount of information but one is 5 times the size of the other! So you start to see that considering dpi is meaningless without also knowing the physical size that you wish to achieve. This concept will be further complicated by the fact that the large format printer will print a specific resolution regardless of the image information that is sent to it as it will interpolate (add or remove pixel information) to suit the specific pitch of the dot pattern. If, for a certain size image, it is given more information than it can print at 600 dpi, it will remove pixels at regular intervals and conversely if it has too little information it will add dots by looking at the dots around the one to be printed and calculating the colour that will produce the smoothest transition. There are limitations in both directions, enlarging and reducing. An image that is over-enlarged will go very soft and blurry at the transitions from one colour to the next and an image that is printed too small will have too little space, at a fixed dot size and pitch, to reproduce the graduation between one colour and the next. Imagine a page of text reduced so small that there is only one dot available for each letter – it would obviously be illegible. A further complication is added when you factor in whether any jpeg compression has been added to the image. At high compression ratios the image can be damaged and show ugly artifacts especially at the edges of image elements and in areas of high contrast. Graduated areas such as sky can also lose the subtlety of the gradient and become stepped and posterised. On the plus side we must consider that we are dealing with large format graphics here and as such they are generally viewed from some distance away. Even with quite low resolution origination, with the judicious use of interpolation to disguise pixellation, it is possible to produce results that are acceptable when viewed from a few feet away. So, to return to the original question of what dpi an image should be saved at, for a particular purpose, it is perhaps more effective to thing in terms of how much information there is – its megabyte size, and its suitability for enlargement (i.e. its image quality and complexity of detail) As a rough guide it may be helpful to think in terms of one image covering the whole of a popup stand. To give a good balance between the quality of the final print and the size of file, I would recommend a file size of around 150Mb – 200Mb. No I haven’t forgotten that the original image would have been 42Mb or less but it is worth increasing the pixel size of the image using interpolation (or ‘resing up’) in order to minimise the enlargement of the pixel shape thus avoiding any blocky shapes or pixellisation. If the image covers about half that area it need be only half the file size and so on. Another way to calculate this would be to save the image at around 25Mb for each square meter of space covered.