Just read it.... What more can I say?
1999 by the Feather River Canyon News - All rights reserved

What is Interpolation?

Today is
Updated  Saturday, November 06, 1999

by Randy Glass - Copyright 1999 by the Feather River Canyon News - All rights reserved.
     These images may not be stored, reproduced, or distributed in any way without the expressed written permission of the Editor of the Feather River Canyon News.
     The information contained herein is supplied where is/as is. You accept all responsibility for using this material. The information on the sites reviewed or linked here is out of our control, and supplied here for your convenience. We accept no responsibility for this information and no warranty as to this information's applicability or safety is expressed or implied. No responsibility or liability is or will be assumed by the author or the Feather River Canyon News for any loss or damage due to its use.

   So you have an image that is about 640x480 and you would like to print it at 8x10 to frame and give to Grandma. The problem is that you know that if you just print it that big that the picture will look like a mosaic tile close-up. All those little pixels that are just about invisible at around 4x5 will start to show like crazy at 8x10. What to do... what to do?

   It is time to interpolate! Interpolation is a way to add pixels that aren't really there to make an image print properly at sizes larger then normal. Let's look at an example to get an idea of what this is all about:

   Lets say you have an image that is 1000x500 pixels in size, and your printer prints at 250dpi, and you print the image at its 'natural' size (not shrunk or expanded). The print size will be 4 inches by 2 inches. But what if you want everything to remain the same as above, and you want to print that image at 8 inches by 4 inches? It doesn't take a math major to see that something has to give. Either the print resolution has to drop to 125dpi or you need more pixels. Well, that's what interpolation does. It fills in the missing pixels (for the above example you would need four times as many pixels- twice as wide and twice as tall; 2x2=4). That's a lot of information to be made up. Why bother? Sure, you can just make the image four times larger and print it, but the results will be less than satisfying. The image will look quite pixelated- blocky.

   There are a number of different interpolation algorithms, and depending on how 'smart' the interpolation method is in your printing application will greatly affect image quality. Interpolation problems show themselves in different ways. Sometimes you will notice a 'blockiness.' Small areas of low resolution that were quite dark in the original image (like a shadow on a rock) will turn to larger black rectangles. Another place to look is fine lines that run at angles such as high contrast edges between a building and the sky, power lines, hand rails, and such. These turn into 'zippers' or 'ziggy-zaggies' when poorly interpolated.

   Take a look at these example images. They will help explain interpolation a bit:

Let's make believe that this is a line in a digital image (well, it actually is).
If we view that line in a larger size (say 300% of the original size) the pixelization (zipper edges) become more apparent. We have all seen this in digital images when we try to view or print them using too much magnification or enlargement.
This is roughly what Interpolation tries to do. It uses mathematical computations to figure out how to make the larger image look like the original and by changing pixels it can help remove the pixelization (the green represents what the interpolation would try to do, but the green would would actually be blue).
   That would seem to make interpolation the great digital image savior, but it doesn't always work that way. The quality of the final image depends on a lot of factors including the resolution of the original image, how much larger it is going to be made, and what interpolation method is used.
Here is an example of an image that was printed at a size quite a bit larger than the original and without interpolation.
Here is that same image but this time it was printed with interpolation. Although the zaggies are gone, the image became quite soft. Choosing an interpolation method that is not quite so severe as the one used here might have improved this image. The better applications allow the user to choose among several interpolation methods.
   So how does interpolation do all this? A complicated series of mathematical computations are done within the application, usually as the image is being spooled to print. The interpolation algorithm looks at one pixel, and then all the pixels in the surrounding area, and then decides what pixels to change. By the time I got through understanding all the details and explained it to you, we would be sharing our oatmeal in the home, so let's just leave the details to the experts and just worry about image quality.
   As you can see, there is always a trade off- No free lunches. Interpolated images can become soft. The algorithm can add so many pixels in such a heavy-handed manner that the image gains a soft focus look. When you have a choice, use a higher resolution image and don't try to print it too large. When that can be done (such as when you are printing a bunch of high resolution images at 4x5") it is best to not use any interpolation. But when you want to print an 8x10 from a 1280x960 image, a good interpolation algorithm can be indispensable for producing useable output.

   How do you choose an interpolation method that is best for you? Some applications don't give you a choice. When you do have a choice, experiment. Variables such as what printer, paper, ink, and printing application you use along with the image itself will dictate what method to use.


Back to the FRCN Digital Imaging Main Page
to the FRCN D.I. main page