Printing Isn't an Island - Part 2 The Ins and Outs of Resolution

MonkeyLek · February 6, 2003 12:00 am

You write a great tutorial and put it on your web site. All your friends ask you to send them a copy. You print it out and the graphics look terrible. They are not the sharp, sparkling graphics that you saw on your screen. Some even look pixelated. What can you do? At this stage of the process, all you can do is basically add a Band-Aid or two. But you can prevent this from happening by understanding some concepts. The basic concept is Resolution. Unfortunately, this is one of the most misused words in the language of graphics.

The term Resolution in the field of computer graphics has various meanings depending upon the context in which it is used. When an image is scanned into a computer, this image is called the input image. The technical term for the number of pixels per linear inch of this input image is ppi (pixels per inch) or spi (samples per inch). However, the output, especially when referring to an inkjet, laser, or dye sub type of printer, is in terms of dpi (dots per inch). Unfortunately, these terms are often used synonymously even though they have technical differences. Most scanner manuals refer to dpi and do not even discuss ppi. Some books also talk about monitor resolution in terms of ppi and dpi. One considers screen resolution to be 72 dpi or 96 dpi. (The term lpi (lines per inch) will not even be addressed in this article.)

Imagine that you had a linear inch and that it was comprised of 72 dots (pixels) spaced evenly. Then, imagine that you had a linear inch of 300 dots (pixels) that were spaced evenly. Which would be denser? The answer is obvious. It is also obvious which would look better on paper. The question could be asked why is there no apparent difference when looking at an image on a monitor. The resolution of a monitor is 72 or 96 dots per inch. Thus, the additional dots or more correctly named pixels are not needed for an image on a monitor as they are for an image on paper.

The next question is: "Why cannot I just add pixels to an image to make up the required number for printing specifications?" Programs add pixels through various methods of interpolation. This is called Resampling. If one looks at Adobe Photoshop's methods, one finds three methods: bicubic - which is the best, nearest neighbor - which is the fastest, and bilinear. To just add pixels is similar to a guessing game. For that reason, screen captures have to initially capture the most pixels possible. One cannot just add pixels and hope that the colors and all their nuances will be correct. At the end of this article, I will discuss resampling further.

When the linear dimensions of a file are changed, that technique is called Resizing. For example, if one had a file that is 100 dpi and one increased the dimensions without adding pixels, the number of pixels or dots per inch, for example, would decrease. Conversely, if one decreased the linear dimensions of a file and did not remove pixels, the number of pixels or dots per inch would increase. Often, one uses combinations of the above.

Sometimes it is absolutely necessary to resample because we do not have control over a screen capture. Perhaps it is of a small but important aspect of a page. The following is a way I have found that can sometimes work. Unfortunately, one cannot always tell until the proofs might come back from the printer.

In Photoshop, resample your image in small increments until the desired size is met. Remember that resampling interpolates the color of the surrounding pixels to assign color to new pixels. I have found a few passes works better than one. Then if the image is not clear you will need to sharpen it. First, go to Image>Mode>Lab. Lab is a device independent color space. It will not create halos around the pixels in an image when sharpened. Go to Window>Channels. I keep it in the palette grouping with layers. Select the Lightness Channel. Then Sharpen conservatively using Photoshop's or any other sharpening filter that you like. I use Photoshop's Unsharp Mask filter and set it at Amount: 50, Radius: starting at 1, and threshold 0. I have found that it is better to sharpen more than once that to sharpen a lot at one time for this type of work.

Remember! This might not look good in print so design your article, review, or tutorial so that the particular screen capture that you have to "fix up" can be discarded if it does not look professional.

On the everyday side of things, we all enlarge our images and usually do not notice changes. I even have seen prints of web images that were passable (not for exhibit but to show friends.) The monitor screen as well as the inkjet printer are very forgiving. The real test is when you are doing four color separation for a job and the output must match the input and the input has been severely enlarged either by resampling, resizing, or a combination.

A rule of thumb is that an output resolution of 300 dpi is usually safe for most printing specifications. Why do I say output? I use that term because some scanners can be very confusing. I have seen scanners refer to resolution and never say whether it is input or output resolution. If one is working 1:1 it doesn't matter. But if one has a 3" x 5" print and one wants to enlarge it, it is necessary to know how one's scanner works. If your scanner only give input information, and you set it for 300 and you enlarge your picture, you will not get an output of 300 because you have increased the size of the image without increasing the number of pixels. But if your scanner, and most seem to be of this type now, give output resolution, then it does not matter whether you increase or decrease the size of your scanned image from the original size. Playing with your scanner and checking with image size in Photoshop or any other image editing program will give you the information you need. However, once again, you have to be logical in how large you are going to increase the size of your image. But I am not going to talk about scanners and optical resolution vs interpolated resolution.

The next article in this series will be a discussion on archival printing and how preservation of digital prints differs or does not differ from the preservation of other art media. This article will then be followed in about two weeks by specific tests of archival inks, different substrates to expand ones printing horizons, etc. I am also always open to requests from readers.

All supporting images are copyright, and cannot be
copied, printed, or reproduced in any manner without written permission from the artist.

  • The Paula Sanders Report is a regular Renderosity Front Page featured column, where Paula investigates and comments on graphic software, techniques, and other relevant material through her reviews, tutorials, and general articles.

Article Comments

No article comments have been posted. Sign In or Register to share your thoughts.