Pages

Tuesday, March 3, 2015

Digitizing Genealogy -- Resolution is always an issue

Roger Gilbertson [CC BY-SA 2.0 (http://creativecommons.org/licenses/by-sa/2.0)], via Wikimedia Commons
The manufacturers of both scanning devices and cameras have been locked in a pixel war for many years. As a result of this fixation, claims for high resolution are common. The results? The image shown above is an example of what is called a moire pattern. This is caused by the interference of the light rays and is very common in images where the pixel count exceeds the length of the waves of visible light. You can't ignore physics when you start to scan and take photographs. Before you run out and spend some hard-earned money on scanner or a camera, you should realize that resolution, as such, has its limits. At this point, we need some definitions:
  • Resolution -- the amount of small detail you can see in an image
  • Magnification -- how much an image can be enlarged
  • Contrast -- the difference between the lightest and darkest parts of an image
  • Sharpness or Definition -- the viewer's perception of the results of focus, depth of field, contrast, and the maximum potential detail inherent in the recording media
  • Depth of Field -- the area of sharpness in front of and behind the main object in an image
  • Grain -- the limit of the resolution of the media (from film photography; the size of the silver particles making up the developed film)
  • Image -- the product of a scanning or photograph process
  • Noise -- also known as artifacts, this is the amount of unwanted defects in the image
  • Wave length -- the distance between the tops of the waves in angstroms or Nanometers
  • DPI also LPI also PPI -- dots per inch, lines per inch and pixels per inch, all measurements of the resolution of a image making device
  • Optical resolution -- the actual or physical resolution of a scanner or lens
  • Sensor -- the light gathering mechanism for a scanner or camera
  • Megapixels -- the total number of sensors in a sensor array
  • Digitize -- using an electronic device such as a digital camera or scanner to create an image that can be manipulated and displayed with a computer-based device

OK, so let's start with a reality check. What is the absolute resolution of the human eye? This question means, what amount of detail can you see with your eyes assuming you have perfect vision?

You might be surprised to learn that if you put your eye's resolution into the terms used today for selling scanners and cameras that you can only resolve images at about 74 Megapixels. Interestingly, some high-end cameras are approaching that resolution level. But is resolution the end all and be all of making images? Not at all. The real issue, although the ads would have you believe otherwise, is the whether or not the image coveys the information intended. That is, can you read the document? That has more to do with contrast than resolution.

Here, I need to pause and sort out where I am going with this issue of resolution. There are really two completely different issues from a genealogical standpoint; the digitization of documents (i.e. paper etc.) and the digitization of photographs (i.e. images of ancestors etc.). Obtaining readable and acceptable images of documents is very different from digitizing photographs (unless they are photographs of documents such as microfilm etc.).

Now back to Megapixels and DPI. Digital cameras have, since they were first sold, have been touting the resolution of their sensors in Megapixels. Originally, the cameras were one or two Megapixels and lately, Canon has announced a new EOS 5DS and a companion camera, the EOS 5DSR, both with 50.6 Megapixel sensors. On the other hand, scanners are advertised with DPI ratings. For example, the Epson Perfection V600 Color Photo, Image, Film, Negative and Document Scanner claims a resolution of 6400 x 9600 dpi. So how do you compare the two?

It turns out that the numbers given by the companies that manufacture scanners are mostly exaggerated. The optical resolution of this Epson scanner is actually 6400 dpi. The claimed effective pixel count is 54,400 x 74,880 giving an incredible 4073 Megapixels. But there is a fatal flaw here in this reasoning. If that number were true, the file would be incredibly large. So how do you find out the answer to the question about comparison? You have to actually measure the resolution.

The way resolution is ultimately determined for lenses or for the produced image is to photograph or scan a standard resolution test document such as the 1951 USAF Resolution Test Chart or some other similar test. Here is an example of this type of chart:

EIA Resolution Chart 1956
This type of chart works with scanners, cameras and other optical equipment. But the catch is that buying the standard type of chart can cost from $300 on up.

After all that, for genealogists and archivists, there are standards for resolution. but that is the next post.


7 comments:

  1. Because we planned to scan important documents and photographs we asked the salesman for a really good scanner. He sold us a marvelous machine that can scan from 75 dpi to 4800 dpi. Every time we use it we are warned about getting the pix too large and it defaults back to 600 dpi. (Think maybe we purchased more than necessary, or even usable.)

    Also, last night I scanned photos at the BYU HBLL and noticed that their scanner produced two versions of each, one being extremely large, too large to edit or save. Can't help but be curious, how does that process work?

    ReplyDelete
    Replies
    1. Thanks for the comment. The issues you raise have to do with the settings of the scanner controlled by the software. Most of the programs that they sell with a scanner are adequate but not good enough to control all of the scanner's functions. I don't understand why this is the case, but it is. I use a third-party program for all my scanning that allow for control of all the things you are concerned with. It is called VueScan and is available for Mac and Windows.

      Delete
  2. A simpler and yet still accurate way to do this rather than paying the $300 for the chart is to print the Takinami version (making sure the elements in group 2 are 10 mm long!). http://www.takinami.com/yoshihiko/photo/lens_test/USAF.pdf

    This allows for a (not perfect) but close version of the test without spending so much.

    I think the concern with resolution is that in genealogy, especially with documents, that with a low resolution, letters get disturbed or muddled or mistaken because part of the letter appears to be a resolution error rather than part of a letter itself. We do need a certain level of resolution in order to zoom in pretty far to a document in order to determine if the transcription is right and get the right matches for our people. Nothing irritates me more than seeing names completely screwed up because the indexer or transcriber couldn't zoom in far enough to tell the difference between two lines or imperfections in the ink/paper.

    ReplyDelete
    Replies
    1. You are right, but making a copy is probably a violation of copyright if the copy is made for the purpose of avoiding buying the original. I am not sure that this is needed at all except in some very technical situations. I am continuing the series.

      Delete
  3. I wonder if 'data size' should have been included in the definitions James. I have met several people who believed that increasing the size (the number of bytes) of an image will restore the missing details that they lost when it was shrunk :-)

    I once made a comment on the findmypast feedback page that they had an arbitrarily small limit to the magnification (zoom) that they supported on their census images. The response was that the displayed resolution should be OK for normal eyesight, thus missing the point that I wanted to zoom in to analyse pen strokes, and to determine whether some marks were part of the text or other blemishes.

    ReplyDelete
    Replies
    1. Thanks for the additional comments. I am going to get to all that in future posts.

      Delete
  4. As the hostess of the monthly genealogy scan-and-chat event, Scanfest (http://tinyurl.com/Scanfest), I find this post fascinating and look forward to reading the rest.

    ReplyDelete