Saturday, April 28, 2012

Viewing images of the U.S. Census

The Ancestry Insider had a very interesting comparison of the images for the 1940 U.S. Census, comparing Ancestry.com, Archives.gov, FamilySearch.org and MyHeritage.com. There are other images of U.S. Census out there, all of which were scanned from microfilm copies from the U.S. National Archives.

I thought that there was some additional information about the Census images and digital images in general that might be interesting and helpful to those viewing the images, including the images from previous censuses.

The comment made by The Ancestry Insider which he retracts is “NARA did the image scanning so Ancestry.com’s images can’t be better than everybody else’s.” The real issue is the quality of the original census forms and the handwriting of the enumerators. Literally, an absolutely faithful reproduction is useless if the original forms are unreadable. The reality is that many of the original forms are bad or even missing. For example, the 1890 U.S. Census was destroyed in a fire except for certain preserved schedules.

Here is the official statement from the U.S. Census Bureau concerning the availability of the Census documents:
Microfilm copies of the original population schedules, from 1790 through 1930 (virtually all of the 1890 records were destroyed as a result of a 1921 fire), are available at the National Archives in Washington (www.archives.gov) and its 13 regional archives (see page 4), and many libraries in various parts of the United States. Most have facilities for making paper copies from the microfilm. The National Archives also rents and sells the microfilm rolls (see below). The Reference Branch at National Archives headquarters (see listing on page 4) will accept photocopy orders by mail or through orders online, given exact page numbers; it will not do research.
 The 1940 U.S. Census was the first time that the images were digitized rather than being made available on microfilm. Microfilm is really just 35mm or 16mm film in rolls like movie film. Microfilm is still available and still being used for archive purposes. See Eastman Park Micrographics, Inc. The quality of a microfilm image is dependent on the same factors important for any photograph, the quality of the camera, the quality of the lenses, the lighting and the focus. In order to determine the quality of an image made on microfilm, it is absolutely necessary to compare it to the original document, not an original image of the document. These same factors influence the resolution and quality of digitized images also.

With digitized images, there is an entirely new layer of viewing issues; the device by which the images are viewed. In other words, what you see on the screen is partially dependent on the resolution of your monitor, not so much the ultimate resolution of the image itself. If your monitor is low resolution, the image will degrade to the level that it is viewed at. Some images will look better on a high resolution monitor and will look very much less acceptable when viewed on a lower resolution monitor. The only accurate way to compare two digital images is to look at them with the same type of monitor set at the same resolution and the same magnification.  Even if you compare the images using the same parameters, you mostly end up comparing the two viewing devices rather than the images themselves.

Microfilm images of the various years of the U.S. Census are notoriously spotty in their quality, sometimes this is due to the quality of the original document, but more frequently it is the quality of the individual microfilm images. As the Ancestry Insider pointed out, the "original" digitized images from the National Archives are blurry at magnification. There are some graphics tools such as those in Adobe Photoshop, that can be used to sharpen or clarify an image, but basically, there is a upper limitation imposed by the resolution of the original scanned image.

For example, I went to the National Archives' site 1940census.archives.gov and downloaded a file of the census from Apache County, Arizona. Here is a copy of the downloaded .jpeg file:


By opening the file in Adobe Photoshop, I find that the file is 7148 pixels by 5452 pixels at 72 dpi. Using the Magnifier tool, I can zoom in on the image at 400%. Where it begins to pixelate. Here is a screen shot of a portion of the image at 400%:


Notice that there is no blurring of the image unless you enlarge it on your screen. For example, look at the image above here in the post. Then click on the image to enlarge it. The enlarged image shows blurring or pixelation while the smaller version of the same image does not.

Ok, so I spend a few minutes with Photoshop CS6 on the above image and here is what it looks like:



The point of this is simple, the results are very complex. The so-called "quality" of a digitized image depends on a number of factors, one of which is the degree to which the image is magnified artificially by the zoom-in function of many programs. Some images online are actually made up of a series of images at different resolutions to allow you zoom-in without an apparent loss of quality.

So what does all this mean for the U.S. Census images?

You can get better visual image quality by using a higher resolution monitor.
You can get "blurring" from a bad photograph or poor digital image.
You can get higher contrast by using programming techniques and tools available in image processing programs like Photoshop.
Any image, at any resolution, will show pixelation by zooming in 400%.

If you are dissatisfied with the quality of any image in any program, find the image in another online program and see if the quality is better. If you like, you can download the image from the National Archives site and manipulate the image yourself with Photoshop or a similar program. Good Luck. 

No comments:

Post a Comment