Roughly 125 million photoreceptor cells are intermingled nonuniformly over the retina of the human eye. Rods (roughly.12.700 per inch) perform the function of a high-speed, black and white film, while the 6 or 7 million cones (roughly 4,200 per inch) perform the function of a low-speed color film.
It could be incorrect to compare the human eye with a scanner or digital camera, but the disparity may be narrowing. Both operate by evaluating the tonality and color of a single spot of reality, a process known as sampling.
|
|
Whether sampled by a row of sensors on a moving bar (scanner), or with a stationary array of sensors (digital camera), the image or object is converted to a collection of discrete areas called "picture elements" or "pixels". Each sensor assigns a single value to the area it "sees" (in this case zero for black and 255 for white),
Dividing the image into smaller squares, by increasing the number of sensors, will increase the pixel density and record more detail as shown below. |
|
The New Leica Manual , Morgan & Lester, 1951, states that the average eye, at a viewing distance of 10 inches, can distinguish individual lines when they are no less than about 1/100 inch apart. Thus, in the final print viewed from a distance of 10 inches, "any detail 1/100 inch in size or smaller will be acceptably sharp." This observation forms the basis for the calculation of most depth of field tables, and is also why we often do not see silver grains, or the dots that make up prints from imagesetters or digital printers.
SPI (SAMPLES per inch). Describes the number of samples (pixels) both horizontally and vertically in each square inch scanned, or recorded by a digital camera. A 600spi scan of a one inch document would supply 600 pixels horizontally and 600 pixels vertically, a total of 360,000 pixels. One would hope that each sample covered exactly a 1/600 inch square.
PPI (PIXELS per inch). Since the pixel is the result of sampling, has the same general meaning as "spi".
DPI (DOTS per inch). A printers term describing the fineness of the mesh or "screen" used to print halftones. It is frequently, INCORRECTLY, used to describe scanner resolution. It is TOTALLY independent of sampling frequency (spi) or the actual pixel count of the image (ppi).
So just how small do the pixels have to be? Or, how many pixels per inch are needed to produce a print of acceptable quality?.
To investigate the effect of pixel count on prints made with the ePixel.com laser printer, six ../../IMAGES/digital of the Luscombe tail shown above left were submitted via the internet with a request for 4x6 inch prints. The source was a 600 spi (samples per inch) scan of a silver print made from a Minox negative, from which the files with reduced pixel counts were made with Photoshop. All were compressed with the JPEG algorithm at "maximum" setting.
These examples, from a portion of each resulting print, illustrate the inability of low pixel counts to render detail, while demonstrating the equally important factor of diminishing improvement at higher counts.
There is a great visual difference between 50 and 150 pixel per inch ../../IMAGES/digital, and much less apparent difference between 200 and 300.
The 200, 250 & 300 ppi prints are quite similar although repeated shufflings and reorderings resulted in the correct order each time. The fact is that behind Plexiglas, on the wall, or in a scrapbook, all three would be accepted. BUT, there IS a difference. Although higher pixel counts can be utilized by some equipment, 300 pixels per inch has emerged as a useful compromise between file size and print quality.
The 150ppi print closely resembles the quality associated with traditional photofinishing, and may even provide 4x6 inch prints which are more pleasing than traditional prints of the same size.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Although the ../../IMAGES/digital shown here clearly illustrate the effect of pixel density, the differences are much less evident in the full sized prints.
|
|
RETURN HOME