pixelated perceptions


Digital images are made of pixels, shorthand for picture elements – tiny area segments of uniform color and intensity. The digital aspect of imaging encodes pixels electronically into sequences of zeroes and ones that computers can understand. The number of pixels per inch (abbreviated ppi) vertically and horizontally determines how well an image is resolved. When viewed from a suitable distance (a common standard is 14 inches), the regular grid of pixels in a digital picture can look like a smoothly continuous image.

Although one normally thinks of it as a continuous medium, photographic film is a clumpy complex of chemical dyes and light-sensitive silver-halide grains that are layered in a plastic emulsion. While similar to pixilation in effect, the nonuniformity of film – its perceived grain – is more correctly described as noise and occurs randomly on a spatial scale much larger than the dimensions of light-sensitive grains. A technology exists to quantify film nonuniformity and relate it to pixilated spatial resolution. The highest resolving power, equivalent to 4065 ppi, was realized by Fujichrome Velvia (ISO 50). While Velvia-50 is no longer produced, this wonderful fine-grained film has been used in much of the photographer’s portfolio, and a few rolls remain viable in photographers’ freezers worldwide. Current contenders for top-film honors offer a resolving power of 3300 ppi – a bit more than three ten-thousandths of an inch, or about one-half the diameter of the finest human hair.

Even the eye sees in a pixilated way, of course. Photons activate retinal cells (the smallest of which are one to two ten-thousands of an inch in size), and the brain processes their minute electrical signals into an illusion of smoothness and continuity. Digital technology is quite analogous to visual perception, with computer software replacing the brain’s image-processing functions.


creating pixels


Remastering a picture from photo-transparency begins with an electronic scan of the film. The eye (more precisely, the retina) of the scanner or of a digital camera is a cluster of metallic cells (a charge-coupled device, abbreviated CCD) that produce electrical signals in response to light. Each cell creates an image pixel. The scanner is effectively a digital camera with a self-contained light source and system of glass elements which project an image of the film onto the CCD array.

A scanner cannot resolve the source film on a scale smaller than the CCD’s photo-responsive cells. Optical resolution, quoted in pixels per inch, is the metric for this limiting CCD scale. Relatively inexpensive scanners offer optical resolutions of 4800 ppi, and top-of-the-line equipment is capable of more than 20,000 ppi. In other words, the fundamental resolving limit of scanning technology rivals and exceeds that of film. Nevertheless, the quality of a scan and ultimately how much an image can be enlarged also depend critically on scanner optics and the path of light through the transparency, and it is here that scanner technology diverges most in cost and quality.

The resolving power of the scanner’s glass elements must be superior, exceeding both the resolving power of the film and the optical resolution of the CCD array, for a scan to resolve film details clearly and accurately. Light path matters because the basic colors (cyan, magenta, and yellow) of the image form in individual layers stacked within the film’s thickness. Light must pass through the transparency perpendicular to the sandwich of layers to assure that the color images overlap perfectly when projected onto the CCD. Warping or wrinkling of the transparency surface likewise distorts what the CCD sees.

Flatbed scanners hold the source transparency flat by its edges but generally do not flatten warps or wrinkles. The compactness of flatbed scanners makes it difficult to achieve perfectly perpendicular illumination through the source. Flatbed scanners also tend to be economical, and that means some compromise is made in the quality of optical elements. Drum scanners control the transparency surface precisely, curving it along a cylindrical surface that is perpendicular to the light path. They use a long optical path for added precision, and many use photomultiplier tubes instead of CCDs. Drum scanners are very expensive, but they compromise little and assure the highest-quality scan.

enlargement


To produce an output image that is larger than the source image, scanner or other image-processing software scales pixel dimensions to span more area. Therefore, enlargement reduces image resolution. It is generally recognized that photo-quality prints require image resolutions of at least 300 pixels per inch, less than which the artifacts of pixilation become evident without the aid of a magnifying lens. However, perception of quality depends somewhat on the sharpness of the source image. A digital original resolved to fewer than 300 ppi at its output dimensions can still yield a photo-quality print. Image-processing software can add pixels during enlargement using various interpolation schemes. Interpolation is not true resolution, however, and can degrade image quality when applied inappropriately.

The electronic data resulting from a scan and enlargement are adjusted, to correct for the color biases of the scanner and the printing media, to remove defects caused by dust or scratches on the original image, and perhaps to reveal or refine details and to crop the image to final dimensions. Image-processing software and color-corrected, computer-controlled hardware facilitate digital adjustment, with minimal proof-printing. More importantly, digital technology expands the creative possibilities of once-traditional silver-halide darkrooms. When the desired output image is achieved, the pixel data can be saved electronically, enabling the print to be reproduced essentially identically at any time in the future.

printing


Scanners and digital cameras create pixels. Computer software manipulates pixels. Monitors display pixels. Printing creates dots – tiny sprays of ink that actually penetrate and then are protected by the surface coating of the paper. Printer software translates pixels into dots. The dots sometimes differ in size within an image, and they are neither the same size nor shape as the image pixels. The number of dots per inch can determine how well a printed image is resolved (if there are fewer dots than image pixels). Fundamentally, dot resolution determines the degree to which a printed image looks truly photographic (rather than speckled like newsprint). High-end inkjet printers using resolutions of at least 1440 dots per inch can produce images that are practically indistinguishable from photographic prints.

The word giclée, pronounced zhee-clay, is often associated with high-quality inkjet prints. It is a nice francophonic word that was invented originally to distinguish between a low-resolution proof and a fine-art print produced on an Iris Graphics inkjet printer. Today, giclée signifies the production of high-resolution inkjet prints in a process using light-fast inks of six or more colors.


references

Digital Images: A Practical Guide, Adele Droblas Greenberg and Seth Greenberg, Osborne McGraw-Hill, 1995.

Link to details about film resolution.
digital darkroom
copyright notice

All rights are reserved. All images are Copyright © David Tubbs (except where indicated). Using images from this website for any purpose is unlawful without prior licensing agreement made by the photographer. The content of thegreensignature.com is Copyright ©2007-2009 David Tubbs and may not be reproduced or used without prior written permission.