pixel-power

Digital Directions: Got Enough Pixel Power?
By Michael J. McNamara
provided by Popular Photography

How many megapixels do you need to get the quality you want? As you can see here, it all depends on print size.

Shooting with a digital camera for the first time can be a thrilling event—as more and more of you are discovering these days. But trying to figure out how to set all of the camera’s controls in order to maximize both image quality and performance can be frustrating. For example, a typical 4MP camera user must choose from as many as eight resolutions, ranging from 640x480 pixels to 3200x2400 pixels, and at least four different compression levels. But a typical camera manual doesn’t tell you which setting to use to make a photo-quality 5x7- or 8x10-inch print or which setting best captures images for your personal web page or for e-mailing. Choose the highest resolution setting, and you’ll barely get a single image on the puny memory card that ships with the camera. Select a low-resolution setting, and you’ll get dozens of images on the card that can be e-mailed in a flash, but aren’t fit to print. How in blazes do you figure out which setting works best for the application you have in mind?

I’ve developed a simple resolution chart that you can use to choose the resolution and compression settings that you need. It starts with a column listing the most common pixel resolutions for a variety of digital cameras and their equivalent in terms of megapixel ratings.

To give you a visual comparison between a variety of megapixel resolutions and what you can expect from a properly exposed ISO 100 color print film, take a close look at the detailed enlargements on this page. The detail in the first five eyes comes from 11x16-inch prints made from their respective megapixel file sizes on an Epson Stylus Pro 7600 printer. The original images were accurately focused and exposed using a manually focused digital SLR. (On the other hand, a 4MP point-and-shoot camera with a so-so lens or focusing system might not capture as much detail as a good 2- or 3MP camera.) Compare the digital prints to the detail from a popular ISO 100 color-negative film optically enlarged to 11x16 inches by an automated minilab printer.

What most POPULAR PHOTOGRAPHY & IMAGING staff members noticed was the lack of grain and the pleasing skin tones on the 11x16-inch prints made with a 9MP camera. (The shot on the previous page was taken with the Super CCD-based Fujifilm S2 Pro digital SLR in 12MP mode.) However, as we reported in our S2 Pro test in the November 2002 issue, the 12MP setting delivers closer to 9MP resolution.) The higher detail reproduced in the film print wasn’t as noticeable when both prints were viewed at normal distances (arm’s length). In fact, when viewed more closely, its sharpness and noticeable film grain actually worked against it, which is why most pro photographers choose a higher speed, lower contrast film such as Kodak Portra 160NC Pro or Fujicolor NPS 160 Pro for portrait shots.

The right formula
The fact that a 9MP digital capture could compete head-to-head with film in an 11x16-inch print really didn’t surprise us. Years ago, the editors at POPULAR PHOTOGRAPHY & IMAGING (PPI) determined that an acceptable photographic-quality print could be produced by using a file with as little as 200 pixels per inch (ppi—sound familiar?) resolution. You can use this finding as a quick rule of thumb to determine the photographic print-size capability of any digital camera. Simply divide the effective horizontal and vertical pixel resolution of any camera by 200. For example, a file from a 3MP camera capturing a 1500x2000-pixel image should be able to produce a 7.5x10-inch “photo-quality” print on a decent six-color inkjet printer.

Our 200 ppi resolution rule of thumb was based on comparisons between digital prints produced by the best inkjet and dye-sublimation printers available at the time and prints from color-negative film made at a top minilab. All were displayed at standard viewing distances and judged by both professional and amateur photographers. For example, 4x6-inch prints were judged from about one-foot away from the subject, while 8x10-inch prints were viewed at arm’s length. The pros among the group, who were used to viewing most enlargements up close or with a loupe, found the inkjet prints made at settings of 250–300 ppi more to their liking, but agreed that prints made with files set at 200 ppi would have acceptable photographic quality for consumers. That’s why both of these resolution settings and their corresponding print sizes are listed in the resolution chart.

If you’re confused about our use of pixels per inch to describe the resolution of an image from a digital camera or film scan, that’s not surprising. Most inkjet printers list their resolution in terms of dots per inch (dpi), and not ppi. That’s as it should be, since the dots produced by an inkjet printer are not the same as pixels. Originally used to describe individual picture elements (that’s where we get “pix-els” from) along a TV scan line, pixels now refer to the photo-sensitive elements in a camera or scanner’s CCD or CMOS sensor. The term is also used to describe the individual elements that make up a digital picture file, as well as the output resolutions of a computer monitor or a digital projector. In a digital file, each pixel contains color information corresponding to a point in the original scene. When these pixels are organized in a grid on a computer monitor or LCD screen, you get a picture.

When you choose to print an image file, printer driver software turns every pixel into dots corresponding to a printer’s color ink set. Unless the original pixel color is exactly the same as one of the inks found in the printer, different color dots must be mixed together to produce the desired color.

It’s the merging of all these microscopic dots of various colors (and in some cases, sizes) that produces the illusion of a continuous tone print with millions of colors (but without the tile pattern that you might see if you zoom into an image on your computer monitor). Smaller droplet sizes, in some cases as tiny as 3 picoliters (i.e., 3 trillionths of a liter!) allow the printer to place more dots next to each other, creating smoother tonal gradations, finer detail in highlight areas, and wider color gamuts. (Note: Because of the way different inkjet printers overlap and merge colors, a print from an HP Photosmart thermal inkjet printer featuring 600 dpi resolution might rival the image quality from an Epson micro-piezo printer featuring 1440 dpi. And dye-sub printers with only 200 dpi resolutions can readily produce a photo-quality print, but don’t do as well generating sharp text.)

Until the majority of photographers know the difference between ppi and dpi, or how to resize an image file for printing without excessive interpolation (see sidebar), I’d like to see a section of every camera and printer manual dedicated to explaining the formula for converting pixel counts to print sizes. Unfortunately, what that formula is and how to apply it is an ongoing disagreement among the manufacturers of camera and printing equipment, and all too often we hear about a new 2MP camera that is just great for making photo-quality 8x10 to 11x14 inch prints. Or worse, we find a manufacturer promoting the megapixel resolution of its inkjet printers. Is either possible? You do the math.

Interpolation: Can you cheat reality by adding pixels?
Shooting with a digital camera involves foresight. You’ve got to anticipate final-image size and set resolution accordingly, plus adjust compression levels, white balance modes, and even choose between still and movie modes—all before composing a shot. Under these circumstances, sooner or later you’ll wind up taking a great picture at a resolution setting that’s too low for the print size it deserves. Is there any way to produce decent prints from a less-than-adequate image?

Fortunately, most imaging programs allow you to resize an image file quite easily by adding pixels in a process known as interpolation. This helps smooth out jagged lines and it reduces the effects of pixelization that would otherwise occur if you printed a straight image file too large. It works best if you just want to add 50 percent more pixels to the original file. Increasing a file size beyond that may result in soft images with color banding in fine gradations, and other artifacts.

To get the best results, mix smart interpolation with careful sharpening. For example, let’s say your image is from a 2MP (1200x1650 pixels) camera. Following the 200 ppi rule, you should be able to make an 6x8 1/2 -inch print from this image without using interpolation. But in order to make an 8x10 print, you’ll need to boost the pixel count to 1600x2000 pixels via interpolation. (If your program offers you a choice, use bicubic interpolation.) Once the resizing is complete, use the unsharp mask filter to sharpen the image (that’s right, the unsharp mask filter actually sharpens an image!). I recommend experimenting with the unsharp masking controls a bit, starting at 100 for Amount, 0.3 for Radius, and 0–3 for Threshold. First, increase the Amount setting, then increase the radius by 1/10 point at time. If you’re using Photoshop, you might try switching the color mode to Lab and apply the unsharp mask to the L level only. This sharpens detail without blurring colors. Other sharpening plug-in filters, such as the Sharpen Filter from Nik Effects, can be used to improve the look of an interpolated file.

For images destined for poster size, there are stand-alone programs and plug-in filters such as Genuine Fractals and VF Zoom, that use fractal algorithms to analyze an image and resize it 100–200 percent without obvious artifacts. But even these have their limits, and both work better if you use images from 3MP-or-higher cameras. So don’t expect to turn a 1.5MP image into a stellar 8x10-inch print.