Image Differences

Now that we've explored the various imaging methods, we should recap and highlight some of the different techniques you can use in building images suitable for output on halftone and contone devices. We say "recap," because we've mentioned most (if not all) of these in previous chapters, though never in one place.


The first and foremost difference between contone and halftone imaging is the required image resolution. It's quite a bit harder to work out the resolution needed for halftone output than it is for contone, so we'll deal with halftone output first.

  • Resolution requirements for halftone output. The resolution of the output device isn't directly relevant in determining the resolution you need for the image. It's the halftone screen frequency that matters. You never need an image resolution above two times (2x) the halftone screen frequency (and often you can get almost-equivalent results with as little as 1.2x or 1.4x). That means that even if you're printing on a 2400-dpi imagesetter, your image resolution can (and should) be much lower. For instance, printing at 150 lpi, you never need more than a 300-ppi image, and usually no higher than 225 ppi (we generally use the 1.5 multiplier; see Chapter 3, Image Essentials.)

  • Resolution requirements for contone output. The resolution needed for a contone output device is easy to figure, but it can sometimes be hard to deliver. Your output resolution should simply match the resolution of the output device. If you're printing to a 300-dpi dye-sub printer, your image resolution should be 300 ppi. When printing to a 4 K film recorder, your image should have a horizontal measure of 4096 pixels, or about 60 MB for a 4-by-5 print. An 8 K film recorder really wants 240 MBan 8192-by-10240-pixel image.

    In truth, many high-resolution film recorders are more forgiving, and you can halve the resolution. For instance, we know of few people who actually send a 960 MB image to a 16 K film recorder, and we know quite a few who get good results sending a 60 MB file to an 8 K film recorder (about half the amount of data it "requires"). Sending less than a full 60 MB to a 4 K film recorder, however, is a much more marginal proposition. Make sure, though, that you send an integral multiple of the device's resolution. If you send 4095 pixels to a device that wants 4096, it'll either barf when it gets the file, or you'll get some very strange interpolation artifacts.

    The appropriate resolution for stochastic screening is less clear, but in general you rarely need over 300-ppi images.

  • Resolution requirements for inkjet output. The necessary resolution for today's photorealistic inkjets is to some extent a guessing game. In part, it depends on the paper stockmatte papers generally require less resolution than glossy ones. Anecdotal evidence suggests that a resolution around 240 ppi is sufficient for most images, but if you're really picky, you may want to determine the ideal resolution for a particular paper stock yourself using good old trial and error. It's certainly possible to send too much data to an inkjet printer, not only increasing print times unconscionably but also degrading the image: you do not want to send a 1440-ppi image to a 1440-dpi inkjet!

    Synthetic targets composed of black and white line pairs show an improvement when they're printed at an integral divisor of the printer resolution, such as 360 ppi on a 1440-dpi inkjet, but it's uncertain how applicable this is to images with more natural content. Bruce usually prints to 1440-dpi inkjets at 360 ppi if the image contains enough real pixel data to start with. On small prints (8-by-10-inches or less) he may even send 480 ppi and print at the much slower 2880-dpi setting. However, although Bruce insists the benefit is real, it takes careful sharpening to see it. If the image doesn't contain enough pixels to print at these fairly high resolutions, we'll simply send what we have rather than interpolating.

Tonal and color correction

We talk a great deal about compressing tonal range ("targeting") for halftone output in Chapter 6, Image Adjustment Fundamentals, so we won't go into it here. Contone output needs less in the way of tonal and gamut compression than halftone output, because contone devices generally have a greater dynamic range and a wider gamut than do halftone devices. However, this can bring its own problems, particularly when you have a scanner with a tendency to oversaturate some colors, as do many inexpensive scanners (and even some expensive ones). Keep a watchful eye on saturated colors. Some dye-sublimation printers feature a magenta that's almost fluorescent!


As we noted back in Chapter 9, Sharpeness, Detail, and Noise Reduction, contone images need less sharpening than halftone images. But that doesn't mean they don't need any at all. Halftones, again because of their coarse screens and significant dot gain, mask details and edges in an image; sharpening can help compensate for both the blurriness of the scan and the blurriness of the halftone. And, halftones being what they are, you have a lot of room to play with sharpening before the picture becomes oversharpened (most people end up undersharpening).

In contone images, however, there's a real risk of oversharpening. Not only should you use a lower Amount setting for unsharp masking, but also a smaller Radius. Where a Radius less than one is often lost in a halftone image, it's usually appropriate in contone images. In this context, inkjet printers tend to behave more like contone devices than halftone devices.

Image mode

This last item, image mode, isn't really dependent on what output method you're using. However, because we still see people confused about image mode, we thought we'd throw in a recap here, too.

Again: if you're printing to a color contone device that outputs to film (or if the image is only seen on a color screen), you should leave your image in RGB mode. Contone and hybrid devices that print on paper use CMYK inks or toners, but in most cases you'll get better results sending RGB and letting Photoshop or the printer handle the conversion. If you have a good profile for the output device, you can preview the output using Proof Setup, and convert the image from your RGB editing space to the device's space at print time (we discuss this in the next section). If you're printing separations, though, you need to send a CMYK file.

We've tried many times to build Photoshop Classic CMYK setups for CMYK dye-sublimation printers, but it simply doesn't work. Photoshop's separation engine is geared toward halftone output, where the ink density remains constant and the dot size varies. It simply can't handle the variable density on dye-sublimation printers. It would work for inkjets if we could control the inks directly, but since we can't, it doesn't.

Real World Adobe Photoshop CS2(c) Industrial-Strength Production Techniques
Real World Adobe Photoshop CS2: Industrial-strength Production Techniques
ISBN: B000N7B9T6
Year: 2006
Pages: 220
Authors: Bruce Fraser © 2008-2017.
If you may any questions please contact us: