All photographs copyright Jay Torborg.

Scaling/Resampling

You may need to resample your image to size it for sending in email, displaying on the web, or for printing to a large size. For down-sizing (reducing the resolution), Photoshop's Image Size command works well (make sure interpolation is set to Bicubic). For upscaling however, this algorithm is not as effective.

The conventional wisdom is to scale your image to 300 dpi (or some similar fixed resolution) before printing on a photo inkjet printer. I don't recommend doing this. The print driver will resample the image as part of its screening algorithm no matter what resolution you send it, so I recommend scaling your image for the printer only if the resolution is less than 200 or more than 400 pixels/inch. 

Use the Image Size command to change to the size you want to print, but make sure the Resample Image box is unchecked. If the Resolution is between 200 and 400 ppi, I've found that you'll get a better print than if you resample.

The maximum resolution you can send to your printer will depend on the printer (actually the print driver). If the resolution gets too high, the print driver will not effectively average the pixel values and some information in the image will be lost. In the worst case, this will result in aliasing artifacts due to undersampling. You should experiment with your particular printer, but I've generally found that resampling images down to 360 ppi will result in very sharp images that print faster.

I've found that the best way to upscale (increase resolution) images that are less than 200 ppi is using an algorithm called Stair Interpolation. Fred Miranda has a free Photoshop action on his website for doing this. It can scale by 1.5x, 2x, 3x, or 4x, and does so by using multiple successive bicubic interpolations in small steps. Use the smallest scale factor to get above 200ppi for your printed image.

Unsharp Mask

Note that if you used the Photoshop Actions for digital camera noise reduction on the previous page, this action performs an unsharp mask as part of its process. However, many images can benefit from a little more unsharp mask after this process, particularly if you have resized the image as discussed above.

Once you've finished all the tonal corrections on your image, you can sharpen it using Unsharp Mask. This filter is one of the most frequently used but often misunderstood functions in Photoshop (or any other image editing application). Given that it is a sharpening tool, you may wonder why it has such a strange name.

The name comes from the darkroom technique from which the algorithm derives. The algorithm works as follows. A blurred (or unsharp) version of the image is subtracted from an original version to create a mask. Since the blurred version will have edges that are "spread out", when it is subtracted from the original you're left with just edge information. This mask is then added back to the original which increases the contrast of the edges and makes the image look sharper.

There are three parameters for the unsharp mask - amount, radius, and threshold. For those that are interested in the algorithm, the amount is a multiplier on the mask before it is added to the original (100% = 1.0); the radius determines how much of the neighboring pixels to use in the calculation of the blurred image (in the same way that the gaussian blur filter works); and the threshold determines the difference in value between adjacent pixels before the unsharp mask will be applied.

The practical application of unsharp mask is fairly straightforward. The radius is determined by how sharp the original image is. You can get a feel for this by looking at your image at 200%-400%. If you can see sharp transitions that span only one or two pixels (that aren't just film grain or noise), then you have a very sharp original and you should use a radius of 0.5 or less. This will be typical of images  from professional digital cameras and images that have been scaled down significantly in resolution. Images from consumer digital cameras and from most film scanners may not be as sharp and may need a higher radius filter to be sharpened effectively. For digital images that are scaled up as described above, the unsharp mask radius required will scale accordingly.

If you use a radius that is too large, you will notice obvious sharpening artifacts, particularly around high contrast edges. If you're not sure what radius to use, set the amount to 150% and adjust the radius upward until you just see a halo around the highest contrast edge in the image, then back down slightly.

The Amount parameter determines how much contrast will be added to the edges. If you've got the radius adjusted correctly, the amount should range from 100% to 200%. Make sure you're looking at the image at 100% scale and adjust the amount until you think it looks best. If you're going to be printing the image on an inkjet printer, you can often get away with a little over-sharpening since the printer will dither the image and a small amount of sharpness will be lost.

The threshold parameter is useful to reduce sharpening of noise and film grain. This parameter should range from 0 for a low ISO image from a pro digital camera, to 10 for a grainy film scan or high ISO digital camera image. The closer to 0 you can get away with, the more natural the sharpening will look since it will be applied consistently to the entire image.

Occasionally, the unsharp mask will result in color artifacts around the edges. This is because the unsharp mask works independently on each color channel. If this happens, you can convert the image to LAB color space first, then perform the unsharp mask only on the Lightness channel, then convert back to RGB. The Lightness channel is selected in the Channels window.

For the lion pair image, I used a radius of 0.3, an amount of 150%, and a threshold of 0.

Adding a Drop Shadow

I like to add a drop shadow to my images before posting them to the web. Here's how to do it.

From the Layer menu, select Layer Style -> Drop Shadow... (you can also get to the same place by double clicking on the layer in the Layers window). This will bring up the drop shadow dialog box. For the shadow on the image below, I selected an opacity of 75%, an angle of 120 degrees, a distance of 15 pixels, a spread of 0%, and a size of 15 pixels. You should experiment with these parameters to get the results you want.

Now that you've got the shadow, you can flatten the image to get rid of the layers, and then you can save as a JPEG for the web.

Results of Edits