Partial Pixels

Facebooktwittergoogle_pluslinkedin

Today’s Question: While I was watching one of your video lessons that covered sharpening in Photoshop, it struck me that you were using a value of less than one pixel to define the sharpening effect. How can a sharpening filter work on partial pixels?! Shouldn’t the value for the size (Radius) only be in whole numbers?

Tim’s Answer: I think the way I would describe this (understandable) source of confusion is that the number of pixels for the Radius setting is really a shorthand of sorts that goes beyond a simple number of pixels. In other words, you aren’t really having an effect that goes down to a level smaller than a single pixel, as that’s not possible with pixel-based editing. Instead, you are having an effect that is tapered across multiple pixels.

So, when you set the Radius to a value of a half pixel, you aren’t really having an effect on half of a pixel at a contrast edge, but rather are using a value of 0.5 as part of the (somewhat complex) formula that determines how much contrast is enhanced along existing contrast edges within the photo.

Keep in mind that the size of the sharpening effect (often referred to with a “Radius” value) interacts with the strength of the sharpening effect (often referred to with an “Amount” value). In addition, with many sharpening filters there are additional values you can adjust or factors that are used to determine exactly how the contrast enhancement of sharpening is applied to various areas throughout the image.

So, the Radius setting really relates to the size of the transition area for the contrast being enhanced. In some cases, with a very strong sharpening effect, the formula will cause a change with values that suggest granularity of one-tenth of a pixel. That’s not because the pixels are being divided into smaller units (they aren’t), but rather because the formula is blending the effect into the image. So in some ways you can think of the Radius setting as defining the degree of transition for the effect.

To get a sense of this, you can create a simple graphic in Photoshop. For example, create an image that is half black and half white, with a hard transition from black to white. Then apply a very slight blur to the image, so that there is a transition from black to white that spans perhaps ten pixels in the image. Then choose Filter > Sharpen > Unsharp Mask from the menu to bring up the dialog for the Unsharp Mask dialog (I recommend this filter for the test because it is a simpler sharpening filter than Smart Sharpen). Set the Amount to 500% (the maximum value) and the Threshold to 0 (the minimum value), and evaluate the effect in the image as you adjust the value for Radius.

If you conduct this little experiment, you’ll see that even a change of 0.1 or 0.2 pixels for Radius will have an impact on the sharpening effect in the photo, with higher values causing the gradation in the image to become smaller than they appear with a lower value for Radius. Again, this is just part of the complexity of the formula in play for this (and other) sharpening filters.