Part 13 - Sharpening, Less is Better

For those just getting started with digital imaging software, the "sharpening" tool is often misunderstood and misused. This post will provide a basic understanding of what sharpening is, and what sharpening isn't. Two of the most important points to remember are:

• sharpening isn't a quick fix for out-of-focus images (out of focus images will always be out of focus)
• all digital images require some degree of sharpening.

So what is sharpening?

Software capable of performing sharpening uses sophisticated algorithms and programming that has matured over the past two decades to selectively increase the contrast along borders of different densities appearing in an image (for example, the edge lines between the veins in the flower petals below). The result is the "illusion" of increased image sharpness -- when in reality it's selective contrast that has been achieved.

Sharpening is one subject that is best explained using lots and lots of picture examples. Most technical articles about sharpening generally use professionally made, resolution test charts to show the effects of focus and sharpening. I'm using a real world picture to help beginners recognize sharpening effects as they would appear in their own prints or monitor. The picture below will be used to demonstrate the operation and effects of sharpening.

This flower picture may well be one of the most carefully set up and taken images I've ever shot. I wanted a picture that was as absolutely "in focus" and sharp as my camera could capture for this sharpening discussion. Here's how the picture was taken:

• outdoor lighting with a slightly overcast sky
  
• no wind to cause subject movement
• shutter speed, 1/250th of a second
• aperture, f/11
• ISO setting, 125
• camera mounted on a very stable tripod
• focus point, center of flower
• image saved in RAW format.
  

Once this picture was taken, I took a second picture of the same flower but just slightly out of focus:

The second image is just BARELY out of focus

To see the effects of sharpening, we need to examine these images more closely in detail. The red box on the picture below will be enlarged for both flower images to reveal the impact sharpening has on our photography:

Both images were stored as RAW files. Below is this portion of each image (in focus and out of focus picture) as it was brought into Photoshop without any sharpening applied in the RAW converter:

The "in focus" image with NO sharpening applied

The "out of focus" image with no sharpening applied

With the images enlarged, it's easier to see that the bottom picture is not quite in focus compared to the top photo. Remember these are the images brought directly from the camera's RAW files and opened in Photoshop.

Because of the limitations of our photo sensors and the way light is collected/recorded at each pixel site, it is necessary to apply a small amount of sharpening to EVERY image. For this example, all sharpening will be done inside Photoshop using the software's "Unsharp Mask" feature -- located inside the "filter/sharpen" menu. (You're right. "Unsharp" seems counter-intuitive. But it's a term inherited from the world of the printing press.)

Unsharp Mask control window next to our flower image

It is very easy to over sharpen an image (more on over sharpening later). As a starting point, try these settings then click OK. Notice, if you check the preview box the sharpening effect can be seen on your image before you click OK. In most cases, this is just enough to overcome the shortcomings of the camera's sensor without leaving any unwanted traces (called artifacts) in the images. The result of applying this degree of sharpening to our two flower images is shown below:

"In focus" flower with normal sharpening applied

"Out of Focus" flower with normal sharpening applied

Both flowers look better. The in-focus flower is very sharp and shows all the detail in the center of the flower and nearby petals. The out-of-focus flower appears to be more in focus, but still is not as sharp as the flower that was shot in focus. This relationship of apparent focus between the two flower images will ALWAYS exist.

So what happened to the image when sharpening was applied?

Unsharpened image enlarged to show individual pixels

If we enlarge the picture further, the individual pixels that make up the flower image can be seen. This first picture shows the "in focus" image BEFORE sharpening was applied.

Same image after sharpening and enlarged to show individual pixels

Photoshop searched the first image to detect the lines that identify the border between areas of different densities. Once these borders were located, the software increased the contrast level between the two adjacent areas to create the sharpening illusion. If you look closely at each picture you will see areas where one pixel has been lightened while the adjacent pixel has been darkened -- as compared to the unsharpened picture.

It would seem that if more sharpening was applied then the out of focus picture could be salvaged and made to appear in focus. There is a point of diminishing returns with sharpening. Apply too much sharpening and image quality suffers as well as becoming visibly noticeable and distracting in the final print.

Visible artifacts become distractions when too much sharpening is applied

Above is the out of focus picture for the last time. It has been sharpened to the point where it approximates the sharpness of the original "in focus" picture. This much sharpening is too much. Artifacts or imperfections are noticeable along the sharp edges of the elements that make up the picture. (This is a very small picture and at monitor resolution, but as an actual print you would be disappointed.)

In the end, the sharpest prints start with original camera files that are "in focus". As far as sharpening is concerned, like so many other things, "less is better".

If you save your files in JPEG format in your camera, I recommend setting the camera's internal sharpening feature to minimal or off. It's my experience that the sharpening functions in today's imaging software do a better job.

If you have questions or comments, please let me know.

Part 12 - From Pixels to Print

We know what pixels are. We also know that the camera doesn't produce a picture but instead a data file that describes the specific amounts of the primary colors (red, green and blue) for every pixel that comprises the picture.

So, if we really don't have a picture, how is the picture displayed on our computer monitor and then as an image from our inkjet printers? Key to understanding these technological miracles is remembering that there are two ways of reproducing the visible spectrum:

• By using the three additive colors (also called primary) -- red, green and blue -- in nearly infinite density combinations to replicate all visible colors
• By using the three subtractive colors (also called secondary) -- cyan, magenta, yellow -- in finite density combination to reproduce all visible colors.

The iDarkroom uses both methods:

• The computer monitor displays images using the Primary Colors -- red, green and blue.
• The inkjet printer produces images using the Secondary Colors -- cyan, magenta and yellow (and adds a black ink to provide contrast and a truer black than can be produced by man made cyan, magenta and yellow inks together).

Click here to review my earlier post Color Primer for more details.

The following graphic borrowed from Wikipedia is the best illustration I have found to help understand monitor vs. printer methods of creating an image as well as the relationship between pixel's per inch (PPI) and dots per inch (DPI):

Each grid represents a 10 by 10 pixel area of a much larger multi-million pixel image

Starting with the monitor

The data contained in the original image file is saved in terms of red, green and blue per pixel (a value between 0 and 255 for each primary color). As shown above, a typical sRGB computer monitor is capable of displaying all 256 values of red, green and blue at each pixel location. At 256 x 256 x 256, this means each pixel can display all 16.7 million colors that a typical DSLR's image sensor is capable of recording.

This display approach is straightforward. The only additional math or adjusting of the monitor that must be done (via the ICC profiles used by your image editing program) is to take into account the color space of the original image file. If you've set your camera, for example, to use the Adobe RGB color space, the computer will adjust the pixel color values to present the correct Adobe RGB colors on the RGB monitor.

Moving on to the inkjet printer

The "original image file" to "inkjet print" relationship is more complex for several reasons:

• Inkjet printers use the secondary colors to reproduce the visible color spectrum -- cyan, magenta and yellow. So, RGB values contained in the original file must be converted to secondary colors as well as to the color space of the printer/paper combination.
  
• Inkjet printers place "dots" of color on paper, but these dots are limited to the color cartridges in your printer. In today's photo inkjet printers, this typically means 8 different colors. For example black, photo black, matte black, cyan, light cyan, magenta, light magenta, and yellow are used by many entry level photo printers.
• Printer resolution is expressed in terms of the number of "dots" it can place within one inch. Dots-Per-Inch (DPI).
• Inkjet "dots" of ink are necessarily much smaller than a pixel. (Explaining why printer DPI numbers are so much larger than the PPI of the image file.) This is critical since the "dots" of ink can only be one color and one density. In order to provide the illusion of continuous tone, tiny dots of ink are placed in extremely close proximity to each other to provide the subtle shades of color required in a photograph.

In fact, the dots placed on paper by an inkjet printer are so small and so close together that our eye cannot distinguish the individual dots and colors without considerable magnification. Our brain visually blends these areas to produce a shade of color. (Much the same way modern military camouflage blends and marries into the surrounding natural colors to become part of the background.) The right hand portion of the illustration above reveals that only secondary ink dots are present and that there are many, many "dots" of secondary color in each pixel area. The higher the DPI, the more dots a printer can produce per inch and the more detailed and continuous the printed image appears.

Once again, the multiple variables involved in the process of creating a print -- camera, to monitor, to final print -- are apparent. And, once again, it's obvious with all the necessary conversions and adjustments taking place behind the scenes that -- even if we make NO image adjustment of our own -- unless the entire workflow is calibrated and in control, the odds of any picture being an accurate reflection of the original scene are slim.

But knowing what's happening throughout the workflow and taking the necessary steps to calibrate your iDarkroom are the first steps to stunning prints.

If you have any questions or comments, feel free to pass them along.