Sharpening – Monster Under the Bed – Debunked

Please keep in mind that in the artistic chain of creating a great image, a processing technique is only secondary to your creativity and your emotional connection with the subject. You should spend most of your time in the field concentrating on composition and light. Only when you master this part, can you complement your image with subtle processing techniques. The simpler and faster the method, the more time you will have to create great photographs. Don’t fall into a hole of never-ending alterations. Get the simplest processing software you can get. Playing for hours with your imagery in Photoshop won’t make you a better photographer; in fact, quite the opposite.

    

There is no other topic that garners more attention in digital post-processing than sharpening. In fact, the question about “how to get sharp photos” is one of the most common we receive. What some photographers don’t realize is that the quest for sharp images doesn’t begin in front of the computer. It starts much earlier.

It begins in the field. Simple things like the way you hold the camera could have a huge impact on your photos. We admit that we are not big fans of tripods. Most of our work has been shot freehand. For us the freedom of movement and creativity that flows from it is much more important than laboratory-sharp photos. Having said that, even while shooting freehand, there are ways to get your photos sharp.

• Avoid using the LCD at the back of the camera and instead use your viewfinder. Press the camera tightly to your body for stabilization.
• Make sure your focus point is exactly where you want it to be.
• Most importantly, pay attention to the lenses you are using. It is so common to see people with huge megapixel-count cameras with cheap zoom lenses attached to them. Get high quality lenses first; only then worry about the camera. For example, the Fuji primes are very sharp, so less sharpening may be required.

All right. You are back at home with your stunning photos and they are ready for processing. Let’s get started.

The sharpening techniques listed below are discussed in the context of the Fuji X-series cameras but they apply to any camera. These are only suggestions and the technique could vary depending on personal preferences and subject matter.

In order to provide you with the best information, we went straight to Brian Griffith, the creator of Iridient Developer. Here is the information he shared with us:    

All of the methods in Iridient Developer are very high quality implementations done with 32 bit/channel floating point processing using a special perceptual colour space where the colour/chroma is separated from the luminance detail to avoid colour artifacts. All of them can give very, very good results; finding what works best for you just takes experimentation and practice. To get a feel for the “character” of the various methods, it helps to push the adjustments to the extremes (right end of the sliders) and then maybe 1/2 and then fine tune from there. Most people will quickly find a method and a range that works for them.

Note: the four major methods of sharpening found in Iridient Developer (and in some other programs) are discussed in the context of the characteristics of the X-Trans cameras. Brian writes:

The X-Trans cameras lack anti-aliasing filters and use a sensor that has more “green” coloured photosites, which tends to have a strong correlation with luminance detail versus colour and is often used with prime lenses that are quite sharp. These three factors all contribute to the X-Trans cameras in general being quite sharp to begin with and needing less aggressive sharpening than most other digital cameras. For any of the sharpening methods that take a radius parameter (USM, DoG, RL) you’ll want to keep it fairly small, as there just isn’t much blur in these images to begin with. For RL Deconvolution I recommend around 0.30 to 0.5 for the X-Trans cameras and this is a decent starting range I think for USM and DoG too.

Unsharp Mask

This is a very, very popular sharpening method. Some people have used unsharp masking in Photoshop for many, many years and are just incredibly comfortable with this method and its parameters. There are numerous books, magazines and online articles that have been published over the years that cover USM so it tends to be one of the better known and understood methods. Using too much amount or too big a radius can lead to “halo” artifacts, a little of which can actually sometimes be a good thing for printing. But when viewed at high resolution on screen, “halos” can quickly get a lot of criticism online as “poorly” done sharpening.

Hybrid

Super fast, only two sliders and can often get away with just adjusting the sharpen strength so this is probably the easiest to use and is the default option for most cameras including the X-trans models in Iridient Developer. This method can also produce “halos” like USM when overdone, but tends to be subtler.

DoG

A Google search for “difference of Gaussians” will turn up lots of info on this one. WikiPedia says, “Difference of Gaussians algorithm is believed to mimic how neural processing in the retina of the eye extracts details from images destined for transmission to the brain.” It’s a popular form of edge enhancement for microscopic imaging. This and the hybrid are quite similar. DoG provides a bit more fine control over the parameters, and has more slider adjustments than Hybrid. Typically you’ll want the Sharpening Radius to be greater than the Noise Radius by a factor of 1.5 to 3.0 or so… it can be used to approximate a Laplacian of Gaussians filter (yet another edge-sharpening method).

R-L Deconvolution

It’s more of a “deblur” process than the others, which tend to be more edge-contrast enhancing. Can really improve fine micro detail like natural textures (sand, stone, leaves, hair, etc.), artificial textures and fine patterns much better than the traditional edge-sharpening methods, but if you are really looking for crisp sharp edges, it may not quite give that same super sharp look either. Can avoid some of the “halo” type over-sharpening artifacts unless really super overdone, but can have some “swirly” sort of “false grain” artifacts of its own.

Some people really, really like the look of RL Deconvolution. In some programs, similar deconvolution-type sharpening methods may be termed “smart” sharpening. If you are after a more natural look of detail without quite so much edge sharpness and contrast and really want to bring out very fine texture details (hair, eyelashes, leaves, grass) RL Deconvolution is one I would definitely recommend.

The RL Deconvolution is the method we use. After experimenting with different tools we found this process the best in treating the X-Trans sensor files. With the most recent X-T1 and Fuji X100S cameras we use 0.39. Indeed, we confirm Brian’s point that this method significantly improves micro contrast and details.

Finally, let me finish where I started. Don’t pull your hair out over sharpening or any other processing dilemma. First, make sure you capture a strong photograph. Pay attention to your subject, composition and light. Only then can you complement your visual masterpiece with some processing technique, if it’s even needed.

All images presented here were captured with the Fuji X-T1, XF 14mm F2.8 and XF 56mm F1. 2.

Sincere thanks to Brian and his team at Iridient Digital for sharing his insights about this important subject with us.

 

2014 © Olaf Sztaba Photography. All rights reserved.