What Causes "Soft"?

INT South-Africa D900 0001 bythom

A complaint I see often is this: "my images look soft." This is usually followed by "did I get a bad copy of [lens or camera]?" While on occasion out of curiosity I'll ask for a sample image that supposedly displays this softness, I've pretty much stopped doing that these days. 

Why? Because I've almost never found that the customer had gotten a "bad copy". It was always something else causing the softness. 

One problem is that the Internet has proven to be a great way to start conspiracy theories, promote disinformation, and incite paranoia. Which then in turn makes people fearful that they have a problem. I’ve had people tell me that "Nikon is intentionally providing poor products" (not true), while the fanboys are out there trying to make themselves feel superior by spreading "Nikon isn't as good as [X; but typically Sony, which should tell you something by itself]" statements (also generally not true). 

In an attempt to provide useful information, I'm going to run through a number of things that can generate "softness" in an image. This is by no means a complete list, but in order to get to the "I got a bad copy" conclusion you pretty much have to eliminate all the following things:

  • Antialiasing — All digital cameras antialias when rendering final pixel values, and cameras with low-pass filters do so more than ones without. The "turn off sharpening in Adobe” notion to see apples-to-apples comparisons in raw conversion is nonsensical, by the way. Any knowledgable post processor knows they have to adjust for antialiasing in their raw conversions, and what they have to do will be different for various cameras/sensors due to differences at the sensor itself (pixel pitch, low pass filtration, etc.). Way to reduce: use "capture sharpening" techniques as described by Fraser, et.al.
  • Noise reduction — While some of the post processing noise reduction engines (e.g. Topaz AI DeNoise) go to great lengths to avoid acuity destruction while removing noise, all of the in-camera and most of the post processing techniques people are using for noise reduction will have an impact on detail and edges. Way to reduce: use lower ISO values and lower High ISO NR values.
  • Compression — With JPEG compression you're throwing data away, sometimes considerable amounts of it. The good news is that cameras have gotten far better at making that not obvious. I'm flabbergasted to see how good Nikon Basic compression looks compared to Nikon Fine compression these days. You really have to pixel peep (see below) to detect any differences. But the one difference I do tend to still see is that as detail gets near the Nyquist frequency, high levels of compression can get messy, and often that mess produces a bit of blur, which some might interpret as "unsharp." Way to reduce: use higher quality compression levels.
  • Motion (subject) — As I outlined 25 years ago in my seminal Nikon Field Guide, you need to pay attention to some specific shutter speed baselines to eradicate subject motion. A rule of thumb is that shutter speeds slower than 1/60 won't remove even minor human subject motion. Those of us who photograph sports tend to have sport-specific minimum shutter speeds we use to "stop" motion. Ditto wildlife photographers. Way to reduce: use higher shutter speeds or slow the subject down.
  • Motion (camera) — The ubiquity of image stabilization has people thinking that camera motion is a thing of the past. It is not. When you see "5 stops CIPA reduction", that's on a mechanical platform that vibrates/moves in a particular pattern, and the 5 stops should, at best case, be thought to mean "if you can get clear, sharp results without IS at 1/500 second, you might get usable results at 1/15 second." Note the two words I put emphasis on. IS is not an absolute, it's an improvement over it not being used. Way to reduce: use higher shutter speeds or stabilize the camera better.
  • Shutter shock — Typically this shows up as a "bounce" or "echo" on edges and detail. On DSLRs, the mirror flip had a tendency to add shock, but on mirrorless cameras the "double clutch" of the mechanical shutter definitely can be a problem. Shutter speeds from 1/15 to 1/250 tend to be the ones where you see shutter shock easily, but other shutter speeds may have a component of it in their results, too. Way to reduce: use First Curtain Electronic Shutter; on the more sophisticated cameras there are usually other options, as well (mirror lockup on top DSLRs, for instance). 
  • Missed focus (1) — It used to be that this was always my fault (e.g. pre autofocus). These days, most of us just blame the camera ;~). Quite a few sins lie in this category, so it's often one that I end up having to spend a lot of time helping users remove their incorrect assumptions and overconfidence in the automated aspects of the system. Lately, eye detect has created a whole new category of missed focus, where the focus is on the eyelashes, not the pupil. But autofocus has always had issues that can slightly (or worse) miss focus. Tracking motion on some cameras is not 100% precise, sometimes because the camera maker assumes that focus within the depth of field (improperly calculated by most makers) is good enough. Way to reduce: learn the focus system for your camera, know how to confirm focus, and when and how to override what the camera is doing.
  • Missed focus (2) — Field curvature and focus shift can cause a camera to focus at a different point than you expect. Focus shift, in particular, is the cause of a lot of “soft” comments, because the user didn’t even know that focus shift can exist, let alone how to correct for it. Way to reduce: learn whether your lens has focus shift in it when stopping down apertures. Learn whether you camera will adjust for that (DSLRs won’t, some mirrorless will, up to a point). Learn what to do if you are using a lens with focus shift.
  • Improper AF Fine TuneIf mount(s), focus sensor position, mirror position, or even lens motors are out of adjustment, the camera may be placing focus improperly because it’s assuming that those things are correct. Way to reduce: use the AF Fine Tuning adjustment in the camera.
  • Wide open — This used to be more common than it is today, but traditionally lenses tended to be sharpest stopped down two stops. That was particularly true of fast prime lenses. So when you tested a fast lens wide open (and on a close test chart, see below) the results were indeed softer than the lens was capable of. Way to reduce: stop down the lens. Mirrorless users may be able to find fast aperture lenses that do better wide open than traditional DSLR lenses.
  • Low MTF — A relative to the above (Wide open), many older lenses simply have a lower overall MTF (all else equal) and that will show up as a lack of contrast and acuity. Way to reduce: get a better lens, or stop down to the best MTF aperture for the lens.
  • Too close — One of the most common problems I see is that someone tested their new gear at a comfortable indoor distance for them, often far closer than they'd actually use the lens (especially true for wide angle landscape use and telephoto wildlife use). After all, you'd need a really big test chart for a 20mm lens to see any differences at its normal use distance, and you don't have a chart that big. So the common practice for wide angle lenses is to frame the entire chart (which probably isn't over 36" long axis) with the lens, which puts the test distance considerably below where the lens was optimized for. With telephoto lenses, no one wants to set up a test chart across the street, so they have the chart far closer to the lens than most of the subjects they'll be taking photos of. Way to reduce: back up and use the right chart. Test at the distances you actually photograph at, in the light you actually photograph in.
  • Too far — While Nikon used to make some telephoto lenses that had pronounced softness at distance (the 200-400mm f/4, for example), in general the biggest problem with long distance images tends to be atmospheric conditions between the camera and subject being photographed. Way to reduce: get closer.
  • Pixels (unpeeped) — Here's one that tends to catch people by surprise: if you're not looking at your pixels at 100%, exactly what is causing them to be resized and how is it doing that? I find this problem most often happens when someone selects "Fit to Screen" for viewing. When that happens, something is resizing the data, and how well or how poorly it does that can be an issue. With Adobe products, 50% and 100% are really the only magnifications you should probably be using. Anything else tends to mask what is actually going on in the pixels unless you know what to look for. (Related: if you see something on the Web that "looks soft," don't immediately blame the equipment that took the image. My site and most other sites on the Internet now have to use an image manager that resizes our images to your window/device. You can thank Google for that, because they’ll demote a site in searches if it doesn’t correctly [according to Google] resize images for smaller screens, like those on phones. If you want to see something closer to actual pixels from my site—caution, images have already been downsized—right click the image and choose "Open Image in New Tab." Then look at the image in the new tab. You’ll see the reduced size image the way it is supposed to look.) Way to reduce: as noted, find the sizing that has the least impact on visual appearance.
  • Pixels (over-peeped) — A corollary to the above is when the user starts pushing the pixels to 200% or 300% so as to see exactly what's going on. Unless you're trained to understand what happens when you over-peep and know how your software is handling that, you're going to see, well, some blurriness. Because your system is inventing pixels that aren't there, and how it does that can cause issues that look like softness. Way to reduce: don't over-peep.
  • Filter useOne of my pet peeves is that I keep finding people using US$2000+ lenses who are putting US$10 filters on the front of them, ostensibly to "protect" the lens. "Needing a filter" is an old dealer up-sell that netted them more profit while making your photos worse. Maybe you'd even come back in and buy "better" gear ;~). You don't need a UV filter; your camera has a UV filter built-in. You don't need a skylight filter; your camera can deal with skies just fine (though I'm not a fan of Fujifilm's tendency to make skies go cyan in their JPEG rendering). Coatings on the front of lenses can deal with most non-severe touches. The front glass on most lenses is highly resistant to breaking (though that filter isn't, and can create jagged edges that scratch the coating on the front element. Yet every filter I've tested adds a level of veiling glare (lower contrast) to your results. Some considerably so. About the only exception to that is circular polarizers, which are reducing a form of glare due to a change of orientation of the light waves themselves after reflection. In case after case, I've pulled a filter off someone's lens where they were complaining about softness, and the softness disappeared. Way to reduce: use filters only as necessary, and buy high-quality ones.
  • Back of camera evaluationBefore Nikon made some changes to how they show images on the rear screen under magnification, people would race to the highest magnification (see over-peeped, above) and declare the image to be too soft. "Missed focus" they'd mumble. Nope. Two things: (1) they over-peeped; and (2) the thing they're looking at came from the embedded JPEG in the camera, and sometimes from a reduced size JPEG in the older cameras. Note all the things that JPEGs can cause, above. It doesn't help that you're looking at a striped LCD display that has the pixels of a SVGA monitor, much of the time. So many things are going on in the camera trying to show you that enlarged view that you can't really trust it for serious evaluation. Way to reduce: look at the real image under controlled conditions.
  • Image stabilization (IS) aliasingThis typically happens one of two ways: (1) you’re using IS (VR in Nikon parlance) on a tripod and the system isn’t well suited to absolute steady so the motors jitter a bit; or (2) you’re using really fast shutter speeds that are high enough that the slower IS movement is not keeping up. Way to reduce: turn IS off when you’re on a tripod, turn IS off if shutter speed is above 1/1000 second. 

I'm pretty sure I'm missing quite a few possibilities. But the ones above are enough to get you started on your hunt for the cause of the softness you think you see.

This article originally appeared in News/Views, and has been archived here for easier access.

Looking for gear-specific information? Check out our other Web sites:
DSLRS: dslrbodies.com | mirrorless: sansmirror.com | Z System: zsystemuser.com | film SLR: filmbodies.com

bythom.com: all text and original images © 2022 Thom Hogan
portions Copyright 1999-2021 Thom Hogan-- All Rights Reserved
Follow us on Twitter@bythom, hashtags #bythom, #sansmirror, #dslrbodies, #zsystemuser