The Importance of Focus Precision
Sharp photos depend upon sharp focus. You might be very surprised at just how sensitive your lens can be to focus changes. I wanted to show you an experiment that gives very precise numbers on how the resolution changes with errors in focus.
Nikon Z8 camera mounted on a linear slide
As shown above, I start by mounting my camera onto a linear slide. This slide can be moved with a micrometer in very small steps, so that I can shift my cameras’ focus very precisely. I conducted these tests using a 135mm lens at f/2.8 mounted on my Nikon Z8. Note that this isn’t a particularly fast lens, but even at f/2.8 you’ll find that focus precision matters enormously.
Using the MTFMapper program created by Frans van den Bergh, I repeatedly photographed a new utility knife blade at different distances and then processed the photos in his software. I focused the lens only once, while the linear slide was near its midpoint (12mm), before starting the test. I could have of course used a more conventional focus chart to get the resolution measurements, too.
A utility knife blade in silhouette
The subject, shown above, is the edge of a very sharp and straight knife blade. The software doing the analysis is capable of analyzing a single edge that you specify. To get the best results, the edge should have high contrast; I used a light to make the blade show up in a silhouette.
If you look very carefully, you can see the little number 37.6 shown on top of the blade edge, which is where the software made the resolution measurement.
Since MTFMapper uses LibRaw to decode raw files, it uses zero sharpening (sharpening would falsely increase resolution measurements). For raw formats that LibRaw doesn’t support (such as the Z8/Z9 high-efficiency raw), I use the Adobe DNGConverter to make DNG raw files; these files also have zero sharpening applied. The downside to this DNG converter program is that it strips out some exif data, such as the focus distance.
Resolution versus focus distance
As shown above, I made a plot of the measured resolution of the blade edge photographs at different distances. I had attempted to focus the lens while the camera was placed at a setting of 12mm on the linear slide rail. The measurements show that in fact the sharpest photo was at a position of 15mm, where I got an MTF50 resolution measurement of 37.6 lp/mm.
The entire range of the focus testing shown is only about 1 inch (27mm).
I had missed focus by only 3 millimeters, while my subject was at a distance of 7 feet (2.13 meters). The MTFMapper program is able to tell the difference in resolution even with a 1 millimeter focus error! I had used “focus peaking” with a magnified view and manual focus to get the best focus I could manage. The camera focus-peaking feedback (set on ‘low sensitivity’) got me to within about 3% of optimal focus.
Granted; these resolution differences are finer than what you can probably perceive yourself, unless you need to crop or print big. Also, telephoto lenses are far more sensitive to focus errors.
Image sharpness is more sensitive to focus than most people could imagine. This little exercise shows why people that measure lens resolution have to be so careful in controlling focus (and vibrations), or else their measurements are just wrong.
In a more general sense, you want those feathers, hairs, and eye lashes/reflections to be totally sharp. The best lens you can buy won’t give you that unless you also nail the focus. A cheap lens that is correctly focused will usually give better results than an expensive lens that is slightly out of focus. I have found that my mirrorless cameras achieve more accurate autofocus than my DSLR cameras, and my lenses don't need focus calibration on mirrorless, either.