Camera Lens Resolution Measurement using a Razor Blade
When you read about lens sharpness results on the internet, can you believe them? Maybe. I believe that you see more variation in lens resolution measurements at different internet sites than in the lenses themselves.
What follows is a technique to inexpensively get some very accurate resolution numbers.
At the core of my lens resolution testing is the program MTFMapper, written by Frans van den Bergh. His software, documentation, and printable test chart files can be found here.
The MTFMapper program can do many different things related to camera lens testing, which includes resolution, contrast plots, sensor alignment, chromatic aberration measurement, and focus analysis. I’m only going to discuss resolution measurement in this article.
There are many elements that are required to get good, repeatable lens resolution measurements. Normally, these elements include the following:
Stable camera/lens support
Quality test chart
Even test target illumination
Proper target alignment
Precision linear rail with micrometer adjustment
Un-sharpened, raw-format photos of test target
In this article, I’m going to discuss a way to avoid the need for the “quality test chart”. Frans’ MTFMapper program will let you get by with only a single high-contrast edge to measure, which can be supplied by a simple razor blade or utility knife blade. A good reason that this option exists is because it isn’t trivial to get a large, accurately-printed, quality media, and properly-mounted test chart. I inspected a blade edge under very high magnification to convince myself that it had no defects or roughness that would degrade the measurement results of the program.
A Sharp Target Edge
The ideal lens resolution target is a chart with many edges that can be measured by the testing program. Although printable files are supplied with the program, those files still need to be printed and mounted. How big should it be printed? How good of a printer is needed? What kind of material should be used for the chart? How do I mount the printed test chart? These issues have kept many people from even attempting to make their own lens resolution measurements.
A back-lit blade target
It might sound a little crazy, but a really good target to test lens resolution is a razor blade, photographed in silhouette. The MTFMapper program has the ability to measure a single straight edge in a photograph. If you can provide a high-contrast, in-focus, very straight edge, then you can get a really good resolution measurement of that edge.
In the photo above, you can see an example of how to mount a blade whose flat face is parallel to the camera sensor. I use forceps held with a clamp. If you look closely, you will notice a number superimposed over the razor’s edge that says “37.6”. This number is the MTF50 resolution value, measured in line pairs per millimeter (this is a very old lens that isn’t as sharp as modern lenses). A light was placed to cause the blade edge to be in silhouette, since you want a high-contrast edge to get a quality resolution measurement.
The edge to measure can be placed anywhere in the photograph, which is handy when you’re interested in resolution in maybe a corner of the frame. You can also orient the edge to enable measurement in meridional (tangent) or sagittal (wheel spoke) directions. If your lens has a curved focus plane, then you can focus on the blade wherever it is located in the frame to get a better resolution measurement.
An optimal edge angle is 5 degrees or 85 degrees, but the program is fairly flexible. Please avoid perfectly vertical or perfectly horizontal edges, due to the way the programs’ mathematics work. You don’t have to use the whole edge, either; you can measure just a piece of it.
Take a photo of the edge in RAW format. You might need to convert the photo into DNG format, using the free Adobe DNG converter program, if MTFMapper doesn’t understand your camera’s raw format. Don’t ever give MTFMapper jpegs!!!! Since jpegs have varying amounts of sharpening applied, the subsequent resolution measurements using these files will be totally bogus and useless.
The most reliable results are obtained by using a remote release or the self-timer and an electronic shutter or mirrorless camera and manual focus. If you take a few shots with the same setup but get different measurements, then the first suspect is unwanted vibrations.
Edge measurement in MTFMapper
To perform the resolution measurement, you need to do the following:
In MTFMapper, first make sure the Settings,Preferences has the pixel size of your camera sensor (measured in microns).
Select File | Open with manual edge selection…
Browse to your camera’s RAW (or DNG) file of the blade
The “Select one or more edge ROIs” dialog opens up (shown above)
Left-mouse-click the beginning of the blade edge to measure
Left-mouse-click the end of the blade edge to measure
A small cyan rectangle is drawn over the region to analyze.
Click the Accept button
Open the resulting “annotated” file on the right-hand side of the program.
If your lens is in proper focus and your camera is steady, then you should now have a very accurate measurement of the lens resolution at that location.
Aye, there’s the rub. If your photo of the edge to measure isn’t in proper focus, then the results turn into garbage-in-garbage-out.
Camera mounted on a linear slide with micrometer
You may think that using contrast-detect focus and/or focus peaking will get you perfect focus. Think again. It’s true that mirrorless cameras generally yield vastly better focus, but optimal resolution measurements actually require millimeter-level accuracy. Manual-focus lenses are even harder to focus properly.
Shown above is a camera on a linear-travel slide controlled by a micrometer that is capable of repeatable movements as small as 0.01mm. Using a system such as this, it’s easy to make a series of photos that start before the expected correct-focus zone and then travel up to and finally past the zone of correct focus. Just let MTFMapper analyze each shot to let you find the peak resolution; it’s rarely in the shot you anticipated.
I like to use focus-peaking to at least get the lens into the general zone of correct focus while in the middle of the focus rail travel. I manually focus with the least sensitive focus peaking level (1), and with the most screen magnification that still displays peaking. Sometimes I’m forced to use the ‘medium’ focus peaking level, because the magnified screen doesn’t display any peaking feedback.
I have found that some lens/aperture combinations can show focus changes (and resolution changes) by movements as small as one millimeter, even when focusing on targets a few meters away. The MTFMapper program is sensitive enough to detect the smallest focus (and resolution) changes, well before your own eyes can detect it. Most lens focus rings are much too coarse to change focus by these small amounts, and autofocus is almost never this precise, either (even using pure contrast-detect).
These fine changes will be lost if you don’t also trip the shutter using remotes/self-timers/mirror-lockups, etc. to rid any vibration. And turn off lens vibration reduction.
It doesn’t have to be very expensive or complex to measure your lens’ resolution. You mostly just need to have the patience to carefully align your target, control the lighting, and nail the focus. And never use jpeg for measurements.