Comparing Two ‘Identical’ Lenses: A Reality Check
You assume that it doesn’t matter which copy of a new lens you buy. When a reputable lens manufacturer makes a particular lens model, you’d think that they’re all about the same, right? Think again.
I have read that you should only worry about lens quality variation when you buy the cheap consumer models. My own experiences say, to borrow from the French, “au contraire, mon frere”.
I used to have two copies of the FX Nikkor 85mm f/1.4 AF-S. This is not a cheap lens, and is considered premium “pro” gear. One lens copy was about 25% lower resolution than the other; I ended up selling the soft lens copy (at a 50% loss!).
18-140 Nikkor at 140mm
I presently have two copies of the Nikkor 18-140mm AF-S DX f/3.5-5.6 G ED VR lens. It’s not a professional lens by any means, but it has pretty good sharpness, decent focus speed, small, light, and is very versatile with its large zoom range.
I thought it would be an interesting exercise to compare several properties of this lens model. When a lens manufacturer puts together a complicated lens like this, it’s impossible to assemble the parts identically from one lens to the next.
Tiny parts tolerances along with small assembly variations can have a bigger impact on the final optical characteristics than you might imagine.
I used the free MTFMapper program and a 4-foot-by-five-foot mounted test chart to get the measurements that follow. This is the same program that NASA used to evaluate the lenses on the Mars Perseverance Rover.
Nikkor 18-140mm: 17 elements in 12 groups. Courtesy Nikon
Resolution
Pretty much everybody understands that lens copies may differ slightly in resolution, so I’m going to start the lens comparison with that parameter.
A proper lens resolution analysis needs to show you the whole lens results, and not just a measurement from the lens center or from an edge. To better understand optical characteristics, you also need resolution information in both the sagittal (think wheel spokes) and meridional (circle tangent) directions.
I used a 24MP DX camera for all of the measurements that follow. Vibration reduction was turned off, and the camera was mounted on a sturdy tripod. The mirror was flipped up, to minimize any vibrations.
The resolution measurements were made using the same resolution chart at the same distance with the same aperture, and the same lighting for both lenses. If measurements are taken at a different camera-to-subject distance or aperture, the resolution will change.
I used a linear translation stage to carefully shift the camera/lens between each resolution test shot by 1 millimeter, and then picked the sharpest one. Even contrast-detect focus isn’t quite accurate enough to nail the focus, and slightly missed focus really impacts resolution.
Lens ‘A’ MTF50 resolution: 18mm f/3.5
Lens ‘B’ MTF50 resolution: 18mm f/3.5
Notice that lens ‘A’ above has a slightly lower peak resolution (63.1 lp/mm) versus than lens ‘B’ (65.3 lp/mm). This sort of difference cannot be seen in photos, and it takes a computer analysis to note this small of a difference.
The peak resolution for lens ‘A’ is pretty much in the optical center. For lens ‘B’, you’ll notice that peak resolution is below the center. The lens B was probably assembled slightly off-center.
Also note that the general shape of the meridional-direction resolution is nearly a perfect circle. The sagittal-direction shape is very different, yet quite similar between the two lens copies.
The drastic drop-off in resolution away from the center is the price you pay for a super-zoom lens. A landscape photographer wouldn’t be very pleased with the quality of the photos along the edges, at least not with the lens aperture wide-open. The lens edge resolution is improved by stopping down, but it never approaches fixed-focal length lenses or pro lenses.
Lens ‘A’ MTF50 resolution: 140mm f/5.6
Lens ‘B’ MTF50 resolution: 140mm f/5.6
At the longest zoom setting of 140mm, lens ‘A’ is clearly better, having a peak MTF50 resolution of 50.2 lp/mm, while lens ‘B’ only reaches 40.7 lp/mm.
Notice how similar the meridional and sagittal plot shapes are between the two lenses. They may not be identical twins, but they’re clearly siblings.
On most lenses, the meridional resolution is worse than the sagittal. This is expected from the design of the lens.
If I had to choose which lens to keep, it would be lens ‘A’, at least based upon resolution. The lens ‘B’ is about 19% lower in peak resolution.
MTF50 Contrast
The contrast plots are what most photographers are familiar with. In case you didn’t know, though, nearly all lens manufacturers only publish ‘theoretical’ values, and not actual measured values. Also note that theoretical plots don’t even consider the camera sensor; my MTF50 contrast plots include the camera sensor effects. My contrast plots additionally include measurements at 50 lp/mm, and not just 10 and 30.
18mm f/3.5 and 140mm f/5.6 MTF 10,30 : Courtesy Nikon
Note that in Nikon’s theoretical MTF contrast plots above, they expect the meridional-direction (M) results to be quite a bit worse than the sagittal-direction (S) results.
Lens ‘A’ 18mm f/3.5 MTF 10,30,50 Contrast Plot
Lens ‘B’ 18mm f/3.5 MTF 10,30,50 Contrast Plot
Lens ‘A’ 140mm f/5.6 MTF 10,30,50 Contrast Plot
Lens ‘B’ 140mm f/5.6 MTF 10,30,50 Contrast Plot
The contrast plots show that lens ‘A’ is clearly superior. The ‘B’ lens has a wider spread in measured readings, indicating that its lens elements aren’t as carefully aligned as they are in lens ‘A’.
Lateral Chromatic Aberration
Nearly every photo editor can correct for lateral chromatic aberration, so this lens defect isn’t as important as resolution and contrast.
18mm f/3.5 lens ‘A’ lateral chromatic aberration
18mm f/3.5 lens ‘B’ lateral chromatic aberration
140mm f/5.6 lens ‘A’ lateral chromatic aberration
140mm f/5.6 lens ‘B’ lateral chromatic aberration
You can’t tell much of a difference between these lenses in regards to lateral chromatic aberration. The values are a bit larger than most lenses, but that’s another price you pay for a super-zoom lens. Fortunately, photo editors can mostly eliminate it.
Summary
Identical lenses are a fantasy. If manufacturers could make lenses that were imperceptibly different from one to the next, they would probably have to cost tens of thousands of dollars.
Generally speaking, about the best you can do is to focus-calibrate your lens to your camera, assuming that your camera supports that feature. Mirrorless cameras are of course better at getting accurate focus, but lens manufacturing variations are still going to bite you.
It would be interesting to know what “meets factory specifications” actually means. I think that the manufacturer is just hoping that you don’t look too closely at what you’re buying.
Don’t get me wrong. I still use these lenses quite a bit. It’s a lot better to get the shot than get nothing, and lenses as portable and versatile as these encourage you to bring along your ‘real’ camera and not just depend on your cell phone camera.
Komentar