MTF Contrast Plots: How Useful are They?
The only camera manufacturers that presently show the public actual measured lens performance data are Leica and Zeiss, and possibly Sigma. The other manufacturers only show “theoretical” performance, typically in the form of an MTF contrast plot. These idealized plots typically are calculated at both 10% and 30% contrast, and separated into meridional and sagittal directions.
Another aspect of these theoretical plots: they don't consider the camera sensor being used. Since I typically mount the lens on a camera to use it, I'm kind of interested in the results of the whole combination.
This begs the question: do the theoretical MTF plots have any basis in reality? What if you could estimate your car’s pollution output for the DMV instead of them making you get it measured? I thought so.
I figured I’d try to answer these questions. As always, trust but verify.
I use the MTFMapper program to measure lenses (mounted on cameras). The recent versions of this program let you display (measured) MTF contrast plots, so that they look just like the manufacturer plots. I’m presently using MTFMapper version 0.6.7, which is for 64-bit Windows. The download site for this free software is here.
Before you can make lens measurements, you will need to print, mount, and photograph a resolution chart. My chart is about 40” X 60” in size, so that I can take measurements at realistic focus distances. The measuring program can use a few different resolution chart designs, and I am using the newest design.
Chart used to make resolution/contrast measurements
Another thing: most manufacturers only show their theoretical MTF contrast plots at the widest lens aperture. Most photographers want to know what aperture gives them the best resolution and contrast, plus how much quality difference there is between aperture settings.
To answer these MTF questions, I picked on a pair of pretty good lenses: the Nikkor 85mm f/1.4 AF-S and the 105mm AF-S Micro f/2.8 G.
These lenses should have decent quality control, plus they're primes, so the theoretical MTF values should have the best chance to match real measurements.
I used a Nikon D610 and un-sharpened 14-bit RAW files for the tests. All pictures were taken using a heavy tripod, Live View mode, “mirror-up”, remote release, and contrast-detect focus. I picked the best results from each aperture, out of a minimum of 10 shots at each aperture. I used “cloudy bright” daylight illumination. I used this full-frame camera so that I could get the same range of information as that from the Nikon web site data.
Nikkor 85mm f/1.4 AF-S Lens
85mm MTF Contrast Plot from Nikon Web Site (theoretical plot)
I grabbed a screen shot from the Nikon site, showing you how the 85mm lens should perform at f/1.4. This plot would assume that their manufacturing plant is capable of flawless lens assembly and their parts exhibit no process variation. Here, the “S” stands for “sagittal”, or the “spoke direction” from the lens center. The “M” stands for “meridional”, or tangential direction. “S10” is for 10% MTF contrast measurements in the sagittal direction, while “S30” is for 30% MTF contrast (represents resolution) measurements in the sagittal direction.
85mm MTF Measured Contrast Plot, f/1.4 (peak MTF50 41.8 lp/mm)
The theoretical and actual MTF contrast plots look quite a bit different. Surprisingly, some aspects of the measured plot actually look better than the theoretical. The pink-ish and blue-ish bands around the plot lines show the actual spread of measurements taken; the dark lines are the average of the measurements. Note that the measurements stop at about 18mm (sensor edge), starting from the lens center. The Nikon-supplied plot has measurements out to the corner, or about 22mm.
So, how about the other apertures for this lens? What follows are the measurements at other apertures, to give you an idea of how much the lens improves as you stop it down (until diffraction starts to mess up the resolution).
85mm MTF Measured Contrast Plot, f/2 (peak MTF50 45.2 lp/mm)
85mm MTF Measured Contrast Plot, f/2.8 (peak MTF50 55.2 lp/mm)
85mm MTF Measured Contrast Plot, f/4 (peak MTF50 58.6 lp/mm)
85mm MTF Measured Contrast Plot, f/5.6 (peak MTF50 56.9 lp/mm)
85mm MTF Measured Contrast Plot, f/8 (peak MTF50 53.5 lp/mm)
85mm MTF Measured Contrast Plot, f/11 (peak MTF50 45.2 lp/mm)
85mm MTF Measured Contrast Plot, f/16 (peak MTF50 36.8 lp/mm)
Notice how the astigmatism vanishes at about f/8 (no more separation between sagittal and meridional measurements). By f/4, even the edge performance is excellent. When you only see the wide-open MTF plot, you don’t get any of this insight.
105mm AF-S Micro f/2.8 G Lens
105mm f/2.8 MTF Contrast Plot from Nikon Web Site (theoretical plot)
Above, I show the Nikon web site plot of the 105mm f/2.8 Micro Nikkor (at f/2.8).
Now, it’s time to see how this compares to reality.
105mm MTF Measured Contrast Plot, f/2.8
The f/2.8 measured plot differs a bit more from the theoretical plot than the 85mm did. Nothing in the measured plot is as good as Nikon’s claims. I don’t have another 105mm lens to compare to this data, but I’ll bet it would be different from the above data, too.
105mm MTF Measured Contrast Plot, f/4
105mm MTF Measured Contrast Plot, f/5.6
105mm MTF Measured Contrast Plot, f/8
105mm MTF Measured Contrast Plot, f/11
105mm MTF Measured Contrast Plot, f/16
Although the 105mm measurements don’t stack up to the Nikon claims, try to keep in mind that measurements higher than 0.5 (50% contrast) at MTF 30 lp/mm are really good. Again, keep in mind that these plot measurements extend to the frame edge, versus Nikon’s frame corner, when comparing plots.
2-D plot of MTF50 performance, 105mm @ f/2.8
I added an MTF50 plot at f/2.8 to show how much more informative that style of plot is, compared to the mere MTF contrast plot. You get to see the performance all over the surface of the sensor.
Conclusion
I pretty much expected that the real measured lenses wouldn’t look quite as good as Nikon’s fantasy plots would imply. The measurement plots bear this out. I’ll bet that Canon et al. would show a similar trend.
Personally, I still think that the 2-D plot measurements in an MTF50 chart give much better information about lens performance than these MTF contrast plots. The MTFMapper program is capable of providing both kinds of information, so you get to choose.
I have attempted to provide all the information that you will need to make similar measurements for yourself. No two lenses are going to be identical, so you need to always keep that fact in mind when looking at lens measurements.
