Sample variation in the Fujifilm GF 100-200 at 200mm

When I measured the performance of the GF 100-200mm f/5.6 lens at 200 mm, I got good, but not great results. In fact, it was about the weakest performance I’ve seen with the excellent line of GF lenses. Ever since I published those results, people have been saying that their copy of the GF 100-200 is better than the one I tested. I now have some results from raw files that a reader sent me, and the people who have been pointing at sample variation have a point.

Here are the results with the GFX 100S. Explanation will follow below.

MTF stand for modulation transfer function. It is a measure of system contrast versus spatial frequency. MTF50 is the spatial frequency at which the contrast is half of what it is at a frequency equal to zero. It’s a good measure of perceived sharpness under some common viewing conditions. I’m expressing MTF50 in cycles per picture height.

In the center at f/5.6 and f/8, Mike got significantly better results than I did. At the right side of the frame, with a radial edge, my results and Mike’s results were about the same.

There were some differences in methodology that you should know about:

1. Mike took several pictures at each f-stop. I picked the sharpest. I used focus bracketing to get the sharpest image in my series.
2. For Mike’s images, I used Matlab standard demosaicing (gradient-corrected linear interpolation), and calculated the luminance MTF. For my images, I white balanced the raw channel MTFs to get luminance. This difference favors Mike’s numbers, but not nearly enough to explain the differences seen at f/5.6 and f/8 in the center.

Here’s a look at the center MTF curves for the sharpest f/5.6 image in Mike’s series:

The overshoot in the edge profile is evidence that the MAtlab demosaicing is doing some sharpening. Quite a bit of longitudinal chromatic aberration (LoCA) is present, as shown by the difference between the green, red, and blue curves.

At f/8:

The effective LoCA is less at f/8 than at f/5.6.