Sony Lens Quality

Marc Heijligers, 06-02-2016

Recently I bought a Sony A7RII camera, and I considered a couple of lenses to build up a system. The lenses I’ve considered are:

• Sony FE 35mm F2.8 - compact, and reported to be very sharp.
• Sony FE 28mm F2.0 - reported to render nicer than the 35mm F2.8, but less compact.
• Sony FE 55mm F1.8 - reported to be ultra sharp, and a very nice bokeh.
• Sony FE 35mm F1.4 - bulky and heavy, but adored about its bokeh and rendering.

To determine which lens to keep, I’ve compared properties like sharpness, the nature of the bokeh, and the way the region between a sharp and unsharp region smoothens. Compared to what you see on other sites that compare lenses, the pictures I made are not so interesting from photographic view, but they are easy to reproduce in moderate controlled conditions, and contain sufficient properties to compare the above ingredients. The

During the comparison of various lenses under similar conditions, I found out that there is a huge variation in field sharpness with similar lenses. So for example, with the 35mm F2.8 lens, for one particular lens the left side of the field is more blurred, whereas with another copy of the same lens the right side of the field is blurred. Another noticable difference was the amount of color fringing. This raises the suspicion that internal lens elements are titled/skewed.

Test Conditions

The first test is to test the field sharpness of the lens. For this I use a bookshelf, which has a flat surface and a lot of books where the text on the book spines can used to test the subjective sharpness at a distance of 1.5 to 2m (not fully scientific, but sufficient for the practical use cases of most of us). The camera is put on a tripod, where care is taken that the camera is perpendicular to the bookshelf (using the level indicator of the camera, visually checking the camera is in the middle and the bookshelf plates are symmetrical, and double check with LensALign that the book spines are within the DoF range). The focus point for all photos is always in the centre (though other focus points have been checked for sanity, as ell as multiple pictures to exclude focus differences). Photos have been made with Steadyshot OFF, Silent Shooting ON, and a timer (but the differences are hardly visible with other settings in this setup). Pictures are taken in RAW, and processed in Lightroom with the Camera Neutral setting, and the Lens corrections enabled (also a default use-case for most - for sanity I’ve checked the difference between JPG, and RAW with corrections turned off, but the conclusions also hold under those conditions). Three regions are considered in the tests, one in the centre, one at the bottom-left, and one at the top-right. To investigate color fringing, tests have also been done closer to books with a high level of contrast.

The second test is to test the lens for sharpness at further distance. This also rules out the problem with the camera might not be absolutely perpendicular to the bookshelf. For this purpose I photographed a (boring) village scene from the balcony. Also here three regions are considered the centre, the left, and the right side.

Although the above tests are not scientifically set up, and hence absolute statements on the amount of sharpness cannot be given, the setup is constant when exchanging lenses, and therefore relative difference between lenses are conclusive, which is the purpose of this test.

Expectations

No lens is perfect, hence we cannot expect ultimate sharpness overall. On the web, test sites like SLRgear, DXOmark or DPreview use better test conditions, and give a good indication about the expected performance.

The figures underneath give the field response for different lenses from SLRgear at open aperture and at F5.6 (where most lenses should be decently sharp over the whole field, with a rare exception of the 35mm F2.8, which on other sites like DPreview improves when stopping down, so also here we see sample variations). The graphs from SLRgear already hint at some field variations (e.g. for the 35mm@f5.6, one can see in one corner a blur factor of 3 pixels, whereas in another corner it is just underneath 2 pixels):

Results

The summarized observations per lens type are shown below:

Conclusions

It is obvious that there are quite some variations per lens, which given the price level of those lenses is a bit surprising. I would expect a better quality control with less variations. My recommendation is to always quickly check your lens directly after you bought it, and make sure you have a dealer who provides good service in cases you want to exchange a lens.