Exactly twenty years of digital photos and ten years since the first mirrorless experiment, I “celebrated” the double event with a bad surprise: my standard zoom, the Sony Zeiss Vario-Tessar T* E 16-70mm F4 ZA OSS, broken. I didn't drop it: it did everything by itself, as Sony repair lab confirmed to me. A mechanical failure. After more than 400€ of repair, unfortunately the lens came back with a bad performance through the whole range and worse at the wide end — something that could be appropriate for a much cheaper lens. Back to the lab, the long end improved a bit, while the wide end remained the same. At this point the lab told me it was `within the specifications` and nothing else could be done. After ten years I understood that my copy, while clearly decentered, was placed towards the top of the list; and I felt the surprising experience that a good deal of reviewer had with this lens, asking their provider to change the copy multiple times before finding a decent one to test.

My relationship with this lens has always been a kind of love-get bored thing (love-hate would be excessive). Looking at my catalog, there are literally hundreds of excellent photos that I took with it: mostly landscape, but also architecture and a few plants. All photos in this post before the evaluation section below have been taken from the gallery of Sony Zeiss Vario-Tessar T* E 16-70mm F4 ZA OSS best shots.

	Tamron 17-70mm ƒ/2.8	Sony 16-55mm ƒ/2.8	Sony 16-70mm ƒ/4	Sigma 18-50mm ƒ/2.8
wide range loss	16-17 mm	-	-	16-18 mm
tele range loss	-	55-70 mm	-	50-70 mm
stabilisation	yes	no	yes	no
weight	525 g	494 g	305 g	290 g
length	119 mm	100 mm	75 mm	76 mm
price	610 €	980 €	650 €	500 €

When I took the photo below I could clearly see green flares dancing in the left bottom part of the photos; but in the end they are barely visible — I didn't even need to fix them in post-processing. Note also how the sun is properly rendered, with the transition from the disk to the sky being not too harsh. Highlights have been fixed in post-processing to give the sun a more distinct shape.

In this version of the photos highlights are the natural ones, so the sun appears fuzzy as it came out, and again the transition area is quite clear.

50mm

At the wide end and full open the lens renders the sun as a nice circular blob with a good transition from the center towards the border. Indeed in the photo below the border of the sun is a bit too marked, but it has been my fault as I over-exposed the photo. I didn't do the same error going at the shorter focal length (see below).

Stopping down makes sun stars to progressively appear, while keeping the good overall rendering. At ƒ/22 they might be too invasive, as for instance the one pointing down and leftwards in the photo below. A very good behaviour in the end, even though as usual in this scenario each shot is a story apart and one should work by trial-and-error.

On the other hand Sun shots with the Sony lens have been often troubling. When the lens is full open, the sun is rendered as a blob (which is not necessarily a problem), but not perfectly round, as a sort of protuberance sometimes emerges at the top left side. Stopping down there are some traces of sun stars, partial and totally irregular (typically a strike from the top-left towards the bottom-right appears).

Wide end

At the wide end the behaviour of the Sigma lens is substantially similar.

The Sony lens works a bit better at short focal lengths and wide open and the sun is rendered with a more regular round blob.

On the other hand, stopping down the situation worsens, as irregular sun-stars appear (and the diagonal strike is more evident); furthermore green flares appear, together with the worse kind of artifact ever seen: a sort of red-green smeared lines that are literally impossible to fix (they are more evident when clicking on the photo for a larger rendering). These aberrations did literally ruin a not negligible amount of photos, specially when I had to stop down for maximising the depth of field.

To be honest, it was not a total disaster; by slightly adjusting the composition, anyway, I was able to take a number of good shots.

Close up

A feature of the Sigma 18-50mm lens is the magnification ratio, 0.35x at all focal lengths — on an APS-C sensor it means that you can vertically fill the sensor with a subject 9/10mm tall. In comparison, the Sony 16-70mm is capable only of 0.23x at the long end, and much inferior at the wide end. This means that in theory the Sigma lens can be used a second-choice lens for close-up photos of critters and flowers, for instance when you are on a landscape trip without other lenses and occasionally find a small subject. But given that this is not the primary purpose of the lens and that it has been designed for small size and weight, it's clear that there are not corrections for defects such as chromatic aberrations and so on.

Flowers of wild carrot (Daucus carota) are a perfect test for testing chromatic aberrations, being white, large and relatively flat. As it can be seen in the test shots below, taken at 50mm, LoCa is massive at ƒ/2.8 and still high at ƒ/4 (in this shot the flower was slightly pushed forward by a slight breeze, so the DoF band moved backard — that's why it appears less blurred than the shot at ƒ/5.6). Aperture at ƒ/5.6 produces good results and definitely the problem disappears at ƒ/8).

Of course the problem is that at ƒ/8 the creamy bokeh has been lost.

In the bee photos below (cropped at 1:1), the left one at ƒ/2.8 shows large amounts of LoCa that can be seen on the out-of-focus parts of the flower, on the antennas and the blurred hairs. In the right one, taken at ƒ/8, LoCa is negligible.

ƒ/8 actually seems to always fix the problem, so if you manage to find a good separation with the background and get a still creamy bokeh, you can produce very good photos. As it can be seen by looking at the marginal part of the photos, slightly beyond the focal plane, the defocus transition is quite good.

Given that the rules governing chromatic aberrations are complex and quite unpredictable, each shot is a story apart and it might happen to have things going better than expected; for instance, in the shot of Pyrrhocoris apterus below, taken at ƒ/4, LoCA gave just a light cyan touch to some portions of the flower. I could easily fix it by desaturating the cyan and blue channels, but even without the fix the shot was decent.

A similar shot of a crab spider, taken at 24mm ƒ/4, didn't require any correction. So I'd say that the correct approach is to try and take multiple shots bracketing the aperture, even when it's not needed by DoF requirements.

This long-legged critter has been used as a test for an attempt to fix LoCA in post-processing (also because I really wanted to stay at maximum aperture for the bokeh). I removed the massive LoCA by completely desaturating the cyan and blue channels; but this, unfortunately, destroyed the bluish background, upon which the green of the subject popped out. Only green remained in the background, mixed with completely desaturated patches. I had to apply a mask to re-colorise the background, giving a uniform green tone. The results are not exceptional, but it's also due to my non-excellent skills with masks. This is to say, anyway, that a photographer with good post-processing capabilities might be able to pull a good image even when the LoCA is massive.

50mm with colour fix done properly (add before/after).

AT 20mm everything is fine.

At 18mm LoCa has a much slower impact, even at maximum aperture; but with backlight it intensifies again, even though the shot is not ruined. The bokeh is pleasant, even with bright spots on the background.

A thing that must be said with close up is that auto-focusing often struggles to get a lock, and it's not unlikely that it's imprecise. It's quite a different behaviour than the one of specific macro lenses, such as the Sigma 105mm F2.8 Art. Often small manual adjustments are required, either using DMF or slightly moving back and forth, as it's usual to do with manual lenses. Trying AF-S for dealing with moving subjects, including slight motion due to the breeze, is much less effective (it works very well with truly macro lenses).