Tag Archives: lenses

The sound of pennies dropping

I shoot a Canon EOS 7D. As well as some Canon lenses, I also own some Sigma lenses. I acquired an ƒ/2.8 70–200mm Sigma a little while ago.

Now, something about using Sigma lenses on Canon bodies bugs me. See if you can spot it in this screenshot from Aperture (bear in mind the image was shot using the 70–200mm):


See the lens model at the bottom? It seems the Canon “brain” sees the Sigma lens, but identifies it as an ƒ/2.8 50–150mm plus a 1.4x extender! Do the maths, and you’ll see this odd combination actually works out as a 70–210mm. It’s not just this lens, either. All my Sigma lenses, both EF and EF-S compatible, are reported incorrectly in the EXIF from the camera. Weird but true.

It would be really nice if there was a way to correct this, either in the camera or within Aperture. I haven’t found a way yet.

Testing a new toy

I have wanted an ƒ/2.8 70–200mm lens for some time. I wanted the Canon variant, but sadly finances don’t quite stretch that far at the moment. After a brief flirtation with the Canon ƒ/4 70–200mm, I decided I really did want the ƒ/2.8 after all.

After comparisons around the main brands, I decided the Sigma 70–200mm ƒ/2.8 II APO EX DG Macro fitted the bill nicely. Reviews were good, if not completely stellar. Users seemed to be happy with the quality, and already owning two other Sigma lenses meant I knew what the build quality would be like. All in all, I felt my choice would be a competent and useful lens that would serve me well until I was wealthy enough to buy the Canon lens.

Then, my hand was forced a little in my decision. The lens in question had been launched in 2008, and was ripe for replacement. Many places still seemed to stock it, but when I went to one of my favourite suppliers, they only had the new version in stock. The new version, sporting improvements such as image stabilisation and slightly better optics, was twice the price. There were still resellers, through the Amazon marketplace, who were stocking the old variant I was after. I did some bean counting, and worked out that, yes, I could afford the lens, and it was duly ordered.

After a week or so, the hefty box arrived from Germany. The lens is a beast. It is built like a tank, and weighs about the same. Attached to the camera, you know it’s there. As with the other Sigma lenses I own, the autofocus is fast and silent, just occasionally getting confused. The macro feature, which I thought I’d probably not really use much, turns out to be quite useful.

I tried the lens out indoors. The ƒ/2.8 constant aperture was very useful. AF would latch on to almost anything no matter the light level. Hand-held, the zoom and focus rings were comfortable to use. The weather, it being November, however, didn’t really let me test the lens outside. I tried some shots from our back garden when we had a few minutes of watery morning sunlight, but that was all I managed.

To say the results were disappointing is understatement. I had been aware of the reviews commenting on softness at full zoom, and a measure of “chromatic aberration” (colour fringes caused by the optics). However, they seemed far worse than I had anticipated. While I had not expected pin sharp detail, I had expected things to be slightly less fuzzy than what I was seeing.

I tested the lens to see if it needed calibrating. My camera, the Canon EOS 7D, has a micro-adjustment feature, which allows you to compensate for lenses that consistently misfocus. I tested the Sigma, and found it was just about right.

So, it wasn’t the lens focus at fault. Perhaps it was technique. This morning, I awoke to bright autumnal sunlight. I went back into the garden, tooled up with the camera and lens. I wanted to see if I could settle whether the fuzziness was just a feature of the lens optics, or whether I needed to refine the technique in order to get the best from it.

These images are unprocessed, converted to JPEG from the RAW images using Apple Aperture 3. The only processing that took place was the RAW conversion filter.

On the surface, they don’t look too bad. The bokeh, one of the key features I was after with the ƒ/2.8, is nice and smooth. However, when you look at them in detail, you can see just how soft the images actually are.


Taken at the full 200mm zoom (310mm effective on my APS-C camera), ƒ/8 at 1/250th second shutter speed. The grab shows part of the image as it appeared in Aperture on my 15in MacBook Pro, with the full image filling the window.


This grab is zoomed in to 100% in Aperture. I believe this equates to one image pixel per screen pixel, but I stand to be corrected. The image is soft. There is detail there, but it’s not sharp. A little disappointing, but still a usable image provided it’s not enlarged too far.


Again at 200mm, ƒ/3.5 at 1/500th second. Once again, the “fit in window” view.


Here is the 100% view. Note the cyan and magenta fringes on the water droplets?


Finally, another 200mm zoom, ƒ/8 at 1/640th second.


The 100% view.

At the other end of the zoom range, somewhat amusingly known as wide angle, though it’s into telephoto range on my camera at 112mm effective, the images are still soft but exhibit much less colour fringeing. 

Having now had the opportunity to test The Beast in decent light outdoors, my initial disappointment has somewhat evaporated. The lens appears to be operating within normal parameters. The quality and sharpness improves as the aperture is closed down, with the sweet spot around the ƒ/5 to ƒ/8 region—much the same as the cheaper 28–300mm ƒ/3.5–ƒ/5.6 Sigma I also own. I was also disappointed with that lens at first, until I began to work within its limitations. With more in-the-field experience, I think I will grow to like The Beast after all.

As a final note, I uploaded my test shots to Flickr. I export uploads to fit within 1024px square. Flickr performs a sharpening algorithm on uploads. Judge for yourself whether that makes a silk purse from a sow’s ear. I may have to experiment with a little judicious sharpening locally to see if it improves matters on my originals.