This post came about as a response to post on a Google+ group I belong to. The post contained a link to an article at the website Olympus Passion by Chris Corradino (whom I don’t know) called “Full Frame vs Micro 4-3 Revisited with Pro Olympus Lens” where a canon EOS 6d with 17-40mm f4 L lens was compared with an Olympus OMD EM10 with 12-40mm f2.8 Pro lens. Nothing wrong with that, it serves as a potentially useful comparison of two popular cameras and lenses. In the article he posts two pictures of the same scene taken around a year apart, one used a polarizing filter one didn’t and the Canon 17-40L is not the best lens in the line up, and comparing a wide-angle zoom to a standard zoom doesn’t really tell anything. Also Olympus uses in camera software correction of its lenses to the RAW files whereas the Canon doesn’t and one has to manually apply correction when processing in Lightroom or ACR. So I thought about it and decided to do my own test. Now before I start I’d like to say that testing zoom lenses is fraught with difficulty and the reason why is very ably demonstrated by Roger Cicala’s excellent article “Painting Zoom Lenses with a Broad Brush – Roger’s Law of Wide Zoom Relativity” which is enough to make any sane person throw up their hands in horror at the enormity of it all.
I’m not really the scientific type, but I know that to make any form of comparison you have to compare like with like and remove all variables. The cameras and lenses tested were as follows:
- Olympus OMD EM1 mki with Olympus m.Zuiko 12-40mm f2.8 pro
- Canon EOS6d with Canon EF 24-70mm f4 IS L
- Sony A7r with Sony FE 28-70 f3.5-5.6 OSS
Why three cameras and three lenses, well mainly a case of why not, but also I wanted to see how a budget kit lens performed against the two “pro” lenses, and I was using it as a control as I had previously looked at it. Each lens was shot at the wide end and the long end, with the wide end an architectural shot to look at how the lens behaved at infinity and close-ups at the long end.
Typically standard zooms behave best at the wide end and the performance deteriorates as you zoom into the long end. All the lenses were tested at an aperture of f8 for the two shots previously mentioned and wide open to test for vignetting. For each shot the camera was mounted to a tripod, there were no filters on the lenses, any image stabilisation was turned off and the shutter was tripped via the self timer. All metering was done using a handheld incident light meter – Minolta Auto Meter V f. Because the base ISO of the EM1 is 200 all the images on all cameras were shot at value. The only DSLR in the group was used in live view mode to avoid mirror slap. The images were all shot as RAW files and then converted to 8 bit jpgs using RawTherapee (version 4.2.1) which allows you to switch off any embedded lens correction. There was no sharpening, noise reduction or correction for chromatic aberration. Ok that’s the methodology lets look at what happened. Click on the images to see them at full size.
Olympus OMD EM1 mki with Olympus m.Zuiko 12-40mm f2.8 Pro lens
Built for the micro four thirds system this lens is equivalent to a 24-80 mm lens and that’s all the talk of equivalency you’ll get out of me, if you want more see this. The lens was launched at the same time as the EM1 in 2013 and it heralded a new line of “Pro” lenses. Of the three lenses here it is the only one with a metal outer construction and it has a splash proof and dust proof design. Its vital statistics are 84mm in length, a diameter of 69.9mm, has a filter thread of 62mm and weighs in at 382g. Not that it counts for much, but it feels nice in the hand and the manual focus clutch (reminiscent of the ones found on Pentax’s 645 range of lenses, is a very nice touch. Optically there are 14 elements in 9 groups – there are 1 aspherical element, 1 dual-sided aspherical element, 2 ED glass elements, 2 HR glass elements, 1 EDA glass element, 1 HD glass element. It is the most optically complex lens of the group. For bokeh aficionados there are 7 rounded aperture blades. The minimum focusing distance is 0.2m.
At the wide end the centre of the image is what you’d expect from a modern zoom, it is sharp and contrasty. At the edges the image is still sharp but the contrast has fallen off a little giving the appearance of softness. The chromatic aberration (CA) is very apparent. At the long end in the centre the image is still sharp but the contrast is lower than what we found on the wide end. The edges exhibit a little softness. The CA virtually non-existent. The results from the long end caused me a little consternation as it was the first time I’d seen images without any in camera correction applied so I repeated them just be sure and got exactly the same result.
Testing for vignetting and distortion I found that the lens exhibited marked vignetting and barrel distortion at the wide end at f2.8. The vignetting had disappeared on stopping down to f5.6.
At the long end there is only slight vignetting and very mild pincushion distortion. What is interesting is that there is significant variation in the exposure, remember these were metered using a handheld incident meter. The long end is nearly a whole stop darker than the wide end. This shouldn’t happen with a constant aperture zoom.
Canon EOS6d with Canon EF 24-70mm f4 IS L lens
This is the most expensive lens in the group with a retail price of around $1200 AUD. It is also worth pointing out that this is the budget standard zoom in Canon’s L range. This lens was introduced in 2012 and was intended to be a kit lens for then newly announced Canon EOS6d. The body is made of some variant of ABS plastic which is somewhat disappointing considering the price, but with a weight of 600g it has a satisfying heft. The optical construction is 15 lens elements in twelve groups with 2 aspherical and 2 UD elements. The aperture diaphragm is comprised of 9 rounded blades. Unsurprising it is the biggest lens of the trio being 93mm long, 83.4mm in diameter and has a 77mm filter thread. Image stabilisation is built into the lens and is good for four stops. The minimum focus distance is 0.38m, but the lens has a rather nifty macro feature where you press a button and turn the zoom ring and that takes it down to 0.2m and a maximum magnification of 0.7x.
At 24mm the centre of image is extremely sharp and contrasty and the edges are the same. At the tele end it is a repeat. Chromatic aberration is non-existent at both extremes.
Shooting wide open at 24mm vignetting is very apparent, I would say that there’s probably 2 stops difference between the corner and the centre. The barrel distortion is not excessive but is noticeable. At 70mm a small amount of vignetting can be seen and there is some mild pincushion distortion.
Sony A7r with Sony FE 28-70mm f3.5-5.6 OSS
The joker in the pack and the cheapest on test at just under $350 AUD, and boy when you pick this one up it certainly feels like it. When I originally wrote about it on 3rd July 2016 I was very skeptical of its durability, well nothing adverse has happened to mine, but, Kirk Tuck wrote today that his met a tragic end courtesy of a dog’s tail and a hard floor. Well what do get for your $350? Well unsurprisingly at this price point this is largely made of plastic – the cheap kind – and is very light coming in at 295g. There’s not a lot of glass either just 9 elements in 8 groups made up 1 extra low dispersion and 3 aspherical elements. The simpler optical design and low price makes this the only variable aperture lens in the test group. The aperture diaphragm is made up of 7 blades. Size wise it is a compact 83mm long, has a diameter of 72.5mm and has a filter thread of 55mm. The plus points are that it is dust and moisture resistant, has built in image stabilisation, has a respectable minimum focusing distance of 0.4m and comes with a lens hood. As an aside I believe that every lens should come with a hood.
At 28mm the lens is again just like any other modern lens – sharp and contrasty. At the edges there is a drop off in contrast but they remain sharp. Chromatic aberration is very well controlled and easily fixed in Lightroom. The slow variable aperture means that vignetting is not a problem and there is only slight barrel distortion. At 70mm the centre is sharp and has good levels of contrast, the edges are sharp with a slightly lower contrast. There was also some evidence of coma. Again very little vignetting and a slight trace of pin cushion distortion.
So is the Olympus 12-40 the Canon killer that Chris Corroding says? Well without in camera software correction it is only just a bit better than the Sony which is half the price. It is the correction that elevates this lens to very good. Having said that the Canon 24-70 f4 is probably one of their very best zoom lenses and produces very good images straight out of the camera. Apply the Lightroom lens profile and it is even better. Is that surprising? No considering its cost I would hope it be optically very good. This is the rub – for $350 AUD you get a surprisingly good lens with the Sony and most people would be very happy with it. The Olympus at $870 AUD sees some significant improvement. The Canon at $1200 AUD sees only incremental improvement over the Olympus. The law of diminishing returns is very clearly at work here.
When it comes down to sensor performance the Sony A7r rips the head off of the EM1 and EOS6d and spits down the stump. I have been amazed by how well it performs – the dynamic range is very, very good, the lack of an anti-aliasing filter means that with good lenses insane amounts of detail can be rendered, and the high ISO performance is also very good. The Canon EOS6d’s sensor is capable of very nice colour rendition, especially skin tones, but it is not the best when it comes to dynamic range. It is what I’d call workman like. For the Olympus EM1, well the dynamic range is very good, high ISO performance not so. The lack of anti-aliasing filter helps you get the best out of the lenses. I really like my m4/3 Olympus cameras and lenses, I use them more than anything else, but I really feel that the sensors are holding them back. I’ve yet to get any long-term usage out of the new 20Mp sensor that is found in the EM1 Mkii, but I do think that if they got access to the latest BSI sensors from Sony and reduced the base ISO to 100 that there would be quite a sizeable performance boost. But all this is moot as all three are capable of excellent results if I do my part.
It is important to remember that I only have access to one of each lens so I have no idea of what the sample variation is for any of them. What does this all mean in terms of real world usage? Well I’ve used my Olympus 12-40 far more than the Canon 24-70 and I am more than happy with how it performs. I hardly ever use the Sony 28-70 as I only really use the A7r with legacy lenses.