September 24, 2015.
I had the opportunity to try the Canon EOS 5D Mark III in detail and carefully. This DSLR is not a novelty, the camera is on sale for several years. I wondered, what had changed since the end of summer 2011, when my ►happy article was published about the previous model, which brought many pseudo-experts to the brink heart attack.
First impressions are very good. The body of the camera is carefully crafted and a label with the words Made in Japan evokes a slightly sad memory of the D810, the first professional Nikon made outside of Japan - in Thailand. At first glance, it is clear, that the manufacturer has repaired virtually everything, I complained about 5D II for the better.
Switch. The power-on lever has moved up to the left, below the dial to select photographic modes. The operation of the lever is reasonably stiff, but not too much, precise and gives a good impression. Just the slightly unfortunate location on the left would probably not suit me very well (the Nikon system is probably ideal in this), although for example left-handers may be pleased, I don't know. In any case, the camera will turn itself off after the specified time has elapsed, and there is no need to use the power switch unless it is accidentally turned on in a bag or backpack. That is, except for the replacement of memory cards, which is to be performed (according to the instructions manual), when the device is switched off.
Both wheels. The upper one, next to the display, rotates quieter than the 5D II and has less sharp edges. The large one on the back side is also better processed and also runs quieter. Both are now quite pleasant to treat and I would probably be able to get used to them.
The rear wheel lock is now independent of the main switch, which is also good. The switch can be set to lock one, two or even all three movable controls (including the control mainly for selecting focus points). It's handy and very similar to Nikon's no less practical Lock function.
Top small buttons. They now have a matte finish, but they are still small and have to be pressed relatively deep. I was also bothered by the impossibility of setting the backlight time of the upper display, the preset one is too short.
Magnify the image after photographing. Yes, it works. If you set the magnifier button in this way, pressing it will enlarge the preview on the rear 3.2" monitor to approximately 100%, so as with Nikon, you can now very well assess the sharpness of the image. If you shot in Live View mode, the image will be enlarged symmetrically from the center. If you use focus points in the viewfinder, the area around the point(-s) used for focusing will be displayed.
The image on the monitor is again (I mentioned it in the article about 5D II) very high quality, sharp, colors clean and relatively accurate. You can also set the automatic backlight intensity. The only drawback is the still impossibility to view other images using the wheels after viewing the current preview - here, unlike Nikon, you must first press the Play button.
Focus points. Although I photograph mostly from a tripod and usually use Live View mode (it reminds me of working with a large format camera), of course I often use focusing through the viewfinder. This is now, unlike the previous model, extremely sophisticated, allows a number of precise settings and for photographers of moving objects is certainly very pleasant to work with. Canon has done a lot of work here. Not that Nikon's existing focusing systems are bad, on the contrary, much has improved here, too, but Canon's autofocus is a little further in conjunction with the quiet and lightning-fast Ultrasonic engines. But as I said, it's not very much my field, so understand this statement as a personal feeling rather than a carefully verified fact.
Memory card doors are now better secured against buckling, as they no longer extend to the bottom of the camera and the lower part of the palm no longer refutes them when held. A spring has also been added to the hinge, so that no longer close spontaneously as the, four years ago mentioned, shutters at the cottage. Paradoxically, I have only now realized the disadvantage of this solution - if you have the camera on a tripod, remove the card and want to upload images to a computer, you must close the door completely to avoid unnecessary dusting on the card contacts. This wears everything unnecessarily, for a door without a spring it was enough them to shut to.
Two memory cards are a handy thing, but there is one big mistake. If you photograph on card 1 and remove it to transfer images to a computer, even when the camera is turned off, the camera willingly and quietly switches to writing to card 2 and does not change its decision even after card 1 is inserted and formatted. So what happens after taking more pictures - turn off the camera, remove the card 1 again, insert it into the reader and ... nothing. Photographs are on the card 2. It's very awkward, not to mention the possibility of accidentally deleting images. I tried different settings and nothing, it is always necessary to switch the camera after inserting the card to write to card 1 again. The only solution (if I did not miss something) is not to put card 2 into the camera at all ...
Live View. I praised the previous model and so it hasn't changed much. Only the focus point still does not move "smoothly" (it is never smooth, but simply so gently, that the jumping is not so visible), but it jumps around the individual positions. It still can't be moved to the edges, and sometimes it bothered me, especially with partial images for collagues, and I had to sharpen it by hand (which is not a problem, but it's just really that hard, to use the edges as well? It's possible with Nikon.) But otherwise, working on a larger monitor is a real joy. The exposure simulation is very accurate under at least average lighting conditions, what you see on the monitor is also in the image, even taking into account that I photograph exclusively in raw and the preview is always jpg or a similar image format. Therefore, slight deviations may occur when viewing in Lightroom or C1, but I was always pleased with the exact and precise work even.
How do I photograph in Live View mode?
After switching on, I first switched off all information on the display with the Info button, leaving only a preview of the image. I focused the lens by hand and created the composition of the image. When it was done, I drove the focusing rectangle to where I wanted to focus. This is important, because the color of the place affects the exposure, resp. the need for its correction. So if I placed the rectangle in a bright place, the image darkened using exposure simulation and a shorter time was set (in Av, except for it, I only use M). When focusing on a dark place, of course, the opposite was true. With the Info button, I have now switched on display all the data, including the live histograms, which I set to RGB mode when setting up the camera, so I can see each channel separately. For example, I had a window in the frame, behind which the Sun was rising from behind a cloud. I watched the histograms of how sensitive and accurate they reacted, and think over, how good it is, when things don't go away and do carefully. – After adjusting the exposure and checking the depth of field, I was able to start focusing. Here, I missed the option to automatically cancel the focus by pressing the shutter button halfway while working in Live View mode. It would be enough to focus only with the Af-On button. Well, it works, but only by turning off the focusing via pressing shutter button halfway in the user's functions. With the D3x in Live View mode, the shutter does not focus, it just shoots. When we turn off Live View, the camera automatically refocuses by pressing it halfway. Practical trifle.
So, for the most part, I magnified the preview 5x - 10x (Canon could already do a smoother magnification, perhaps just as an option) and let the camera focus. Needless to say, macro photographies are often more accurate and easier to focus manually. You can turn on the two- or ten-second self-timer mode for exposure, which, together with a high-quality tripod, is usually sufficient protection against vibration. Thanks to the permanently folded mirror, the exposition itself takes place practically noiselessly and, of course, without vibrations.
To conclude - Live View photographing is more enjoyable for me with Canon than with Nikon, despite some missing features, such as the more complicated shutter-release button off or the inability to set two white balances (pilot lights/flash discharge tubes) when work with studio flash generators. I got used to it almost immediately and saying goodbye was really unhappy.
You may be wondering, why I just didn't buy this camera, when I worked so well. You're right, I was very upset about not supplementing my existing equipment. After buying a few lenses (I don't need too bright even with Ultrasonic engines) I would have a different gear for certain orders and free creation. And I haven't mentioned a program for photographing directly from a computer, which is free and works similarly to Live View - great. In short, it all looked very good until a fundamental and crucial problem arose. There was nothing less than
image quality. Maybe now sworn Canon fans are already unlocking the colts and surrounding me in a narrowing circle. Calm down, pistols in holsters, sit down. I'll explain.
Quality of the photographies seemed excellent to me at first. Unlike its predecessor, which suffered from an unnecessarily effective AA filter, the 5D III draws, as if the creators of the sensor now forget to put AA there. Suddenly, the photographs from the D3x seemed slightly less sharp, and even the images from the D810, which has a higher resolution, an AA filter missing, and is currently a widely accepted image quality standard, did not seem much better. Images from 5D III, in conjunction with prime lens, are sharp, perfectly sharp, virtually error-free. I had an all-plastic (including mount!) Canon EF 50 mm/1.8 II (yes, the cheapest lens that Canon makes, sharp as a razor and surprisingly good due to its price, if we do not want to photograph in difficult lighting conditions, need water resistance and durability at all) and the Canon EF 100 mm/2.8 Macro USM lens, again excellent glass, sharp and very high quality. When shooting in balanced lighting conditions, where there were no strong sources of backlight in the frame, the results were excellent with the correct exposure. What I liked most was the precise presentation of details, the drawing of contrasting edges and clean, vivid, accurate colors. I sometimes bother with them with all Nikons, paradoxically even after a complete color calibration of the entire chain. In Adobe Lightroom 6.1, the Canon 5D III managed to make do with the basic conversion profiles, and I usually only slightly increased the contrast of the images, ev. corrected white balance. Noise appears minimally, the camera can handle even high ISO sensitivities without major problems, however, I worked mainly from a tripod to ISO 100. And that was paradoxically a stumbling block.
Above is the entire photograph, below enlarged crops.
Canon EOS 5D Mark III with Canon EF 50 mm/1.8 II lens.
Shutter speed 1/60 second, aperture f/2.8, ISO 100. Photographed with a tripod.
The first problem appeared in the studio, where I photographed a test image of a table lamp. It has ventilation holes at the top, through which the light of the bulb shines intensely. It's a tough test for both the camera and the lens, but all the Nikons I've tried (D80, D300, D700, D7100, D800, D810, D3x) can handle it without hesitation. But there was a problem with the 5D III. It was also visible on the rear monitor and, after opening in the Lightroom, cannot be overlooked. To prevent the translucent light from being overexposed, the rest of the image is logically dark and there are also deep shadows. And it was in these dark areas that quite a lot of noise appeared - mostly yellow and purple, as if I had an ISO set of at least 12 800 (but it was 100). What's more, and what was worse, the image was torn by vertical irregular stripes, that resembled traces of water, when we let it run down on the painting. In contrast, the result of the 5D III is very unaesthetic to disgusting. I've heard of this problem before, it's called low iso shadow banding. Well, I thought, we'll see what happens outside the next day. And it happened.
Again. Balanced scene - excellent. Sun or other backlight source in the frame - streaks and color noise at ISO 100. Most strangely, the night photographs at the railway station did not show a problem, and it was in the frame also the signal lights, that glow very intensely. However - always only by light of a certain color, ie spectrally narrow. Likewise, the sodium lamps, the enemy of photographers with a CRI of 0, did not irritate the 5D III sensor in any way with their yellow-orange nature. Thus, it was clear, that a light source with a broader spectral characteristic is needed to cause the problem.
The culmination was photographing a shiny spoon on a black background. Here was only partial light, the object was illuminated from above and from the front. The intensity of stripes and colored spots was beyond any tolerance.
Canon EOS 5D Mark III, Canon EF 100 mm/2.8 II Macro USM.
Exposure time 1/125 second, aperture f/5.6, ISO 100. Photograped with tripod.
What did Canon say?
I tried to ask on Canon's website. The answer was, that they were sorry, but they had never heard of anything like it, and let me take the camera to a service, that it was probably broken. So I wrote, whether all the links that Google finds, when entering the above words, are somehow fancy, or are there so many cameras faulty? They requested raw for the sample, and I still had to reassure them, that I hadn't modified the files in any way. The answer came fairly quickly. It is said, that they know about this phenomenon and with these cameras (5D III) under these lighting conditions it is normal(!). They said to be working to improve the quality of their sensors.
The last sentence pleasantly surprised me, hopefully it will work out soon. I have about 40 .cr2 files from 5DSR from various authors and it is true, that neither stripes nor spots appear in the shadows yet. However, in order to be able to judge it responsibly, I would have to use the camera when working on orders, where lighting is complicated.
Why is it hapenning?
I was wondering, why the described phenomenon occurs. Apparently, even with a small overloading of a part of the sensor, a residual signal is generated at other, on the contrary, insufficiently loaded points, to the extent that this is generating the mentioned noise. The vertical artifacts are a mystery to me, they are probably somehow related to the signal reading on the individual buses. But why isn't it with Nikon? One explanation I see.
With an astronomer ►Martin Myslivec (site is in Czech language only, but photographs definitely worth a look) we compared camera sensors in terms of their suitability for astrophotography. While the RAW was unchanged at Canon (30D, 40D), at Nikon (D3X, D70) there were special artifacts in the Dark Frames. Martin tried to adjust RAW in many ways, but it was not possible to get rid of them. It was clear, that with the Nikons we don't get (then certainly, as I don't know today) pure RAW, but already noiseless and "improved". And this may be the reason for the artefacts, but also the tack sharpness of Canon (if not crazy with the AA filter, of course). Not that Nikon images aren't sharp. They are, absolutely perfect. But the detailed sharpness, that accentuates the almost invisible surface structures with the eye, the hair edges precise and as if cut out - this is higher compared to the 5D III only with the D810 without an AA filter and sometimes not even with it. The optics of both manufacturers are top-notch, so signal processing must be to blame. In addition, when removing filters on APS-C sensors (AA, IR, corrective - it bothers in astrophotography), Martin found, that many (if not all) cameras with a piezo dust collection system do not have an AA filter at all! Without anyone mentioning it anywhere (as far as I know)! The manufacturer simply calculated (and mostly correctly), that most people do not attach such a sharp lens to the body, that moiré appears. Another interesting finding, that Martin saw with his own eyes under a microscope, was the confirmation of the function of a double birefringent filter - one point is really divided into four by the AA filter (birefringent filters are two, rotated 90 degrees). Yuck …
The second option is a different processing of the incident light. While on the Nikon/Sony, the light is imminently transmitted by the AD converter immediately after it generates an electrical impulse in the individual cells, on the Canon it is first mixed with the carrier frequency and led to the AD converter, which is located further away from the sensor. It is also possible, that the signal is preamplified, to pass the path better. This is said to reduce the dynamic range and damage the data in various other ways, but it is also said to reduce noise at high sensitivities without (apparently) having to interfere with it as much as Nikon does (did?). Hard to say.
Large errors cannot be tolerated with photographic cameras designed primarily for work. You can lose customers and your reputation. In exaggeration, I think, that this camera should be provided with "Do not use in backlight" and "Not suitable for professional work" signs by the manufacturer. Yes, some people don't mind these things, others "drown" the shadows and with them stripes and spots, the third one still publishes photographs only on the internet and it is not visible there. However, if you are photographing for customers, who require full-resolution images, or are enlarging your free fine art to large formats, I cannot recommend this camera. Body design, controls, Live View, autofocus, vivid and bright colors - all this is excellent and there is not much to complain about. But the main thing, image quality, is good only occasionally, under certain conditions, and you never know exactly when image damage will occur. It's such a major limitation, that Canon should finally do something about it. And maybe, given my previous 5DSR RAW tests, they've already succeeded.
© Martin Mojzis, 2015.
Photographs: © Martin Mojzis.
►To the beginning of the article.
►Overview of all articles.