SUNDAY 7th FEBRUARY 2010
Sense and Sensitivity
Some notes on the EOS 7D
Having used an EOS 40D since 2007, and a 30D and 20D before that, it's been a fair old wait for Canon to stop warming over their xxD line once every 18 months and bring out something decent. And then they did - the 7D. So I got one.
On paper, it sounds like the perfect birding camera - 1.6x sensor, eight frames per second, 18 Megapixels, HD video, shit-hot autofocus etc etc, but apart from the usual studio tests you find dotted around the interweb, I couldn't really find many useful bird-related 7D samples, so hopefully some unscientific field tests and pictures will be of more practical use to other bird and wildlife photographers thinking of getting one, with particular emphasis on those upgrading from one of the xxD bodies.
Compared to the EOS 40D
The composite below shows how much more Gadwall you get with the 18.0MP EOS 7D in comparison with the 10.1MP EOS 40D. That's a significant amount of extra duck. Having said that, I don't know why they had to give it quite so many pixels, and after using the camera for over a month I'm still not convinced it wouldn't have been more sensible to have given it less. But never mind.
Another thing that struck me when doing some controlled tests between the two cameras is that the 7D tends to meter 1/3 stop more than the 40D. But pulling the exposure back by 1/3 in RAW then puts the 7D files a touch underexposed in comparison - I suppose the sensitivity of the metering and sensor are calibrated slightly differently between the two bodies. White balance under artificial light seems quite different (less yellowy) on the 7D as well. However, regular Canon users will be pleased to learn that on a grey cloudy day the 7D underexposes just like all the others!
Resolution and noise
Here are some proper pictures to look at. Click on the thumbnails below for DPP RAW screengrabs of 16 images at 100%, at various ISOs from 200 to 3200. All are shown with Auto Lighting Optimiser, luminance noise reduction and sharpening at zero (some of those settings may differ from how the info panel says they were shot in-camera). I've pulled up the info panel and tool palette to show what's what.
With 40D RAWs, I used to move both NR sliders in DPP to zero before converting them to a TIFF, and then use a Neat Image plug-in where necessary in Photoshop, but one thing that surprised me when I first did this on the 7D was the amount of chrominance noise in dark shadows at ISO 800 and above. This manifests itself with a nasty spattering of magenta speckles. In fact you can see it beginning to appear on some ISO 400 shots. At one point I even downloaded some 7D RAWs from the Internet to check that mine Sense and Sensitivity - notes on the EOS 7D faulty (they were all the same). Luckily it can be removed fairly painlessly without losing any detail using the chroma slider, but you might need to turn it up a bit.
The appearance of noise at mid to high ISOs tends to be more or less noticeable depending on how it's exposed, so I've tried to give a few examples of different lighting for each sensitivity to give a general indication. Each file is approx 1MB. Don't forget to view them at 100%!
Given that there really is a silly amount of pixels on the sensor, I think that what I've seen so far is more or less in line with my expectations. It's easy to get a bit anal about sensor noise; when I compare some of these samples to slides taken a few years ago on the fabled Kodachrome 64, I know which I'd rather have. That said, I'd happily trade a few of those Megapixels for less noise at 800 ISO. But you can't have it all, I suppose.
Most of my photography is of birds using a telephoto lens, rarely at more than f/8, but during the summer when there are butterflies and moths about I do have cause to use a smaller aperture every now and then. On the one hand, having 18MP should be a benefit when photographing micro-moths at 1:1 or less, but on the other hand you don't want all those Megapixels to be showing up your lens at high f/numbers.
Much has been made by hysterical people on the internet that diffraction will cause the 7D to take crap photos at small apertures. I already know that my macro lens is best at around f/6.3 or f/7.1 with my 40D, so I did a test to see how it looked on the 7D. The feather below was photographed at ISO 100 using a Sigma 150mm f/2.8 at 1:1, at various apertures from f/2.8 to f/16. The lens was set at the same focus point for all shots, and the samples (converted from RAWs) are shown at 100% with sharpness and NR off.
The two thumbnails below show the 7D on the left and the 40D on the right (click to enlarge). You should be able to tell from the file names which aperture is which. It looks as though f/5.6 is sharpest on both bodies in these shots, just a shade sharper than f/8, and it does appear that the lower resolution of the 40D is a little more forgiving as you move further away from the optimum sharpness of the lens. For the sake of completeness I did a comparison of the 1/3 stops between f/5.6 and f/8, which I reckon shows that both are best at around f/6.3 to f/7.1 (as expected). I also did an upsize/downsize comparison (f/6.3).
It seems from these that at optimum apertures the 40D shots are a little bit sharper at pixel level than those of the 7D, and although the latter does resolve more detail, I suspect that for many people the 7D's mighty 18 Megapixels is a good example of the law of diminishing returns in action.
Bear in mind that the set of sequences above is as much a test of the optics as of the camera, and that the image produced by a macro lens at 1:1 probably isn't the best subject for drawing firm conclusions across the board; the results aren't necessarily going to translate to what you would get using a wide angle when photographing a pretty view, for example. Nevertheless, if you're a fan of f/16 then perhaps the 7D might not be for you.
Probably the biggest reason I decided to get a 7D. Over the years, having put up with the quirks, shortcomings and foibles of the 20D, 30D and 40D, and binned a load of otherwise good shots due to stroppy autofocus, it was about time Canon put a decent AF system into one of its 1.6x sensor cameras.
Not only does the 7D have 19 AF sensors compared to the 40D's nine, they're bigger as well. Or at least that's what their viewfinders would have me believe, as shown on the overlay below.
I’ve owned four Canon xxD cameras before the 7D, all of which were a bit wobbly out of the box when it came to AF accuracy. I was expecting the AF on this one to be right on the money from the word go, but it was by far the worst of the lot. Using it with a 500mm lens at f/4, it was hopeless. Good job the 7D has AF microadjustment. According to the manual, ‘Normally, this adjustment should not be required’. Luckily in my case I was able to use it to correct a camera that would otherwise have gone back to the shop.
The problem I always had with the 40D wasn't that servo AF was rubbish; far from it, just that I found it very difficult to use when photographing birds in flight. As the overlay above shows, you're pretty much restricted to using that tiny central sensor if you want the birds body in focus. And therein lies my problem. With an erratically-moving bird, inevitably at fairly close range, just keeping the sensor on the target for any length of time with a heavy 500mm lens I found very difficult with the 40D.
Having only ever used the xxD series cameras, I didn't really know what to expect when photographing flying birds using the 7D, with its reportedly 1D-like performance. Compared to the 40D it does appear faster, and with 19 sensors and a plethora of selection options is clearly superior when tracking a flying bird against both clear and cluttered backgrounds. As far as servo performance goes, it's this improved usability of the 7D's AF that I found to be the most significant difference over that of the 40D.
As I'm pretty much restricted to a few hours each weekend to get out taking pictures, I've not really given the AF a properly thorough test. It took a while to get the microadjustment sorted, and even after I think I've got it bang on, every now and then it will do whatever it fancies and I'll get a sequence which is inexplicably way off. Those instances are in the minority though, and overall it's much better than the 40D, especially with flying birds.
Not exactly a deal-maker for me, although having seen some of the stuff shot with the 7D on the interweb I can quite see it being something I'll be using in the future. Having never previously shot a second of film in my life, my three attempts at video using the 7D so far have been mediocre at best, which I wholly attribute to my woeful technique.
Sorry if you were hoping for something a little more in-depth here.
With 100% frame coverage, the viewfinder and eyecup are a bit bigger than on the 40D (like on the 1D series), which does make it nicer to look through. Another change is the transmissive LCD which is rather fancy, and lets you hide the inactive focus points, giving you an uncluttered field of view. On the downside, the bigger viewfinder means that the anglefinder I use with my 40D doesn’t fit. Furthermore, the eyecup size (Eg) is only shared with the 1D Mark III, so none of the third-party anglefinders yet have adapters to fit. Bah.
With a RAW file size of up to 25MB, you’ll probably want to buy a few more CF cards. Also bear in mind that if you have a PC that’s a few years old, you might find that you can make a cup of tea in the time it takes DPP to open a RAW file.
It’s a different battery to the one in the 40D - an LP-E6, the same as in the 5D Mark II. Seems to last pretty well though - longer than the battery in my 40D, that's for sure.
If you don’t already, you ought to use RAW to get the most out of all those 18MPs. On Canon’s latest versions of DPP, the ‘standard’ JPEG picture style not only gives you a big wedge of sharpening, but also a hefty dollop of luminance noise reduction on top of it, to give a nice mushy finish to all your shots that would otherwise have loads of fine feather detail. Then there’s also that chrominance noise I mentioned earlier to keep an eye on. Shoot RAW and use the Neat Image plug-in in Photoshop, I would.
I think that just about covers it.