I probably should have started my 5D blogging with this post, but it’s only just dawned on me that it might not be a common thing for people to choose to do…
When I started shooting with my 20D I was impressed with it’s low-light ability compared to 35mm film but after the novelty of changing ISO between shots wore off, I naturally started experimenting with ISO 1600, H (ISO 3200) and also H-2 (two stops under exposure, on the basis that shooting in RAW mostly allows up to 2 stops of recovery in post). The results weren’t pretty, although I did spend some time underexposing ISO 1600 to compare to H, the resulting banding noise patterns were very obvious, and very digital.
Fast forward to this year, and my love of trying to take candid pictures in dark places (eg: wedding receptions) hasn’t abated but DSLR’s are now sporting ISO 12,800 as a normal top range option, and the recent crop of compacts such as the Panasonic Lumix and micro 4/3 systems were also looking very attractive until I started playing with the data provided by DxOMark (NB: Flash required for a fair amount of the data display).
DxOMark have a standardised approach to testing cameras, sensors (and more recently lenses) to come up with a ‘magic’ number for the quality of a camera – I’m not about to dive off into a Zen and the Art of Motorcycle Maintenance discussion on quality here, but one aspect of their testing is very interesting to me: what they term Sports shooting. Here’s an extract from their page on the Use Case Scores:
“When shooting a moving scene such as a sports event, action photographers’ primary objective is to freeze the motion, giving priority to short exposure time. To compensate for the lack of exposure, they have to increase the ISO setting, which means the SNR will decrease. How far can they go while keeping decent quality? Our metric, Low-Light ISO, will tell them.”
So how does my 20D stack up against the 5D and more interestingly, against recent cameras such as the 7D, 550D and the 5D Mk II (ok, not that recent, but at least it’s current) ? Go to the menu on the left, click on “Camera Sensor” and then “Compare Sensors” and you should get this page. Note that the 550D has a top ISO setting of 12,800, compared to the 20D’s 3200, but only a marginal increase in the Sports rating.
I should stop here and point out that the 550D has many more compelling features than the 20D, not least of which is the pixel count: to get that many pixels into the same area and still manage to edge out the 20D in terms of noise is a truly remarkable achievement, and yes, I have ignored the potential improvement in image quality if I pixel-binned in post the 17MP 550D down to an 8MP 20D equivalent. HD video, Live View and more are also 100% ignored…
What interested me most was the exposure (sorry) of the internal logic inside Canon: what appears to have been done is to design a sensor with as many MP as possible, but with performance that can be measurably and accurately shown to be not worse than previous generations. Yes, if they made an 8MP sensor today it would have a far, far better low-light ability (they could make it 16MP in reality and pixel-bin, for example) but I wanted to know which camera gives the best low-light performance and will still accept my collection of lenses.
Go to “Camera Sensor” and “Sensor Rankings” to get this information – I can’t offer a direct link this time, but choose “Canon” as the manufacturer, leave the sensor size set to “Sensor format” (ie: all sizes), leave “Mpix” and “Price” alone, and click on the “Sports” button.
Now you will have the year of release along the bottom, and the DxOMark scores up the side, with higher being better. If you follow the band of triangles along the graph from left to right just under the 800 line, you’ll notice that most of the xxD and xxxD cameras fall into that very specific quality range, with very little variance: it’s almost as if new cameras were brought out when the sensor tech advanced sufficiently to squeeze in another MP or two and keep the noise inline with past models. The two highest values are recorded for the 5D Mk II and the 1Ds Mk III, but go back to between 2005 and 2006 and there’s the original 5D: a camera whose low light ability is only bested by the flagship 1D series, or it’s own successor…
Factor in the price of the better cameras plus the fact that photography is not income generating for me (it’s not even income-neutral) and going for a 5 year old system seems a lot more logical.