How to choose a camera, the right way


We’re living in a golden age for photography gear: there has never been around such a tremendous offer of exceptional cameras for basically every price point. The problem is: how do you choose?

Obviously you will first narrow down the suitable candidates after careful consideration of:

  • the features you need for the kind of pictures you shoot;
  • assuming you’re starting “fresh”, if that particular brand offers all the lenses you need;
  • even if everyone will deny this, the looks (yes, I’m that shallow; I have to “bond” with a camera, and to be able to do this I have to like it!);
  • and finally price.

Even this way you will most likely be left with way too many candidates, and in this particular case even two is a crowd. How to make a final decision?

First a note: ideally you should be able to play with the various cameras in person, but sadly these days is less and less feasible – now even at the few brick-and-mortar stores left here they don’t put batteries inside the cameras, so you cannot really try them other than just check the handling – so we will have to make do with Camerasize* to compare the relative dimensions (hint: at least on a 21″ iMac, if you zoom two times into their page the images will be life sized, making the comparison much easier), and with Dpreview, Imaging-Resource and the likes for specs and, well, reviews.

*Hint 2: you can also check how big the lenses are! Very handy if you’re putting together a travel camera set, for example

But the final and only winner will be determined, at least for me, by the image quality that each particular camera is capable to deliver. This is an even more difficult decision to make, because like I said these days – thankfully! – there are almost no lemons and so the level is pretty darn high.


Going blind

The solution? A blind test. My system is pretty simple. I download a set of RAW files – not JPGs because you don’t know how the reviewers or the camera processed them – from at least two sources, basically the two I mentioned before plus another one for more “real life” examples if I can find one. If they have name indicatives of their origin I dump the pictures into a renaming program so when I open them I can’t tell from what camera they are without looking at the EXIFs.


At this point I open each group of identical images (same subject, same Iso etc.) in PhotoShop with identical settings, scaling them up directly from CameraRaw to the maximum print resolution I could possibly want to print with that particular kind of camera – for example, if I’m buying a camera for street photography I know I won’t need to print as big as a landscape camera, and even if I did the requirements in terms of sharpness will be more relaxed anyway. If I’m being really picky I will also print the images, obviously just sections of the entire image on A4 paper an not on huge A0 sheets!

Now the real examination begins, on the screen or on the physical prints. You will have to check all four borders and the center of the image, to avoid judging the lens instead of the camera, and every particular area of some interest, for example flat black or colored areas for noise if you’re testing high Iso. For each and every section of the image you’ve examined, determine a winner, a 2nd best, a 3rd best etc., writing the results in a row (in the example below “1”, “2” and “3” refer to the position of each image as opened side to side in PhotoShop, or to a number scribbled on the back of an A4 print).  You will be left with something like this:

Corner sx top:  2  3  1
Corner sx bottom:  3  2  1
Corner dx top:  3  2  1
Corner dx bottom:  3  1  2
Center:  3  2  1
Noise:  1  2  3
TOT instances:1: 1x
2: 1x
3: 4x
1:  1x
2: 4x
3: 1x
1: 4x
2: 1x
3: 1x

And now you can just look at the last row, and see that camera n. 3 is obviously – to your eyes and for your own taste – the best of these particular three, camera n. 2 is quite good as well, but it will be best for you to avoid buying camera n. 1.

You will have to repeat this process with each set of images you’ve downloaded. A well representative set for a landscape camera could be, for example: test charts (for the absolute resolution), a natural subject with lots of fine detail, a set of images showing large swats of sky (to check for noise in the blue channel) or overexposed (to check for the ability to recover highlights and dynamic range). For street photography: test charts (this time looking for noise at high Iso in the flat colored areas, more than for resolution), test charts in low light (Dpreview has them), a portrait in natural light (to check for skin tones, you can find these on Imaging-Resource, even if the model is in reality a dummy).

Once you’ve done all this and you will have a winner and a runner up, you can look at the EXIFs to give an identity to those cameras, and finally decide between them just in terms of size, price, looks or whatever other criterion is important for you.

Just recently, running a test like this for choosing a small travel camera that suited my needs I ended up discovering that I liked best the rendering of a camera with a lot less Megapixels than its competitors, but with much sharper lenses available – and way cheaper to boot. So the hour or so you’ll need to complete the process will not only make you happier with your purchase, but maybe even save you money.