I recently had cause to buy myself a new digital camera. I’ve been taking photos for many years and know what’s important to me when I look for a camera system. Back in the film days (I still use a Canon T90 on occasion) the faster the film, the more ‘grain‘ you got. For landscape photos – where I wanted vibrant colour and plenty of detail – I would choose Fuji Velvia film. This film had an ISO rating of 50, which meant that it needed a lot of light in order to expose correctly. This wasn’t a problem usually as I’d have my tripod handy and could happily let the shutter stay open for several seconds if need be. If I wanted to photograph BMXers at the local skatepark though I’d have put a roll of Ilford Delta 3200 in my camera. This film is black & white and requires five times less light than Velvia to expose correctly, just what you need for capturing fast action. The downside is that the film is very ‘grainy’
Fast forward to digital…
These days ‘grain’ has been replaced by ‘noise‘. It’s kind of the same thing. When you set your digital camera’s ISO to a high number ,the shutter needs to be open for less time, allowing you to freeze fast moving objects. The trade-off is that your image will contain more ‘noise’.
A few years ago all the manufacturers were striving to produce sensors with more megapixels. To the general public more megapixels = better. The megapixel race has slowed a little now and it seems that what we should really all be caring about is the amount of noise a digital sensor produces. Personally there are other features I’m more concerned about than low light ISO performance but I digress. The goal is to have as little noise as possible when using high ISO settings. This makes sense because at certain ISO settings the amount of noise in an image becomes so great that it degrades the image significantly.
So, why am I writing about this at all? Well, when you want to buy a camera it would seem sensible to visit sites like DPReview and Imaging Resource and compare the sample images of various cameras at the same ISO settings so you can see which one will provide you with the cleanest image when available light is low, yes? Well, not necessarily. We need to think about why we want to use a high ISO setting. Let’s put the idea of using large grain/noise to create mood to one side for now and look at the reason why 99% of the time we want to use a high ISO setting. We need a high ISO setting because after choosing the aperture we want to shoot at we see that the recommended shutter speed we need to get correct exposure is too slow for us to hand-hold and we know that the image will come out blurred. So, to get a faster shutter speed we need to either use a larger aperture or set the ISO higher. Assuming we can’t use a larger aperture (confusingly this is a smaller aperture number e.g. f/1.8 is a larger aperture than f/16) then our only option (tripod aside) is higher ISO.
So, given all that why shouldn’t you just head on over to DPReview and compare the ISO 3200 images from the Olympus EP1 against those of the Panasonic G1? Well, not all ISO settings are created equal. Olympus’s ISO3200 setting is actually less sensitive than Panasonic’s ISO3200. I first noticed this when looking at some figures on the DxOMark website where it seems that an Olympus at ISO3200 is actually less sensitive than a Panasonic at ISO 1600! Now, some people will tell you that the way DxOMark evaluate the ‘real’ ISO settings of a camera is wrong and maybe they’re right.
I decided to look at the sample images at DPReview. I took their ISO1600 sample studio images for the Olympus EP1 and compared the EXIF data from the Panasonic G2 at ISO1600. If we accept that the lighting conditions were constant then we should get identical aperture and shutter speeds showing on both shots. Here are the results:
Panasonic G2 – f/6.3 @ 1/1600th second
Olympus EP1 – f/6.3 @ 1/1000th second
We can see that the Olympus required the shutter remain open for longer than the Panasonic. This result ties in with DxOMark’s findings (not exactly it has to be said). As an additional experiment I took a look at the EXIF data for Canon’s 60D for the same scene:
Canon 60D – f/8 @ 1/640th second
More difficult to figure out because DPReview used a different aperture but if you adjust the aperture to f/6.3 then you’d need 1/1000th second shutter speed, the same as the Olympus.
So, what does this all mean? Well, it means that for a given aperture/shutter speed combination on two separate cameras you may find that one of the cameras needs a higher ISO setting than the other in order to provide equivalent exposure. So, should we be comparing noise levels at set ISO settings to determine the quality of a camera’s sensor? No, I don’t believe we should. I think we should be comparing the noise levels of each camera at set aperture and shutter speeds because ultimately people want to know how well a camera deals with a given situation.
I’ve brought this situation to DPReview’s attention and am hoping for a response.
